don't pass nameidata to ->create()
[GitHub/mt8127/android_kernel_alcatel_ttab.git] / include / linux / fs.h
1 #ifndef _LINUX_FS_H
2 #define _LINUX_FS_H
3
4 /*
5 * This file has definitions for some important file table
6 * structures etc.
7 */
8
9 #include <linux/limits.h>
10 #include <linux/ioctl.h>
11 #include <linux/blk_types.h>
12 #include <linux/types.h>
13
14 /*
15 * It's silly to have NR_OPEN bigger than NR_FILE, but you can change
16 * the file limit at runtime and only root can increase the per-process
17 * nr_file rlimit, so it's safe to set up a ridiculously high absolute
18 * upper limit on files-per-process.
19 *
20 * Some programs (notably those using select()) may have to be
21 * recompiled to take full advantage of the new limits..
22 */
23
24 /* Fixed constants first: */
25 #undef NR_OPEN
26 #define INR_OPEN_CUR 1024 /* Initial setting for nfile rlimits */
27 #define INR_OPEN_MAX 4096 /* Hard limit for nfile rlimits */
28
29 #define BLOCK_SIZE_BITS 10
30 #define BLOCK_SIZE (1<<BLOCK_SIZE_BITS)
31
32 #define SEEK_SET 0 /* seek relative to beginning of file */
33 #define SEEK_CUR 1 /* seek relative to current file position */
34 #define SEEK_END 2 /* seek relative to end of file */
35 #define SEEK_DATA 3 /* seek to the next data */
36 #define SEEK_HOLE 4 /* seek to the next hole */
37 #define SEEK_MAX SEEK_HOLE
38
39 struct fstrim_range {
40 __u64 start;
41 __u64 len;
42 __u64 minlen;
43 };
44
45 /* And dynamically-tunable limits and defaults: */
46 struct files_stat_struct {
47 unsigned long nr_files; /* read only */
48 unsigned long nr_free_files; /* read only */
49 unsigned long max_files; /* tunable */
50 };
51
52 struct inodes_stat_t {
53 int nr_inodes;
54 int nr_unused;
55 int dummy[5]; /* padding for sysctl ABI compatibility */
56 };
57
58
59 #define NR_FILE 8192 /* this can well be larger on a larger system */
60
61 #define MAY_EXEC 0x00000001
62 #define MAY_WRITE 0x00000002
63 #define MAY_READ 0x00000004
64 #define MAY_APPEND 0x00000008
65 #define MAY_ACCESS 0x00000010
66 #define MAY_OPEN 0x00000020
67 #define MAY_CHDIR 0x00000040
68 /* called from RCU mode, don't block */
69 #define MAY_NOT_BLOCK 0x00000080
70
71 /*
72 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
73 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
74 */
75
76 /* file is open for reading */
77 #define FMODE_READ ((__force fmode_t)0x1)
78 /* file is open for writing */
79 #define FMODE_WRITE ((__force fmode_t)0x2)
80 /* file is seekable */
81 #define FMODE_LSEEK ((__force fmode_t)0x4)
82 /* file can be accessed using pread */
83 #define FMODE_PREAD ((__force fmode_t)0x8)
84 /* file can be accessed using pwrite */
85 #define FMODE_PWRITE ((__force fmode_t)0x10)
86 /* File is opened for execution with sys_execve / sys_uselib */
87 #define FMODE_EXEC ((__force fmode_t)0x20)
88 /* File is opened with O_NDELAY (only set for block devices) */
89 #define FMODE_NDELAY ((__force fmode_t)0x40)
90 /* File is opened with O_EXCL (only set for block devices) */
91 #define FMODE_EXCL ((__force fmode_t)0x80)
92 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
93 (specialy hack for floppy.c) */
94 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
95 /* 32bit hashes as llseek() offset (for directories) */
96 #define FMODE_32BITHASH ((__force fmode_t)0x200)
97 /* 64bit hashes as llseek() offset (for directories) */
98 #define FMODE_64BITHASH ((__force fmode_t)0x400)
99
100 /*
101 * Don't update ctime and mtime.
102 *
103 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
104 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
105 */
106 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
107
108 /* Expect random access pattern */
109 #define FMODE_RANDOM ((__force fmode_t)0x1000)
110
111 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
112 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
113
114 /* File is opened with O_PATH; almost nothing can be done with it */
115 #define FMODE_PATH ((__force fmode_t)0x4000)
116
117 /* File was opened by fanotify and shouldn't generate fanotify events */
118 #define FMODE_NONOTIFY ((__force fmode_t)0x1000000)
119
120 /*
121 * The below are the various read and write types that we support. Some of
122 * them include behavioral modifiers that send information down to the
123 * block layer and IO scheduler. Terminology:
124 *
125 * The block layer uses device plugging to defer IO a little bit, in
126 * the hope that we will see more IO very shortly. This increases
127 * coalescing of adjacent IO and thus reduces the number of IOs we
128 * have to send to the device. It also allows for better queuing,
129 * if the IO isn't mergeable. If the caller is going to be waiting
130 * for the IO, then he must ensure that the device is unplugged so
131 * that the IO is dispatched to the driver.
132 *
133 * All IO is handled async in Linux. This is fine for background
134 * writes, but for reads or writes that someone waits for completion
135 * on, we want to notify the block layer and IO scheduler so that they
136 * know about it. That allows them to make better scheduling
137 * decisions. So when the below references 'sync' and 'async', it
138 * is referencing this priority hint.
139 *
140 * With that in mind, the available types are:
141 *
142 * READ A normal read operation. Device will be plugged.
143 * READ_SYNC A synchronous read. Device is not plugged, caller can
144 * immediately wait on this read without caring about
145 * unplugging.
146 * READA Used for read-ahead operations. Lower priority, and the
147 * block layer could (in theory) choose to ignore this
148 * request if it runs into resource problems.
149 * WRITE A normal async write. Device will be plugged.
150 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
151 * the hint that someone will be waiting on this IO
152 * shortly. The write equivalent of READ_SYNC.
153 * WRITE_ODIRECT Special case write for O_DIRECT only.
154 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
155 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
156 * non-volatile media on completion.
157 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
158 * by a cache flush and data is guaranteed to be on
159 * non-volatile media on completion.
160 *
161 */
162 #define RW_MASK REQ_WRITE
163 #define RWA_MASK REQ_RAHEAD
164
165 #define READ 0
166 #define WRITE RW_MASK
167 #define READA RWA_MASK
168
169 #define READ_SYNC (READ | REQ_SYNC)
170 #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
171 #define WRITE_ODIRECT (WRITE | REQ_SYNC)
172 #define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
173 #define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
174 #define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
175
176
177 /*
178 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
179 * that indicates that they should check the contents of the iovec are
180 * valid, but not check the memory that the iovec elements
181 * points too.
182 */
183 #define CHECK_IOVEC_ONLY -1
184
185 #define SEL_IN 1
186 #define SEL_OUT 2
187 #define SEL_EX 4
188
189 /* public flags for file_system_type */
190 #define FS_REQUIRES_DEV 1
191 #define FS_BINARY_MOUNTDATA 2
192 #define FS_HAS_SUBTYPE 4
193 #define FS_REVAL_DOT 16384 /* Check the paths ".", ".." for staleness */
194 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move()
195 * during rename() internally.
196 */
197
198 /*
199 * These are the fs-independent mount-flags: up to 32 flags are supported
200 */
201 #define MS_RDONLY 1 /* Mount read-only */
202 #define MS_NOSUID 2 /* Ignore suid and sgid bits */
203 #define MS_NODEV 4 /* Disallow access to device special files */
204 #define MS_NOEXEC 8 /* Disallow program execution */
205 #define MS_SYNCHRONOUS 16 /* Writes are synced at once */
206 #define MS_REMOUNT 32 /* Alter flags of a mounted FS */
207 #define MS_MANDLOCK 64 /* Allow mandatory locks on an FS */
208 #define MS_DIRSYNC 128 /* Directory modifications are synchronous */
209 #define MS_NOATIME 1024 /* Do not update access times. */
210 #define MS_NODIRATIME 2048 /* Do not update directory access times */
211 #define MS_BIND 4096
212 #define MS_MOVE 8192
213 #define MS_REC 16384
214 #define MS_VERBOSE 32768 /* War is peace. Verbosity is silence.
215 MS_VERBOSE is deprecated. */
216 #define MS_SILENT 32768
217 #define MS_POSIXACL (1<<16) /* VFS does not apply the umask */
218 #define MS_UNBINDABLE (1<<17) /* change to unbindable */
219 #define MS_PRIVATE (1<<18) /* change to private */
220 #define MS_SLAVE (1<<19) /* change to slave */
221 #define MS_SHARED (1<<20) /* change to shared */
222 #define MS_RELATIME (1<<21) /* Update atime relative to mtime/ctime. */
223 #define MS_KERNMOUNT (1<<22) /* this is a kern_mount call */
224 #define MS_I_VERSION (1<<23) /* Update inode I_version field */
225 #define MS_STRICTATIME (1<<24) /* Always perform atime updates */
226 #define MS_NOSEC (1<<28)
227 #define MS_BORN (1<<29)
228 #define MS_ACTIVE (1<<30)
229 #define MS_NOUSER (1<<31)
230
231 /*
232 * Superblock flags that can be altered by MS_REMOUNT
233 */
234 #define MS_RMT_MASK (MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK|MS_I_VERSION)
235
236 /*
237 * Old magic mount flag and mask
238 */
239 #define MS_MGC_VAL 0xC0ED0000
240 #define MS_MGC_MSK 0xffff0000
241
242 /* Inode flags - they have nothing to superblock flags now */
243
244 #define S_SYNC 1 /* Writes are synced at once */
245 #define S_NOATIME 2 /* Do not update access times */
246 #define S_APPEND 4 /* Append-only file */
247 #define S_IMMUTABLE 8 /* Immutable file */
248 #define S_DEAD 16 /* removed, but still open directory */
249 #define S_NOQUOTA 32 /* Inode is not counted to quota */
250 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
251 #define S_NOCMTIME 128 /* Do not update file c/mtime */
252 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
253 #define S_PRIVATE 512 /* Inode is fs-internal */
254 #define S_IMA 1024 /* Inode has an associated IMA struct */
255 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
256 #define S_NOSEC 4096 /* no suid or xattr security attributes */
257
258 /*
259 * Note that nosuid etc flags are inode-specific: setting some file-system
260 * flags just means all the inodes inherit those flags by default. It might be
261 * possible to override it selectively if you really wanted to with some
262 * ioctl() that is not currently implemented.
263 *
264 * Exception: MS_RDONLY is always applied to the entire file system.
265 *
266 * Unfortunately, it is possible to change a filesystems flags with it mounted
267 * with files in use. This means that all of the inodes will not have their
268 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
269 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
270 */
271 #define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))
272
273 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
274 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
275 ((inode)->i_flags & S_SYNC))
276 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
277 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
278 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
279 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
280 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
281
282 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
283 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
284 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
285 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
286
287 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
288 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
289 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
290 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
291 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
292 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
293 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
294
295 /* the read-only stuff doesn't really belong here, but any other place is
296 probably as bad and I don't want to create yet another include file. */
297
298 #define BLKROSET _IO(0x12,93) /* set device read-only (0 = read-write) */
299 #define BLKROGET _IO(0x12,94) /* get read-only status (0 = read_write) */
300 #define BLKRRPART _IO(0x12,95) /* re-read partition table */
301 #define BLKGETSIZE _IO(0x12,96) /* return device size /512 (long *arg) */
302 #define BLKFLSBUF _IO(0x12,97) /* flush buffer cache */
303 #define BLKRASET _IO(0x12,98) /* set read ahead for block device */
304 #define BLKRAGET _IO(0x12,99) /* get current read ahead setting */
305 #define BLKFRASET _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
306 #define BLKFRAGET _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
307 #define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
308 #define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
309 #define BLKSSZGET _IO(0x12,104)/* get block device sector size */
310 #if 0
311 #define BLKPG _IO(0x12,105)/* See blkpg.h */
312
313 /* Some people are morons. Do not use sizeof! */
314
315 #define BLKELVGET _IOR(0x12,106,size_t)/* elevator get */
316 #define BLKELVSET _IOW(0x12,107,size_t)/* elevator set */
317 /* This was here just to show that the number is taken -
318 probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
319 #endif
320 /* A jump here: 108-111 have been used for various private purposes. */
321 #define BLKBSZGET _IOR(0x12,112,size_t)
322 #define BLKBSZSET _IOW(0x12,113,size_t)
323 #define BLKGETSIZE64 _IOR(0x12,114,size_t) /* return device size in bytes (u64 *arg) */
324 #define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup)
325 #define BLKTRACESTART _IO(0x12,116)
326 #define BLKTRACESTOP _IO(0x12,117)
327 #define BLKTRACETEARDOWN _IO(0x12,118)
328 #define BLKDISCARD _IO(0x12,119)
329 #define BLKIOMIN _IO(0x12,120)
330 #define BLKIOOPT _IO(0x12,121)
331 #define BLKALIGNOFF _IO(0x12,122)
332 #define BLKPBSZGET _IO(0x12,123)
333 #define BLKDISCARDZEROES _IO(0x12,124)
334 #define BLKSECDISCARD _IO(0x12,125)
335 #define BLKROTATIONAL _IO(0x12,126)
336
337 #define BMAP_IOCTL 1 /* obsolete - kept for compatibility */
338 #define FIBMAP _IO(0x00,1) /* bmap access */
339 #define FIGETBSZ _IO(0x00,2) /* get the block size used for bmap */
340 #define FIFREEZE _IOWR('X', 119, int) /* Freeze */
341 #define FITHAW _IOWR('X', 120, int) /* Thaw */
342 #define FITRIM _IOWR('X', 121, struct fstrim_range) /* Trim */
343
344 #define FS_IOC_GETFLAGS _IOR('f', 1, long)
345 #define FS_IOC_SETFLAGS _IOW('f', 2, long)
346 #define FS_IOC_GETVERSION _IOR('v', 1, long)
347 #define FS_IOC_SETVERSION _IOW('v', 2, long)
348 #define FS_IOC_FIEMAP _IOWR('f', 11, struct fiemap)
349 #define FS_IOC32_GETFLAGS _IOR('f', 1, int)
350 #define FS_IOC32_SETFLAGS _IOW('f', 2, int)
351 #define FS_IOC32_GETVERSION _IOR('v', 1, int)
352 #define FS_IOC32_SETVERSION _IOW('v', 2, int)
353
354 /*
355 * Inode flags (FS_IOC_GETFLAGS / FS_IOC_SETFLAGS)
356 */
357 #define FS_SECRM_FL 0x00000001 /* Secure deletion */
358 #define FS_UNRM_FL 0x00000002 /* Undelete */
359 #define FS_COMPR_FL 0x00000004 /* Compress file */
360 #define FS_SYNC_FL 0x00000008 /* Synchronous updates */
361 #define FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
362 #define FS_APPEND_FL 0x00000020 /* writes to file may only append */
363 #define FS_NODUMP_FL 0x00000040 /* do not dump file */
364 #define FS_NOATIME_FL 0x00000080 /* do not update atime */
365 /* Reserved for compression usage... */
366 #define FS_DIRTY_FL 0x00000100
367 #define FS_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
368 #define FS_NOCOMP_FL 0x00000400 /* Don't compress */
369 #define FS_ECOMPR_FL 0x00000800 /* Compression error */
370 /* End compression flags --- maybe not all used */
371 #define FS_BTREE_FL 0x00001000 /* btree format dir */
372 #define FS_INDEX_FL 0x00001000 /* hash-indexed directory */
373 #define FS_IMAGIC_FL 0x00002000 /* AFS directory */
374 #define FS_JOURNAL_DATA_FL 0x00004000 /* Reserved for ext3 */
375 #define FS_NOTAIL_FL 0x00008000 /* file tail should not be merged */
376 #define FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
377 #define FS_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
378 #define FS_EXTENT_FL 0x00080000 /* Extents */
379 #define FS_DIRECTIO_FL 0x00100000 /* Use direct i/o */
380 #define FS_NOCOW_FL 0x00800000 /* Do not cow file */
381 #define FS_RESERVED_FL 0x80000000 /* reserved for ext2 lib */
382
383 #define FS_FL_USER_VISIBLE 0x0003DFFF /* User visible flags */
384 #define FS_FL_USER_MODIFIABLE 0x000380FF /* User modifiable flags */
385
386
387 #define SYNC_FILE_RANGE_WAIT_BEFORE 1
388 #define SYNC_FILE_RANGE_WRITE 2
389 #define SYNC_FILE_RANGE_WAIT_AFTER 4
390
391 #ifdef __KERNEL__
392
393 #include <linux/linkage.h>
394 #include <linux/wait.h>
395 #include <linux/kdev_t.h>
396 #include <linux/dcache.h>
397 #include <linux/path.h>
398 #include <linux/stat.h>
399 #include <linux/cache.h>
400 #include <linux/list.h>
401 #include <linux/radix-tree.h>
402 #include <linux/prio_tree.h>
403 #include <linux/init.h>
404 #include <linux/pid.h>
405 #include <linux/bug.h>
406 #include <linux/mutex.h>
407 #include <linux/capability.h>
408 #include <linux/semaphore.h>
409 #include <linux/fiemap.h>
410 #include <linux/rculist_bl.h>
411 #include <linux/atomic.h>
412 #include <linux/shrinker.h>
413 #include <linux/migrate_mode.h>
414 #include <linux/uidgid.h>
415
416 #include <asm/byteorder.h>
417
418 struct export_operations;
419 struct hd_geometry;
420 struct iovec;
421 struct nameidata;
422 struct kiocb;
423 struct kobject;
424 struct pipe_inode_info;
425 struct poll_table_struct;
426 struct kstatfs;
427 struct vm_area_struct;
428 struct vfsmount;
429 struct cred;
430
431 extern void __init inode_init(void);
432 extern void __init inode_init_early(void);
433 extern void __init files_init(unsigned long);
434
435 extern struct files_stat_struct files_stat;
436 extern unsigned long get_max_files(void);
437 extern int sysctl_nr_open;
438 extern struct inodes_stat_t inodes_stat;
439 extern int leases_enable, lease_break_time;
440
441 struct buffer_head;
442 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
443 struct buffer_head *bh_result, int create);
444 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
445 ssize_t bytes, void *private, int ret,
446 bool is_async);
447
448 /*
449 * Attribute flags. These should be or-ed together to figure out what
450 * has been changed!
451 */
452 #define ATTR_MODE (1 << 0)
453 #define ATTR_UID (1 << 1)
454 #define ATTR_GID (1 << 2)
455 #define ATTR_SIZE (1 << 3)
456 #define ATTR_ATIME (1 << 4)
457 #define ATTR_MTIME (1 << 5)
458 #define ATTR_CTIME (1 << 6)
459 #define ATTR_ATIME_SET (1 << 7)
460 #define ATTR_MTIME_SET (1 << 8)
461 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
462 #define ATTR_ATTR_FLAG (1 << 10)
463 #define ATTR_KILL_SUID (1 << 11)
464 #define ATTR_KILL_SGID (1 << 12)
465 #define ATTR_FILE (1 << 13)
466 #define ATTR_KILL_PRIV (1 << 14)
467 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
468 #define ATTR_TIMES_SET (1 << 16)
469
470 /*
471 * This is the Inode Attributes structure, used for notify_change(). It
472 * uses the above definitions as flags, to know which values have changed.
473 * Also, in this manner, a Filesystem can look at only the values it cares
474 * about. Basically, these are the attributes that the VFS layer can
475 * request to change from the FS layer.
476 *
477 * Derek Atkins <warlord@MIT.EDU> 94-10-20
478 */
479 struct iattr {
480 unsigned int ia_valid;
481 umode_t ia_mode;
482 kuid_t ia_uid;
483 kgid_t ia_gid;
484 loff_t ia_size;
485 struct timespec ia_atime;
486 struct timespec ia_mtime;
487 struct timespec ia_ctime;
488
489 /*
490 * Not an attribute, but an auxiliary info for filesystems wanting to
491 * implement an ftruncate() like method. NOTE: filesystem should
492 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
493 */
494 struct file *ia_file;
495 };
496
497 /*
498 * Includes for diskquotas.
499 */
500 #include <linux/quota.h>
501
502 /**
503 * enum positive_aop_returns - aop return codes with specific semantics
504 *
505 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
506 * completed, that the page is still locked, and
507 * should be considered active. The VM uses this hint
508 * to return the page to the active list -- it won't
509 * be a candidate for writeback again in the near
510 * future. Other callers must be careful to unlock
511 * the page if they get this return. Returned by
512 * writepage();
513 *
514 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
515 * unlocked it and the page might have been truncated.
516 * The caller should back up to acquiring a new page and
517 * trying again. The aop will be taking reasonable
518 * precautions not to livelock. If the caller held a page
519 * reference, it should drop it before retrying. Returned
520 * by readpage().
521 *
522 * address_space_operation functions return these large constants to indicate
523 * special semantics to the caller. These are much larger than the bytes in a
524 * page to allow for functions that return the number of bytes operated on in a
525 * given page.
526 */
527
528 enum positive_aop_returns {
529 AOP_WRITEPAGE_ACTIVATE = 0x80000,
530 AOP_TRUNCATED_PAGE = 0x80001,
531 };
532
533 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
534 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
535 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
536 * helper code (eg buffer layer)
537 * to clear GFP_FS from alloc */
538
539 /*
540 * oh the beauties of C type declarations.
541 */
542 struct page;
543 struct address_space;
544 struct writeback_control;
545
546 struct iov_iter {
547 const struct iovec *iov;
548 unsigned long nr_segs;
549 size_t iov_offset;
550 size_t count;
551 };
552
553 size_t iov_iter_copy_from_user_atomic(struct page *page,
554 struct iov_iter *i, unsigned long offset, size_t bytes);
555 size_t iov_iter_copy_from_user(struct page *page,
556 struct iov_iter *i, unsigned long offset, size_t bytes);
557 void iov_iter_advance(struct iov_iter *i, size_t bytes);
558 int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes);
559 size_t iov_iter_single_seg_count(struct iov_iter *i);
560
561 static inline void iov_iter_init(struct iov_iter *i,
562 const struct iovec *iov, unsigned long nr_segs,
563 size_t count, size_t written)
564 {
565 i->iov = iov;
566 i->nr_segs = nr_segs;
567 i->iov_offset = 0;
568 i->count = count + written;
569
570 iov_iter_advance(i, written);
571 }
572
573 static inline size_t iov_iter_count(struct iov_iter *i)
574 {
575 return i->count;
576 }
577
578 /*
579 * "descriptor" for what we're up to with a read.
580 * This allows us to use the same read code yet
581 * have multiple different users of the data that
582 * we read from a file.
583 *
584 * The simplest case just copies the data to user
585 * mode.
586 */
587 typedef struct {
588 size_t written;
589 size_t count;
590 union {
591 char __user *buf;
592 void *data;
593 } arg;
594 int error;
595 } read_descriptor_t;
596
597 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
598 unsigned long, unsigned long);
599
600 struct address_space_operations {
601 int (*writepage)(struct page *page, struct writeback_control *wbc);
602 int (*readpage)(struct file *, struct page *);
603
604 /* Write back some dirty pages from this mapping. */
605 int (*writepages)(struct address_space *, struct writeback_control *);
606
607 /* Set a page dirty. Return true if this dirtied it */
608 int (*set_page_dirty)(struct page *page);
609
610 int (*readpages)(struct file *filp, struct address_space *mapping,
611 struct list_head *pages, unsigned nr_pages);
612
613 int (*write_begin)(struct file *, struct address_space *mapping,
614 loff_t pos, unsigned len, unsigned flags,
615 struct page **pagep, void **fsdata);
616 int (*write_end)(struct file *, struct address_space *mapping,
617 loff_t pos, unsigned len, unsigned copied,
618 struct page *page, void *fsdata);
619
620 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
621 sector_t (*bmap)(struct address_space *, sector_t);
622 void (*invalidatepage) (struct page *, unsigned long);
623 int (*releasepage) (struct page *, gfp_t);
624 void (*freepage)(struct page *);
625 ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
626 loff_t offset, unsigned long nr_segs);
627 int (*get_xip_mem)(struct address_space *, pgoff_t, int,
628 void **, unsigned long *);
629 /*
630 * migrate the contents of a page to the specified target. If sync
631 * is false, it must not block.
632 */
633 int (*migratepage) (struct address_space *,
634 struct page *, struct page *, enum migrate_mode);
635 int (*launder_page) (struct page *);
636 int (*is_partially_uptodate) (struct page *, read_descriptor_t *,
637 unsigned long);
638 int (*error_remove_page)(struct address_space *, struct page *);
639 };
640
641 extern const struct address_space_operations empty_aops;
642
643 /*
644 * pagecache_write_begin/pagecache_write_end must be used by general code
645 * to write into the pagecache.
646 */
647 int pagecache_write_begin(struct file *, struct address_space *mapping,
648 loff_t pos, unsigned len, unsigned flags,
649 struct page **pagep, void **fsdata);
650
651 int pagecache_write_end(struct file *, struct address_space *mapping,
652 loff_t pos, unsigned len, unsigned copied,
653 struct page *page, void *fsdata);
654
655 struct backing_dev_info;
656 struct address_space {
657 struct inode *host; /* owner: inode, block_device */
658 struct radix_tree_root page_tree; /* radix tree of all pages */
659 spinlock_t tree_lock; /* and lock protecting it */
660 unsigned int i_mmap_writable;/* count VM_SHARED mappings */
661 struct prio_tree_root i_mmap; /* tree of private and shared mappings */
662 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
663 struct mutex i_mmap_mutex; /* protect tree, count, list */
664 /* Protected by tree_lock together with the radix tree */
665 unsigned long nrpages; /* number of total pages */
666 pgoff_t writeback_index;/* writeback starts here */
667 const struct address_space_operations *a_ops; /* methods */
668 unsigned long flags; /* error bits/gfp mask */
669 struct backing_dev_info *backing_dev_info; /* device readahead, etc */
670 spinlock_t private_lock; /* for use by the address_space */
671 struct list_head private_list; /* ditto */
672 struct address_space *assoc_mapping; /* ditto */
673 } __attribute__((aligned(sizeof(long))));
674 /*
675 * On most architectures that alignment is already the case; but
676 * must be enforced here for CRIS, to let the least significant bit
677 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
678 */
679 struct request_queue;
680
681 struct block_device {
682 dev_t bd_dev; /* not a kdev_t - it's a search key */
683 int bd_openers;
684 struct inode * bd_inode; /* will die */
685 struct super_block * bd_super;
686 struct mutex bd_mutex; /* open/close mutex */
687 struct list_head bd_inodes;
688 void * bd_claiming;
689 void * bd_holder;
690 int bd_holders;
691 bool bd_write_holder;
692 #ifdef CONFIG_SYSFS
693 struct list_head bd_holder_disks;
694 #endif
695 struct block_device * bd_contains;
696 unsigned bd_block_size;
697 struct hd_struct * bd_part;
698 /* number of times partitions within this device have been opened. */
699 unsigned bd_part_count;
700 int bd_invalidated;
701 struct gendisk * bd_disk;
702 struct request_queue * bd_queue;
703 struct list_head bd_list;
704 /*
705 * Private data. You must have bd_claim'ed the block_device
706 * to use this. NOTE: bd_claim allows an owner to claim
707 * the same device multiple times, the owner must take special
708 * care to not mess up bd_private for that case.
709 */
710 unsigned long bd_private;
711
712 /* The counter of freeze processes */
713 int bd_fsfreeze_count;
714 /* Mutex for freeze */
715 struct mutex bd_fsfreeze_mutex;
716 };
717
718 /*
719 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
720 * radix trees
721 */
722 #define PAGECACHE_TAG_DIRTY 0
723 #define PAGECACHE_TAG_WRITEBACK 1
724 #define PAGECACHE_TAG_TOWRITE 2
725
726 int mapping_tagged(struct address_space *mapping, int tag);
727
728 /*
729 * Might pages of this file be mapped into userspace?
730 */
731 static inline int mapping_mapped(struct address_space *mapping)
732 {
733 return !prio_tree_empty(&mapping->i_mmap) ||
734 !list_empty(&mapping->i_mmap_nonlinear);
735 }
736
737 /*
738 * Might pages of this file have been modified in userspace?
739 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
740 * marks vma as VM_SHARED if it is shared, and the file was opened for
741 * writing i.e. vma may be mprotected writable even if now readonly.
742 */
743 static inline int mapping_writably_mapped(struct address_space *mapping)
744 {
745 return mapping->i_mmap_writable != 0;
746 }
747
748 /*
749 * Use sequence counter to get consistent i_size on 32-bit processors.
750 */
751 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
752 #include <linux/seqlock.h>
753 #define __NEED_I_SIZE_ORDERED
754 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
755 #else
756 #define i_size_ordered_init(inode) do { } while (0)
757 #endif
758
759 struct posix_acl;
760 #define ACL_NOT_CACHED ((void *)(-1))
761
762 #define IOP_FASTPERM 0x0001
763 #define IOP_LOOKUP 0x0002
764 #define IOP_NOFOLLOW 0x0004
765
766 /*
767 * Keep mostly read-only and often accessed (especially for
768 * the RCU path lookup and 'stat' data) fields at the beginning
769 * of the 'struct inode'
770 */
771 struct inode {
772 umode_t i_mode;
773 unsigned short i_opflags;
774 kuid_t i_uid;
775 kgid_t i_gid;
776 unsigned int i_flags;
777
778 #ifdef CONFIG_FS_POSIX_ACL
779 struct posix_acl *i_acl;
780 struct posix_acl *i_default_acl;
781 #endif
782
783 const struct inode_operations *i_op;
784 struct super_block *i_sb;
785 struct address_space *i_mapping;
786
787 #ifdef CONFIG_SECURITY
788 void *i_security;
789 #endif
790
791 /* Stat data, not accessed from path walking */
792 unsigned long i_ino;
793 /*
794 * Filesystems may only read i_nlink directly. They shall use the
795 * following functions for modification:
796 *
797 * (set|clear|inc|drop)_nlink
798 * inode_(inc|dec)_link_count
799 */
800 union {
801 const unsigned int i_nlink;
802 unsigned int __i_nlink;
803 };
804 dev_t i_rdev;
805 loff_t i_size;
806 struct timespec i_atime;
807 struct timespec i_mtime;
808 struct timespec i_ctime;
809 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
810 unsigned short i_bytes;
811 unsigned int i_blkbits;
812 blkcnt_t i_blocks;
813
814 #ifdef __NEED_I_SIZE_ORDERED
815 seqcount_t i_size_seqcount;
816 #endif
817
818 /* Misc */
819 unsigned long i_state;
820 struct mutex i_mutex;
821
822 unsigned long dirtied_when; /* jiffies of first dirtying */
823
824 struct hlist_node i_hash;
825 struct list_head i_wb_list; /* backing dev IO list */
826 struct list_head i_lru; /* inode LRU list */
827 struct list_head i_sb_list;
828 union {
829 struct hlist_head i_dentry;
830 struct rcu_head i_rcu;
831 };
832 u64 i_version;
833 atomic_t i_count;
834 atomic_t i_dio_count;
835 atomic_t i_writecount;
836 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
837 struct file_lock *i_flock;
838 struct address_space i_data;
839 #ifdef CONFIG_QUOTA
840 struct dquot *i_dquot[MAXQUOTAS];
841 #endif
842 struct list_head i_devices;
843 union {
844 struct pipe_inode_info *i_pipe;
845 struct block_device *i_bdev;
846 struct cdev *i_cdev;
847 };
848
849 __u32 i_generation;
850
851 #ifdef CONFIG_FSNOTIFY
852 __u32 i_fsnotify_mask; /* all events this inode cares about */
853 struct hlist_head i_fsnotify_marks;
854 #endif
855
856 #ifdef CONFIG_IMA
857 atomic_t i_readcount; /* struct files open RO */
858 #endif
859 void *i_private; /* fs or device private pointer */
860 };
861
862 static inline int inode_unhashed(struct inode *inode)
863 {
864 return hlist_unhashed(&inode->i_hash);
865 }
866
867 /*
868 * inode->i_mutex nesting subclasses for the lock validator:
869 *
870 * 0: the object of the current VFS operation
871 * 1: parent
872 * 2: child/target
873 * 3: quota file
874 *
875 * The locking order between these classes is
876 * parent -> child -> normal -> xattr -> quota
877 */
878 enum inode_i_mutex_lock_class
879 {
880 I_MUTEX_NORMAL,
881 I_MUTEX_PARENT,
882 I_MUTEX_CHILD,
883 I_MUTEX_XATTR,
884 I_MUTEX_QUOTA
885 };
886
887 /*
888 * NOTE: in a 32bit arch with a preemptable kernel and
889 * an UP compile the i_size_read/write must be atomic
890 * with respect to the local cpu (unlike with preempt disabled),
891 * but they don't need to be atomic with respect to other cpus like in
892 * true SMP (so they need either to either locally disable irq around
893 * the read or for example on x86 they can be still implemented as a
894 * cmpxchg8b without the need of the lock prefix). For SMP compiles
895 * and 64bit archs it makes no difference if preempt is enabled or not.
896 */
897 static inline loff_t i_size_read(const struct inode *inode)
898 {
899 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
900 loff_t i_size;
901 unsigned int seq;
902
903 do {
904 seq = read_seqcount_begin(&inode->i_size_seqcount);
905 i_size = inode->i_size;
906 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
907 return i_size;
908 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
909 loff_t i_size;
910
911 preempt_disable();
912 i_size = inode->i_size;
913 preempt_enable();
914 return i_size;
915 #else
916 return inode->i_size;
917 #endif
918 }
919
920 /*
921 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
922 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
923 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
924 */
925 static inline void i_size_write(struct inode *inode, loff_t i_size)
926 {
927 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
928 write_seqcount_begin(&inode->i_size_seqcount);
929 inode->i_size = i_size;
930 write_seqcount_end(&inode->i_size_seqcount);
931 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
932 preempt_disable();
933 inode->i_size = i_size;
934 preempt_enable();
935 #else
936 inode->i_size = i_size;
937 #endif
938 }
939
940 /* Helper functions so that in most cases filesystems will
941 * not need to deal directly with kuid_t and kgid_t and can
942 * instead deal with the raw numeric values that are stored
943 * in the filesystem.
944 */
945 static inline uid_t i_uid_read(const struct inode *inode)
946 {
947 return from_kuid(&init_user_ns, inode->i_uid);
948 }
949
950 static inline gid_t i_gid_read(const struct inode *inode)
951 {
952 return from_kgid(&init_user_ns, inode->i_gid);
953 }
954
955 static inline void i_uid_write(struct inode *inode, uid_t uid)
956 {
957 inode->i_uid = make_kuid(&init_user_ns, uid);
958 }
959
960 static inline void i_gid_write(struct inode *inode, gid_t gid)
961 {
962 inode->i_gid = make_kgid(&init_user_ns, gid);
963 }
964
965 static inline unsigned iminor(const struct inode *inode)
966 {
967 return MINOR(inode->i_rdev);
968 }
969
970 static inline unsigned imajor(const struct inode *inode)
971 {
972 return MAJOR(inode->i_rdev);
973 }
974
975 extern struct block_device *I_BDEV(struct inode *inode);
976
977 struct fown_struct {
978 rwlock_t lock; /* protects pid, uid, euid fields */
979 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
980 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
981 kuid_t uid, euid; /* uid/euid of process setting the owner */
982 int signum; /* posix.1b rt signal to be delivered on IO */
983 };
984
985 /*
986 * Track a single file's readahead state
987 */
988 struct file_ra_state {
989 pgoff_t start; /* where readahead started */
990 unsigned int size; /* # of readahead pages */
991 unsigned int async_size; /* do asynchronous readahead when
992 there are only # of pages ahead */
993
994 unsigned int ra_pages; /* Maximum readahead window */
995 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
996 loff_t prev_pos; /* Cache last read() position */
997 };
998
999 /*
1000 * Check if @index falls in the readahead windows.
1001 */
1002 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
1003 {
1004 return (index >= ra->start &&
1005 index < ra->start + ra->size);
1006 }
1007
1008 #define FILE_MNT_WRITE_TAKEN 1
1009 #define FILE_MNT_WRITE_RELEASED 2
1010
1011 struct file {
1012 /*
1013 * fu_list becomes invalid after file_free is called and queued via
1014 * fu_rcuhead for RCU freeing
1015 */
1016 union {
1017 struct list_head fu_list;
1018 struct rcu_head fu_rcuhead;
1019 } f_u;
1020 struct path f_path;
1021 #define f_dentry f_path.dentry
1022 #define f_vfsmnt f_path.mnt
1023 const struct file_operations *f_op;
1024
1025 /*
1026 * Protects f_ep_links, f_flags, f_pos vs i_size in lseek SEEK_CUR.
1027 * Must not be taken from IRQ context.
1028 */
1029 spinlock_t f_lock;
1030 #ifdef CONFIG_SMP
1031 int f_sb_list_cpu;
1032 #endif
1033 atomic_long_t f_count;
1034 unsigned int f_flags;
1035 fmode_t f_mode;
1036 loff_t f_pos;
1037 struct fown_struct f_owner;
1038 const struct cred *f_cred;
1039 struct file_ra_state f_ra;
1040
1041 u64 f_version;
1042 #ifdef CONFIG_SECURITY
1043 void *f_security;
1044 #endif
1045 /* needed for tty driver, and maybe others */
1046 void *private_data;
1047
1048 #ifdef CONFIG_EPOLL
1049 /* Used by fs/eventpoll.c to link all the hooks to this file */
1050 struct list_head f_ep_links;
1051 struct list_head f_tfile_llink;
1052 #endif /* #ifdef CONFIG_EPOLL */
1053 struct address_space *f_mapping;
1054 #ifdef CONFIG_DEBUG_WRITECOUNT
1055 unsigned long f_mnt_write_state;
1056 #endif
1057 };
1058
1059 struct file_handle {
1060 __u32 handle_bytes;
1061 int handle_type;
1062 /* file identifier */
1063 unsigned char f_handle[0];
1064 };
1065
1066 #define get_file(x) atomic_long_inc(&(x)->f_count)
1067 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
1068 #define file_count(x) atomic_long_read(&(x)->f_count)
1069
1070 #ifdef CONFIG_DEBUG_WRITECOUNT
1071 static inline void file_take_write(struct file *f)
1072 {
1073 WARN_ON(f->f_mnt_write_state != 0);
1074 f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN;
1075 }
1076 static inline void file_release_write(struct file *f)
1077 {
1078 f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED;
1079 }
1080 static inline void file_reset_write(struct file *f)
1081 {
1082 f->f_mnt_write_state = 0;
1083 }
1084 static inline void file_check_state(struct file *f)
1085 {
1086 /*
1087 * At this point, either both or neither of these bits
1088 * should be set.
1089 */
1090 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN);
1091 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED);
1092 }
1093 static inline int file_check_writeable(struct file *f)
1094 {
1095 if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN)
1096 return 0;
1097 printk(KERN_WARNING "writeable file with no "
1098 "mnt_want_write()\n");
1099 WARN_ON(1);
1100 return -EINVAL;
1101 }
1102 #else /* !CONFIG_DEBUG_WRITECOUNT */
1103 static inline void file_take_write(struct file *filp) {}
1104 static inline void file_release_write(struct file *filp) {}
1105 static inline void file_reset_write(struct file *filp) {}
1106 static inline void file_check_state(struct file *filp) {}
1107 static inline int file_check_writeable(struct file *filp)
1108 {
1109 return 0;
1110 }
1111 #endif /* CONFIG_DEBUG_WRITECOUNT */
1112
1113 #define MAX_NON_LFS ((1UL<<31) - 1)
1114
1115 /* Page cache limit. The filesystems should put that into their s_maxbytes
1116 limits, otherwise bad things can happen in VM. */
1117 #if BITS_PER_LONG==32
1118 #define MAX_LFS_FILESIZE (((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
1119 #elif BITS_PER_LONG==64
1120 #define MAX_LFS_FILESIZE 0x7fffffffffffffffUL
1121 #endif
1122
1123 #define FL_POSIX 1
1124 #define FL_FLOCK 2
1125 #define FL_ACCESS 8 /* not trying to lock, just looking */
1126 #define FL_EXISTS 16 /* when unlocking, test for existence */
1127 #define FL_LEASE 32 /* lease held on this file */
1128 #define FL_CLOSE 64 /* unlock on close */
1129 #define FL_SLEEP 128 /* A blocking lock */
1130 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
1131 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
1132
1133 /*
1134 * Special return value from posix_lock_file() and vfs_lock_file() for
1135 * asynchronous locking.
1136 */
1137 #define FILE_LOCK_DEFERRED 1
1138
1139 /*
1140 * The POSIX file lock owner is determined by
1141 * the "struct files_struct" in the thread group
1142 * (or NULL for no owner - BSD locks).
1143 *
1144 * Lockd stuffs a "host" pointer into this.
1145 */
1146 typedef struct files_struct *fl_owner_t;
1147
1148 struct file_lock_operations {
1149 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1150 void (*fl_release_private)(struct file_lock *);
1151 };
1152
1153 struct lock_manager_operations {
1154 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
1155 void (*lm_notify)(struct file_lock *); /* unblock callback */
1156 int (*lm_grant)(struct file_lock *, struct file_lock *, int);
1157 void (*lm_release_private)(struct file_lock *);
1158 void (*lm_break)(struct file_lock *);
1159 int (*lm_change)(struct file_lock **, int);
1160 };
1161
1162 struct lock_manager {
1163 struct list_head list;
1164 };
1165
1166 void locks_start_grace(struct lock_manager *);
1167 void locks_end_grace(struct lock_manager *);
1168 int locks_in_grace(void);
1169
1170 /* that will die - we need it for nfs_lock_info */
1171 #include <linux/nfs_fs_i.h>
1172
1173 struct file_lock {
1174 struct file_lock *fl_next; /* singly linked list for this inode */
1175 struct list_head fl_link; /* doubly linked list of all locks */
1176 struct list_head fl_block; /* circular list of blocked processes */
1177 fl_owner_t fl_owner;
1178 unsigned int fl_flags;
1179 unsigned char fl_type;
1180 unsigned int fl_pid;
1181 struct pid *fl_nspid;
1182 wait_queue_head_t fl_wait;
1183 struct file *fl_file;
1184 loff_t fl_start;
1185 loff_t fl_end;
1186
1187 struct fasync_struct * fl_fasync; /* for lease break notifications */
1188 /* for lease breaks: */
1189 unsigned long fl_break_time;
1190 unsigned long fl_downgrade_time;
1191
1192 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1193 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1194 union {
1195 struct nfs_lock_info nfs_fl;
1196 struct nfs4_lock_info nfs4_fl;
1197 struct {
1198 struct list_head link; /* link in AFS vnode's pending_locks list */
1199 int state; /* state of grant or error if -ve */
1200 } afs;
1201 } fl_u;
1202 };
1203
1204 /* The following constant reflects the upper bound of the file/locking space */
1205 #ifndef OFFSET_MAX
1206 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1207 #define OFFSET_MAX INT_LIMIT(loff_t)
1208 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1209 #endif
1210
1211 #include <linux/fcntl.h>
1212
1213 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1214
1215 #ifdef CONFIG_FILE_LOCKING
1216 extern int fcntl_getlk(struct file *, struct flock __user *);
1217 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1218 struct flock __user *);
1219
1220 #if BITS_PER_LONG == 32
1221 extern int fcntl_getlk64(struct file *, struct flock64 __user *);
1222 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1223 struct flock64 __user *);
1224 #endif
1225
1226 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1227 extern int fcntl_getlease(struct file *filp);
1228
1229 /* fs/locks.c */
1230 void locks_free_lock(struct file_lock *fl);
1231 extern void locks_init_lock(struct file_lock *);
1232 extern struct file_lock * locks_alloc_lock(void);
1233 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1234 extern void __locks_copy_lock(struct file_lock *, const struct file_lock *);
1235 extern void locks_remove_posix(struct file *, fl_owner_t);
1236 extern void locks_remove_flock(struct file *);
1237 extern void locks_release_private(struct file_lock *);
1238 extern void posix_test_lock(struct file *, struct file_lock *);
1239 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1240 extern int posix_lock_file_wait(struct file *, struct file_lock *);
1241 extern int posix_unblock_lock(struct file *, struct file_lock *);
1242 extern int vfs_test_lock(struct file *, struct file_lock *);
1243 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1244 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1245 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
1246 extern int __break_lease(struct inode *inode, unsigned int flags);
1247 extern void lease_get_mtime(struct inode *, struct timespec *time);
1248 extern int generic_setlease(struct file *, long, struct file_lock **);
1249 extern int vfs_setlease(struct file *, long, struct file_lock **);
1250 extern int lease_modify(struct file_lock **, int);
1251 extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
1252 extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
1253 extern void locks_delete_block(struct file_lock *waiter);
1254 extern void lock_flocks(void);
1255 extern void unlock_flocks(void);
1256 #else /* !CONFIG_FILE_LOCKING */
1257 static inline int fcntl_getlk(struct file *file, struct flock __user *user)
1258 {
1259 return -EINVAL;
1260 }
1261
1262 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1263 unsigned int cmd, struct flock __user *user)
1264 {
1265 return -EACCES;
1266 }
1267
1268 #if BITS_PER_LONG == 32
1269 static inline int fcntl_getlk64(struct file *file, struct flock64 __user *user)
1270 {
1271 return -EINVAL;
1272 }
1273
1274 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1275 unsigned int cmd, struct flock64 __user *user)
1276 {
1277 return -EACCES;
1278 }
1279 #endif
1280 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1281 {
1282 return 0;
1283 }
1284
1285 static inline int fcntl_getlease(struct file *filp)
1286 {
1287 return 0;
1288 }
1289
1290 static inline void locks_init_lock(struct file_lock *fl)
1291 {
1292 return;
1293 }
1294
1295 static inline void __locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1296 {
1297 return;
1298 }
1299
1300 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1301 {
1302 return;
1303 }
1304
1305 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1306 {
1307 return;
1308 }
1309
1310 static inline void locks_remove_flock(struct file *filp)
1311 {
1312 return;
1313 }
1314
1315 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1316 {
1317 return;
1318 }
1319
1320 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1321 struct file_lock *conflock)
1322 {
1323 return -ENOLCK;
1324 }
1325
1326 static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1327 {
1328 return -ENOLCK;
1329 }
1330
1331 static inline int posix_unblock_lock(struct file *filp,
1332 struct file_lock *waiter)
1333 {
1334 return -ENOENT;
1335 }
1336
1337 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1338 {
1339 return 0;
1340 }
1341
1342 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1343 struct file_lock *fl, struct file_lock *conf)
1344 {
1345 return -ENOLCK;
1346 }
1347
1348 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1349 {
1350 return 0;
1351 }
1352
1353 static inline int flock_lock_file_wait(struct file *filp,
1354 struct file_lock *request)
1355 {
1356 return -ENOLCK;
1357 }
1358
1359 static inline int __break_lease(struct inode *inode, unsigned int mode)
1360 {
1361 return 0;
1362 }
1363
1364 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1365 {
1366 return;
1367 }
1368
1369 static inline int generic_setlease(struct file *filp, long arg,
1370 struct file_lock **flp)
1371 {
1372 return -EINVAL;
1373 }
1374
1375 static inline int vfs_setlease(struct file *filp, long arg,
1376 struct file_lock **lease)
1377 {
1378 return -EINVAL;
1379 }
1380
1381 static inline int lease_modify(struct file_lock **before, int arg)
1382 {
1383 return -EINVAL;
1384 }
1385
1386 static inline int lock_may_read(struct inode *inode, loff_t start,
1387 unsigned long len)
1388 {
1389 return 1;
1390 }
1391
1392 static inline int lock_may_write(struct inode *inode, loff_t start,
1393 unsigned long len)
1394 {
1395 return 1;
1396 }
1397
1398 static inline void locks_delete_block(struct file_lock *waiter)
1399 {
1400 }
1401
1402 static inline void lock_flocks(void)
1403 {
1404 }
1405
1406 static inline void unlock_flocks(void)
1407 {
1408 }
1409
1410 #endif /* !CONFIG_FILE_LOCKING */
1411
1412
1413 struct fasync_struct {
1414 spinlock_t fa_lock;
1415 int magic;
1416 int fa_fd;
1417 struct fasync_struct *fa_next; /* singly linked list */
1418 struct file *fa_file;
1419 struct rcu_head fa_rcu;
1420 };
1421
1422 #define FASYNC_MAGIC 0x4601
1423
1424 /* SMP safe fasync helpers: */
1425 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1426 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1427 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1428 extern struct fasync_struct *fasync_alloc(void);
1429 extern void fasync_free(struct fasync_struct *);
1430
1431 /* can be called from interrupts */
1432 extern void kill_fasync(struct fasync_struct **, int, int);
1433
1434 extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1435 extern int f_setown(struct file *filp, unsigned long arg, int force);
1436 extern void f_delown(struct file *filp);
1437 extern pid_t f_getown(struct file *filp);
1438 extern int send_sigurg(struct fown_struct *fown);
1439
1440 /*
1441 * Umount options
1442 */
1443
1444 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1445 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1446 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1447 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1448 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1449
1450 extern struct list_head super_blocks;
1451 extern spinlock_t sb_lock;
1452
1453 struct super_block {
1454 struct list_head s_list; /* Keep this first */
1455 dev_t s_dev; /* search index; _not_ kdev_t */
1456 unsigned char s_dirt;
1457 unsigned char s_blocksize_bits;
1458 unsigned long s_blocksize;
1459 loff_t s_maxbytes; /* Max file size */
1460 struct file_system_type *s_type;
1461 const struct super_operations *s_op;
1462 const struct dquot_operations *dq_op;
1463 const struct quotactl_ops *s_qcop;
1464 const struct export_operations *s_export_op;
1465 unsigned long s_flags;
1466 unsigned long s_magic;
1467 struct dentry *s_root;
1468 struct rw_semaphore s_umount;
1469 struct mutex s_lock;
1470 int s_count;
1471 atomic_t s_active;
1472 #ifdef CONFIG_SECURITY
1473 void *s_security;
1474 #endif
1475 const struct xattr_handler **s_xattr;
1476
1477 struct list_head s_inodes; /* all inodes */
1478 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1479 #ifdef CONFIG_SMP
1480 struct list_head __percpu *s_files;
1481 #else
1482 struct list_head s_files;
1483 #endif
1484 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1485 /* s_dentry_lru, s_nr_dentry_unused protected by dcache.c lru locks */
1486 struct list_head s_dentry_lru; /* unused dentry lru */
1487 int s_nr_dentry_unused; /* # of dentry on lru */
1488
1489 /* s_inode_lru_lock protects s_inode_lru and s_nr_inodes_unused */
1490 spinlock_t s_inode_lru_lock ____cacheline_aligned_in_smp;
1491 struct list_head s_inode_lru; /* unused inode lru */
1492 int s_nr_inodes_unused; /* # of inodes on lru */
1493
1494 struct block_device *s_bdev;
1495 struct backing_dev_info *s_bdi;
1496 struct mtd_info *s_mtd;
1497 struct hlist_node s_instances;
1498 struct quota_info s_dquot; /* Diskquota specific options */
1499
1500 int s_frozen;
1501 wait_queue_head_t s_wait_unfrozen;
1502
1503 char s_id[32]; /* Informational name */
1504 u8 s_uuid[16]; /* UUID */
1505
1506 void *s_fs_info; /* Filesystem private info */
1507 unsigned int s_max_links;
1508 fmode_t s_mode;
1509
1510 /* Granularity of c/m/atime in ns.
1511 Cannot be worse than a second */
1512 u32 s_time_gran;
1513
1514 /*
1515 * The next field is for VFS *only*. No filesystems have any business
1516 * even looking at it. You had been warned.
1517 */
1518 struct mutex s_vfs_rename_mutex; /* Kludge */
1519
1520 /*
1521 * Filesystem subtype. If non-empty the filesystem type field
1522 * in /proc/mounts will be "type.subtype"
1523 */
1524 char *s_subtype;
1525
1526 /*
1527 * Saved mount options for lazy filesystems using
1528 * generic_show_options()
1529 */
1530 char __rcu *s_options;
1531 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1532
1533 /*
1534 * Saved pool identifier for cleancache (-1 means none)
1535 */
1536 int cleancache_poolid;
1537
1538 struct shrinker s_shrink; /* per-sb shrinker handle */
1539
1540 /* Number of inodes with nlink == 0 but still referenced */
1541 atomic_long_t s_remove_count;
1542
1543 /* Being remounted read-only */
1544 int s_readonly_remount;
1545 };
1546
1547 /* superblock cache pruning functions */
1548 extern void prune_icache_sb(struct super_block *sb, int nr_to_scan);
1549 extern void prune_dcache_sb(struct super_block *sb, int nr_to_scan);
1550
1551 extern struct timespec current_fs_time(struct super_block *sb);
1552
1553 /*
1554 * Snapshotting support.
1555 */
1556 enum {
1557 SB_UNFROZEN = 0,
1558 SB_FREEZE_WRITE = 1,
1559 SB_FREEZE_TRANS = 2,
1560 };
1561
1562 #define vfs_check_frozen(sb, level) \
1563 wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
1564
1565 extern bool inode_owner_or_capable(const struct inode *inode);
1566
1567 /* not quite ready to be deprecated, but... */
1568 extern void lock_super(struct super_block *);
1569 extern void unlock_super(struct super_block *);
1570
1571 /*
1572 * VFS helper functions..
1573 */
1574 extern int vfs_create(struct inode *, struct dentry *, umode_t, struct nameidata *);
1575 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1576 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1577 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1578 extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
1579 extern int vfs_rmdir(struct inode *, struct dentry *);
1580 extern int vfs_unlink(struct inode *, struct dentry *);
1581 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1582
1583 /*
1584 * VFS dentry helper functions.
1585 */
1586 extern void dentry_unhash(struct dentry *dentry);
1587
1588 /*
1589 * VFS file helper functions.
1590 */
1591 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1592 umode_t mode);
1593 /*
1594 * VFS FS_IOC_FIEMAP helper definitions.
1595 */
1596 struct fiemap_extent_info {
1597 unsigned int fi_flags; /* Flags as passed from user */
1598 unsigned int fi_extents_mapped; /* Number of mapped extents */
1599 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1600 struct fiemap_extent __user *fi_extents_start; /* Start of
1601 fiemap_extent array */
1602 };
1603 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1604 u64 phys, u64 len, u32 flags);
1605 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1606
1607 /*
1608 * File types
1609 *
1610 * NOTE! These match bits 12..15 of stat.st_mode
1611 * (ie "(i_mode >> 12) & 15").
1612 */
1613 #define DT_UNKNOWN 0
1614 #define DT_FIFO 1
1615 #define DT_CHR 2
1616 #define DT_DIR 4
1617 #define DT_BLK 6
1618 #define DT_REG 8
1619 #define DT_LNK 10
1620 #define DT_SOCK 12
1621 #define DT_WHT 14
1622
1623 /*
1624 * This is the "filldir" function type, used by readdir() to let
1625 * the kernel specify what kind of dirent layout it wants to have.
1626 * This allows the kernel to read directories into kernel space or
1627 * to have different dirent layouts depending on the binary type.
1628 */
1629 typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
1630 struct block_device_operations;
1631
1632 /* These macros are for out of kernel modules to test that
1633 * the kernel supports the unlocked_ioctl and compat_ioctl
1634 * fields in struct file_operations. */
1635 #define HAVE_COMPAT_IOCTL 1
1636 #define HAVE_UNLOCKED_IOCTL 1
1637
1638 struct file_operations {
1639 struct module *owner;
1640 loff_t (*llseek) (struct file *, loff_t, int);
1641 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1642 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1643 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1644 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1645 int (*readdir) (struct file *, void *, filldir_t);
1646 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1647 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1648 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1649 int (*mmap) (struct file *, struct vm_area_struct *);
1650 int (*open) (struct inode *, struct file *);
1651 int (*flush) (struct file *, fl_owner_t id);
1652 int (*release) (struct inode *, struct file *);
1653 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1654 int (*aio_fsync) (struct kiocb *, int datasync);
1655 int (*fasync) (int, struct file *, int);
1656 int (*lock) (struct file *, int, struct file_lock *);
1657 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1658 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1659 int (*check_flags)(int);
1660 int (*flock) (struct file *, int, struct file_lock *);
1661 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1662 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1663 int (*setlease)(struct file *, long, struct file_lock **);
1664 long (*fallocate)(struct file *file, int mode, loff_t offset,
1665 loff_t len);
1666 };
1667
1668 struct inode_operations {
1669 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1670 void * (*follow_link) (struct dentry *, struct nameidata *);
1671 int (*permission) (struct inode *, int);
1672 struct posix_acl * (*get_acl)(struct inode *, int);
1673
1674 int (*readlink) (struct dentry *, char __user *,int);
1675 void (*put_link) (struct dentry *, struct nameidata *, void *);
1676
1677 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1678 int (*link) (struct dentry *,struct inode *,struct dentry *);
1679 int (*unlink) (struct inode *,struct dentry *);
1680 int (*symlink) (struct inode *,struct dentry *,const char *);
1681 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1682 int (*rmdir) (struct inode *,struct dentry *);
1683 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1684 int (*rename) (struct inode *, struct dentry *,
1685 struct inode *, struct dentry *);
1686 void (*truncate) (struct inode *);
1687 int (*setattr) (struct dentry *, struct iattr *);
1688 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1689 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1690 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1691 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1692 int (*removexattr) (struct dentry *, const char *);
1693 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1694 u64 len);
1695 int (*update_time)(struct inode *, struct timespec *, int);
1696 int (*atomic_open)(struct inode *, struct dentry *,
1697 struct file *, unsigned open_flag,
1698 umode_t create_mode, int *opened);
1699 } ____cacheline_aligned;
1700
1701 struct seq_file;
1702
1703 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1704 unsigned long nr_segs, unsigned long fast_segs,
1705 struct iovec *fast_pointer,
1706 struct iovec **ret_pointer);
1707
1708 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1709 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1710 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1711 unsigned long, loff_t *);
1712 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1713 unsigned long, loff_t *);
1714
1715 struct super_operations {
1716 struct inode *(*alloc_inode)(struct super_block *sb);
1717 void (*destroy_inode)(struct inode *);
1718
1719 void (*dirty_inode) (struct inode *, int flags);
1720 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1721 int (*drop_inode) (struct inode *);
1722 void (*evict_inode) (struct inode *);
1723 void (*put_super) (struct super_block *);
1724 void (*write_super) (struct super_block *);
1725 int (*sync_fs)(struct super_block *sb, int wait);
1726 int (*freeze_fs) (struct super_block *);
1727 int (*unfreeze_fs) (struct super_block *);
1728 int (*statfs) (struct dentry *, struct kstatfs *);
1729 int (*remount_fs) (struct super_block *, int *, char *);
1730 void (*umount_begin) (struct super_block *);
1731
1732 int (*show_options)(struct seq_file *, struct dentry *);
1733 int (*show_devname)(struct seq_file *, struct dentry *);
1734 int (*show_path)(struct seq_file *, struct dentry *);
1735 int (*show_stats)(struct seq_file *, struct dentry *);
1736 #ifdef CONFIG_QUOTA
1737 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1738 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1739 #endif
1740 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1741 int (*nr_cached_objects)(struct super_block *);
1742 void (*free_cached_objects)(struct super_block *, int);
1743 };
1744
1745 /*
1746 * Inode state bits. Protected by inode->i_lock
1747 *
1748 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1749 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1750 *
1751 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1752 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1753 * various stages of removing an inode.
1754 *
1755 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1756 *
1757 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1758 * fdatasync(). i_atime is the usual cause.
1759 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1760 * these changes separately from I_DIRTY_SYNC so that we
1761 * don't have to write inode on fdatasync() when only
1762 * mtime has changed in it.
1763 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1764 * I_NEW Serves as both a mutex and completion notification.
1765 * New inodes set I_NEW. If two processes both create
1766 * the same inode, one of them will release its inode and
1767 * wait for I_NEW to be released before returning.
1768 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1769 * also cause waiting on I_NEW, without I_NEW actually
1770 * being set. find_inode() uses this to prevent returning
1771 * nearly-dead inodes.
1772 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1773 * is zero. I_FREEING must be set when I_WILL_FREE is
1774 * cleared.
1775 * I_FREEING Set when inode is about to be freed but still has dirty
1776 * pages or buffers attached or the inode itself is still
1777 * dirty.
1778 * I_CLEAR Added by clear_inode(). In this state the inode is
1779 * clean and can be destroyed. Inode keeps I_FREEING.
1780 *
1781 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1782 * prohibited for many purposes. iget() must wait for
1783 * the inode to be completely released, then create it
1784 * anew. Other functions will just ignore such inodes,
1785 * if appropriate. I_NEW is used for waiting.
1786 *
1787 * I_SYNC Writeback of inode is running. The bit is set during
1788 * data writeback, and cleared with a wakeup on the bit
1789 * address once it is done. The bit is also used to pin
1790 * the inode in memory for flusher thread.
1791 *
1792 * I_REFERENCED Marks the inode as recently references on the LRU list.
1793 *
1794 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1795 *
1796 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1797 */
1798 #define I_DIRTY_SYNC (1 << 0)
1799 #define I_DIRTY_DATASYNC (1 << 1)
1800 #define I_DIRTY_PAGES (1 << 2)
1801 #define __I_NEW 3
1802 #define I_NEW (1 << __I_NEW)
1803 #define I_WILL_FREE (1 << 4)
1804 #define I_FREEING (1 << 5)
1805 #define I_CLEAR (1 << 6)
1806 #define __I_SYNC 7
1807 #define I_SYNC (1 << __I_SYNC)
1808 #define I_REFERENCED (1 << 8)
1809 #define __I_DIO_WAKEUP 9
1810 #define I_DIO_WAKEUP (1 << I_DIO_WAKEUP)
1811
1812 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1813
1814 extern void __mark_inode_dirty(struct inode *, int);
1815 static inline void mark_inode_dirty(struct inode *inode)
1816 {
1817 __mark_inode_dirty(inode, I_DIRTY);
1818 }
1819
1820 static inline void mark_inode_dirty_sync(struct inode *inode)
1821 {
1822 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1823 }
1824
1825 extern void inc_nlink(struct inode *inode);
1826 extern void drop_nlink(struct inode *inode);
1827 extern void clear_nlink(struct inode *inode);
1828 extern void set_nlink(struct inode *inode, unsigned int nlink);
1829
1830 static inline void inode_inc_link_count(struct inode *inode)
1831 {
1832 inc_nlink(inode);
1833 mark_inode_dirty(inode);
1834 }
1835
1836 static inline void inode_dec_link_count(struct inode *inode)
1837 {
1838 drop_nlink(inode);
1839 mark_inode_dirty(inode);
1840 }
1841
1842 /**
1843 * inode_inc_iversion - increments i_version
1844 * @inode: inode that need to be updated
1845 *
1846 * Every time the inode is modified, the i_version field will be incremented.
1847 * The filesystem has to be mounted with i_version flag
1848 */
1849
1850 static inline void inode_inc_iversion(struct inode *inode)
1851 {
1852 spin_lock(&inode->i_lock);
1853 inode->i_version++;
1854 spin_unlock(&inode->i_lock);
1855 }
1856
1857 enum file_time_flags {
1858 S_ATIME = 1,
1859 S_MTIME = 2,
1860 S_CTIME = 4,
1861 S_VERSION = 8,
1862 };
1863
1864 extern void touch_atime(struct path *);
1865 static inline void file_accessed(struct file *file)
1866 {
1867 if (!(file->f_flags & O_NOATIME))
1868 touch_atime(&file->f_path);
1869 }
1870
1871 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1872 int sync_inode_metadata(struct inode *inode, int wait);
1873
1874 struct file_system_type {
1875 const char *name;
1876 int fs_flags;
1877 struct dentry *(*mount) (struct file_system_type *, int,
1878 const char *, void *);
1879 void (*kill_sb) (struct super_block *);
1880 struct module *owner;
1881 struct file_system_type * next;
1882 struct hlist_head fs_supers;
1883
1884 struct lock_class_key s_lock_key;
1885 struct lock_class_key s_umount_key;
1886 struct lock_class_key s_vfs_rename_key;
1887
1888 struct lock_class_key i_lock_key;
1889 struct lock_class_key i_mutex_key;
1890 struct lock_class_key i_mutex_dir_key;
1891 };
1892
1893 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1894 void *data, int (*fill_super)(struct super_block *, void *, int));
1895 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1896 int flags, const char *dev_name, void *data,
1897 int (*fill_super)(struct super_block *, void *, int));
1898 extern struct dentry *mount_single(struct file_system_type *fs_type,
1899 int flags, void *data,
1900 int (*fill_super)(struct super_block *, void *, int));
1901 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1902 int flags, void *data,
1903 int (*fill_super)(struct super_block *, void *, int));
1904 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
1905 void generic_shutdown_super(struct super_block *sb);
1906 void kill_block_super(struct super_block *sb);
1907 void kill_anon_super(struct super_block *sb);
1908 void kill_litter_super(struct super_block *sb);
1909 void deactivate_super(struct super_block *sb);
1910 void deactivate_locked_super(struct super_block *sb);
1911 int set_anon_super(struct super_block *s, void *data);
1912 int get_anon_bdev(dev_t *);
1913 void free_anon_bdev(dev_t);
1914 struct super_block *sget(struct file_system_type *type,
1915 int (*test)(struct super_block *,void *),
1916 int (*set)(struct super_block *,void *),
1917 void *data);
1918 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
1919 const struct super_operations *ops,
1920 const struct dentry_operations *dops,
1921 unsigned long);
1922
1923 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1924 #define fops_get(fops) \
1925 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1926 #define fops_put(fops) \
1927 do { if (fops) module_put((fops)->owner); } while(0)
1928
1929 extern int register_filesystem(struct file_system_type *);
1930 extern int unregister_filesystem(struct file_system_type *);
1931 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1932 #define kern_mount(type) kern_mount_data(type, NULL)
1933 extern void kern_unmount(struct vfsmount *mnt);
1934 extern int may_umount_tree(struct vfsmount *);
1935 extern int may_umount(struct vfsmount *);
1936 extern long do_mount(char *, char *, char *, unsigned long, void *);
1937 extern struct vfsmount *collect_mounts(struct path *);
1938 extern void drop_collected_mounts(struct vfsmount *);
1939 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1940 struct vfsmount *);
1941 extern int vfs_statfs(struct path *, struct kstatfs *);
1942 extern int user_statfs(const char __user *, struct kstatfs *);
1943 extern int fd_statfs(int, struct kstatfs *);
1944 extern int vfs_ustat(dev_t, struct kstatfs *);
1945 extern int freeze_super(struct super_block *super);
1946 extern int thaw_super(struct super_block *super);
1947 extern bool our_mnt(struct vfsmount *mnt);
1948
1949 extern int current_umask(void);
1950
1951 /* /sys/fs */
1952 extern struct kobject *fs_kobj;
1953
1954 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
1955 extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1956
1957 #define FLOCK_VERIFY_READ 1
1958 #define FLOCK_VERIFY_WRITE 2
1959
1960 #ifdef CONFIG_FILE_LOCKING
1961 extern int locks_mandatory_locked(struct inode *);
1962 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1963
1964 /*
1965 * Candidates for mandatory locking have the setgid bit set
1966 * but no group execute bit - an otherwise meaningless combination.
1967 */
1968
1969 static inline int __mandatory_lock(struct inode *ino)
1970 {
1971 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1972 }
1973
1974 /*
1975 * ... and these candidates should be on MS_MANDLOCK mounted fs,
1976 * otherwise these will be advisory locks
1977 */
1978
1979 static inline int mandatory_lock(struct inode *ino)
1980 {
1981 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1982 }
1983
1984 static inline int locks_verify_locked(struct inode *inode)
1985 {
1986 if (mandatory_lock(inode))
1987 return locks_mandatory_locked(inode);
1988 return 0;
1989 }
1990
1991 static inline int locks_verify_truncate(struct inode *inode,
1992 struct file *filp,
1993 loff_t size)
1994 {
1995 if (inode->i_flock && mandatory_lock(inode))
1996 return locks_mandatory_area(
1997 FLOCK_VERIFY_WRITE, inode, filp,
1998 size < inode->i_size ? size : inode->i_size,
1999 (size < inode->i_size ? inode->i_size - size
2000 : size - inode->i_size)
2001 );
2002 return 0;
2003 }
2004
2005 static inline int break_lease(struct inode *inode, unsigned int mode)
2006 {
2007 if (inode->i_flock)
2008 return __break_lease(inode, mode);
2009 return 0;
2010 }
2011 #else /* !CONFIG_FILE_LOCKING */
2012 static inline int locks_mandatory_locked(struct inode *inode)
2013 {
2014 return 0;
2015 }
2016
2017 static inline int locks_mandatory_area(int rw, struct inode *inode,
2018 struct file *filp, loff_t offset,
2019 size_t count)
2020 {
2021 return 0;
2022 }
2023
2024 static inline int __mandatory_lock(struct inode *inode)
2025 {
2026 return 0;
2027 }
2028
2029 static inline int mandatory_lock(struct inode *inode)
2030 {
2031 return 0;
2032 }
2033
2034 static inline int locks_verify_locked(struct inode *inode)
2035 {
2036 return 0;
2037 }
2038
2039 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2040 size_t size)
2041 {
2042 return 0;
2043 }
2044
2045 static inline int break_lease(struct inode *inode, unsigned int mode)
2046 {
2047 return 0;
2048 }
2049
2050 #endif /* CONFIG_FILE_LOCKING */
2051
2052 /* fs/open.c */
2053
2054 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2055 struct file *filp);
2056 extern int do_fallocate(struct file *file, int mode, loff_t offset,
2057 loff_t len);
2058 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2059 umode_t mode);
2060 extern struct file *filp_open(const char *, int, umode_t);
2061 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2062 const char *, int);
2063 extern struct file * dentry_open(struct dentry *, struct vfsmount *, int,
2064 const struct cred *);
2065 extern int filp_close(struct file *, fl_owner_t id);
2066 extern char * getname(const char __user *);
2067 enum {
2068 FILE_CREATED = 1,
2069 FILE_OPENED = 2
2070 };
2071 extern int finish_open(struct file *file, struct dentry *dentry,
2072 int (*open)(struct inode *, struct file *),
2073 int *opened);
2074 extern int finish_no_open(struct file *file, struct dentry *dentry);
2075
2076 /* fs/ioctl.c */
2077
2078 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2079
2080 /* fs/dcache.c */
2081 extern void __init vfs_caches_init_early(void);
2082 extern void __init vfs_caches_init(unsigned long);
2083
2084 extern struct kmem_cache *names_cachep;
2085
2086 #define __getname_gfp(gfp) kmem_cache_alloc(names_cachep, (gfp))
2087 #define __getname() __getname_gfp(GFP_KERNEL)
2088 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2089 #ifndef CONFIG_AUDITSYSCALL
2090 #define putname(name) __putname(name)
2091 #else
2092 extern void putname(const char *name);
2093 #endif
2094
2095 #ifdef CONFIG_BLOCK
2096 extern int register_blkdev(unsigned int, const char *);
2097 extern void unregister_blkdev(unsigned int, const char *);
2098 extern struct block_device *bdget(dev_t);
2099 extern struct block_device *bdgrab(struct block_device *bdev);
2100 extern void bd_set_size(struct block_device *, loff_t size);
2101 extern sector_t blkdev_max_block(struct block_device *bdev);
2102 extern void bd_forget(struct inode *inode);
2103 extern void bdput(struct block_device *);
2104 extern void invalidate_bdev(struct block_device *);
2105 extern int sync_blockdev(struct block_device *bdev);
2106 extern void kill_bdev(struct block_device *);
2107 extern struct super_block *freeze_bdev(struct block_device *);
2108 extern void emergency_thaw_all(void);
2109 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2110 extern int fsync_bdev(struct block_device *);
2111 #else
2112 static inline void bd_forget(struct inode *inode) {}
2113 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2114 static inline void kill_bdev(struct block_device *bdev) {}
2115 static inline void invalidate_bdev(struct block_device *bdev) {}
2116
2117 static inline struct super_block *freeze_bdev(struct block_device *sb)
2118 {
2119 return NULL;
2120 }
2121
2122 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2123 {
2124 return 0;
2125 }
2126 #endif
2127 extern int sync_filesystem(struct super_block *);
2128 extern const struct file_operations def_blk_fops;
2129 extern const struct file_operations def_chr_fops;
2130 extern const struct file_operations bad_sock_fops;
2131 extern const struct file_operations def_fifo_fops;
2132 #ifdef CONFIG_BLOCK
2133 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2134 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2135 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2136 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2137 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2138 void *holder);
2139 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2140 void *holder);
2141 extern int blkdev_put(struct block_device *bdev, fmode_t mode);
2142 #ifdef CONFIG_SYSFS
2143 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2144 extern void bd_unlink_disk_holder(struct block_device *bdev,
2145 struct gendisk *disk);
2146 #else
2147 static inline int bd_link_disk_holder(struct block_device *bdev,
2148 struct gendisk *disk)
2149 {
2150 return 0;
2151 }
2152 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2153 struct gendisk *disk)
2154 {
2155 }
2156 #endif
2157 #endif
2158
2159 /* fs/char_dev.c */
2160 #define CHRDEV_MAJOR_HASH_SIZE 255
2161 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2162 extern int register_chrdev_region(dev_t, unsigned, const char *);
2163 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2164 unsigned int count, const char *name,
2165 const struct file_operations *fops);
2166 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2167 unsigned int count, const char *name);
2168 extern void unregister_chrdev_region(dev_t, unsigned);
2169 extern void chrdev_show(struct seq_file *,off_t);
2170
2171 static inline int register_chrdev(unsigned int major, const char *name,
2172 const struct file_operations *fops)
2173 {
2174 return __register_chrdev(major, 0, 256, name, fops);
2175 }
2176
2177 static inline void unregister_chrdev(unsigned int major, const char *name)
2178 {
2179 __unregister_chrdev(major, 0, 256, name);
2180 }
2181
2182 /* fs/block_dev.c */
2183 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2184 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2185
2186 #ifdef CONFIG_BLOCK
2187 #define BLKDEV_MAJOR_HASH_SIZE 255
2188 extern const char *__bdevname(dev_t, char *buffer);
2189 extern const char *bdevname(struct block_device *bdev, char *buffer);
2190 extern struct block_device *lookup_bdev(const char *);
2191 extern void blkdev_show(struct seq_file *,off_t);
2192
2193 #else
2194 #define BLKDEV_MAJOR_HASH_SIZE 0
2195 #endif
2196
2197 extern void init_special_inode(struct inode *, umode_t, dev_t);
2198
2199 /* Invalid inode operations -- fs/bad_inode.c */
2200 extern void make_bad_inode(struct inode *);
2201 extern int is_bad_inode(struct inode *);
2202
2203 extern const struct file_operations read_pipefifo_fops;
2204 extern const struct file_operations write_pipefifo_fops;
2205 extern const struct file_operations rdwr_pipefifo_fops;
2206
2207 #ifdef CONFIG_BLOCK
2208 /*
2209 * return READ, READA, or WRITE
2210 */
2211 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2212
2213 /*
2214 * return data direction, READ or WRITE
2215 */
2216 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2217
2218 extern void check_disk_size_change(struct gendisk *disk,
2219 struct block_device *bdev);
2220 extern int revalidate_disk(struct gendisk *);
2221 extern int check_disk_change(struct block_device *);
2222 extern int __invalidate_device(struct block_device *, bool);
2223 extern int invalidate_partition(struct gendisk *, int);
2224 #endif
2225 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2226 pgoff_t start, pgoff_t end);
2227
2228 static inline void invalidate_remote_inode(struct inode *inode)
2229 {
2230 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2231 S_ISLNK(inode->i_mode))
2232 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2233 }
2234 extern int invalidate_inode_pages2(struct address_space *mapping);
2235 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2236 pgoff_t start, pgoff_t end);
2237 extern int write_inode_now(struct inode *, int);
2238 extern int filemap_fdatawrite(struct address_space *);
2239 extern int filemap_flush(struct address_space *);
2240 extern int filemap_fdatawait(struct address_space *);
2241 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2242 loff_t lend);
2243 extern int filemap_write_and_wait(struct address_space *mapping);
2244 extern int filemap_write_and_wait_range(struct address_space *mapping,
2245 loff_t lstart, loff_t lend);
2246 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2247 loff_t start, loff_t end, int sync_mode);
2248 extern int filemap_fdatawrite_range(struct address_space *mapping,
2249 loff_t start, loff_t end);
2250
2251 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2252 int datasync);
2253 extern int vfs_fsync(struct file *file, int datasync);
2254 extern int generic_write_sync(struct file *file, loff_t pos, loff_t count);
2255 extern void sync_supers(void);
2256 extern void emergency_sync(void);
2257 extern void emergency_remount(void);
2258 #ifdef CONFIG_BLOCK
2259 extern sector_t bmap(struct inode *, sector_t);
2260 #endif
2261 extern int notify_change(struct dentry *, struct iattr *);
2262 extern int inode_permission(struct inode *, int);
2263 extern int generic_permission(struct inode *, int);
2264
2265 static inline bool execute_ok(struct inode *inode)
2266 {
2267 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2268 }
2269
2270 /*
2271 * get_write_access() gets write permission for a file.
2272 * put_write_access() releases this write permission.
2273 * This is used for regular files.
2274 * We cannot support write (and maybe mmap read-write shared) accesses and
2275 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2276 * can have the following values:
2277 * 0: no writers, no VM_DENYWRITE mappings
2278 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2279 * > 0: (i_writecount) users are writing to the file.
2280 *
2281 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2282 * except for the cases where we don't hold i_writecount yet. Then we need to
2283 * use {get,deny}_write_access() - these functions check the sign and refuse
2284 * to do the change if sign is wrong.
2285 */
2286 static inline int get_write_access(struct inode *inode)
2287 {
2288 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2289 }
2290 static inline int deny_write_access(struct file *file)
2291 {
2292 struct inode *inode = file->f_path.dentry->d_inode;
2293 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2294 }
2295 static inline void put_write_access(struct inode * inode)
2296 {
2297 atomic_dec(&inode->i_writecount);
2298 }
2299 static inline void allow_write_access(struct file *file)
2300 {
2301 if (file)
2302 atomic_inc(&file->f_path.dentry->d_inode->i_writecount);
2303 }
2304 #ifdef CONFIG_IMA
2305 static inline void i_readcount_dec(struct inode *inode)
2306 {
2307 BUG_ON(!atomic_read(&inode->i_readcount));
2308 atomic_dec(&inode->i_readcount);
2309 }
2310 static inline void i_readcount_inc(struct inode *inode)
2311 {
2312 atomic_inc(&inode->i_readcount);
2313 }
2314 #else
2315 static inline void i_readcount_dec(struct inode *inode)
2316 {
2317 return;
2318 }
2319 static inline void i_readcount_inc(struct inode *inode)
2320 {
2321 return;
2322 }
2323 #endif
2324 extern int do_pipe_flags(int *, int);
2325 extern struct file *create_read_pipe(struct file *f, int flags);
2326 extern struct file *create_write_pipe(int flags);
2327 extern void free_write_pipe(struct file *);
2328
2329 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2330 extern struct file * open_exec(const char *);
2331
2332 /* fs/dcache.c -- generic fs support functions */
2333 extern int is_subdir(struct dentry *, struct dentry *);
2334 extern int path_is_under(struct path *, struct path *);
2335 extern ino_t find_inode_number(struct dentry *, struct qstr *);
2336
2337 #include <linux/err.h>
2338
2339 /* needed for stackable file system support */
2340 extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
2341
2342 extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
2343
2344 extern int inode_init_always(struct super_block *, struct inode *);
2345 extern void inode_init_once(struct inode *);
2346 extern void address_space_init_once(struct address_space *mapping);
2347 extern void ihold(struct inode * inode);
2348 extern void iput(struct inode *);
2349 extern struct inode * igrab(struct inode *);
2350 extern ino_t iunique(struct super_block *, ino_t);
2351 extern int inode_needs_sync(struct inode *inode);
2352 extern int generic_delete_inode(struct inode *inode);
2353 static inline int generic_drop_inode(struct inode *inode)
2354 {
2355 return !inode->i_nlink || inode_unhashed(inode);
2356 }
2357
2358 extern struct inode *ilookup5_nowait(struct super_block *sb,
2359 unsigned long hashval, int (*test)(struct inode *, void *),
2360 void *data);
2361 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2362 int (*test)(struct inode *, void *), void *data);
2363 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2364
2365 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2366 extern struct inode * iget_locked(struct super_block *, unsigned long);
2367 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2368 extern int insert_inode_locked(struct inode *);
2369 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2370 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2371 #else
2372 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2373 #endif
2374 extern void unlock_new_inode(struct inode *);
2375 extern unsigned int get_next_ino(void);
2376
2377 extern void __iget(struct inode * inode);
2378 extern void iget_failed(struct inode *);
2379 extern void clear_inode(struct inode *);
2380 extern void __destroy_inode(struct inode *);
2381 extern struct inode *new_inode_pseudo(struct super_block *sb);
2382 extern struct inode *new_inode(struct super_block *sb);
2383 extern void free_inode_nonrcu(struct inode *inode);
2384 extern int should_remove_suid(struct dentry *);
2385 extern int file_remove_suid(struct file *);
2386
2387 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2388 static inline void insert_inode_hash(struct inode *inode)
2389 {
2390 __insert_inode_hash(inode, inode->i_ino);
2391 }
2392
2393 extern void __remove_inode_hash(struct inode *);
2394 static inline void remove_inode_hash(struct inode *inode)
2395 {
2396 if (!inode_unhashed(inode))
2397 __remove_inode_hash(inode);
2398 }
2399
2400 extern void inode_sb_list_add(struct inode *inode);
2401
2402 #ifdef CONFIG_BLOCK
2403 extern void submit_bio(int, struct bio *);
2404 extern int bdev_read_only(struct block_device *);
2405 #endif
2406 extern int set_blocksize(struct block_device *, int);
2407 extern int sb_set_blocksize(struct super_block *, int);
2408 extern int sb_min_blocksize(struct super_block *, int);
2409
2410 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2411 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2412 extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
2413 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
2414 extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2415 extern ssize_t __generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long,
2416 loff_t *);
2417 extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2418 extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
2419 unsigned long *, loff_t, loff_t *, size_t, size_t);
2420 extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
2421 unsigned long, loff_t, loff_t *, size_t, ssize_t);
2422 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2423 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2424 extern int generic_segment_checks(const struct iovec *iov,
2425 unsigned long *nr_segs, size_t *count, int access_flags);
2426
2427 /* fs/block_dev.c */
2428 extern ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov,
2429 unsigned long nr_segs, loff_t pos);
2430 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2431 int datasync);
2432 extern void block_sync_page(struct page *page);
2433
2434 /* fs/splice.c */
2435 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2436 struct pipe_inode_info *, size_t, unsigned int);
2437 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2438 struct pipe_inode_info *, size_t, unsigned int);
2439 extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
2440 struct file *, loff_t *, size_t, unsigned int);
2441 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2442 struct file *out, loff_t *, size_t len, unsigned int flags);
2443 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2444 size_t len, unsigned int flags);
2445
2446 extern void
2447 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2448 extern loff_t noop_llseek(struct file *file, loff_t offset, int origin);
2449 extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
2450 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
2451 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2452 int origin, loff_t maxsize);
2453 extern int generic_file_open(struct inode * inode, struct file * filp);
2454 extern int nonseekable_open(struct inode * inode, struct file * filp);
2455
2456 #ifdef CONFIG_FS_XIP
2457 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2458 loff_t *ppos);
2459 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
2460 extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2461 size_t len, loff_t *ppos);
2462 extern int xip_truncate_page(struct address_space *mapping, loff_t from);
2463 #else
2464 static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2465 {
2466 return 0;
2467 }
2468 #endif
2469
2470 #ifdef CONFIG_BLOCK
2471 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2472 loff_t file_offset);
2473
2474 enum {
2475 /* need locking between buffered and direct access */
2476 DIO_LOCKING = 0x01,
2477
2478 /* filesystem does not support filling holes */
2479 DIO_SKIP_HOLES = 0x02,
2480 };
2481
2482 void dio_end_io(struct bio *bio, int error);
2483
2484 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2485 struct block_device *bdev, const struct iovec *iov, loff_t offset,
2486 unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
2487 dio_submit_t submit_io, int flags);
2488
2489 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
2490 struct inode *inode, const struct iovec *iov, loff_t offset,
2491 unsigned long nr_segs, get_block_t get_block)
2492 {
2493 return __blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov,
2494 offset, nr_segs, get_block, NULL, NULL,
2495 DIO_LOCKING | DIO_SKIP_HOLES);
2496 }
2497 #endif
2498
2499 void inode_dio_wait(struct inode *inode);
2500 void inode_dio_done(struct inode *inode);
2501
2502 extern const struct file_operations generic_ro_fops;
2503
2504 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2505
2506 extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
2507 extern int vfs_follow_link(struct nameidata *, const char *);
2508 extern int page_readlink(struct dentry *, char __user *, int);
2509 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2510 extern void page_put_link(struct dentry *, struct nameidata *, void *);
2511 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2512 int nofs);
2513 extern int page_symlink(struct inode *inode, const char *symname, int len);
2514 extern const struct inode_operations page_symlink_inode_operations;
2515 extern int generic_readlink(struct dentry *, char __user *, int);
2516 extern void generic_fillattr(struct inode *, struct kstat *);
2517 extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2518 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2519 void inode_add_bytes(struct inode *inode, loff_t bytes);
2520 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2521 loff_t inode_get_bytes(struct inode *inode);
2522 void inode_set_bytes(struct inode *inode, loff_t bytes);
2523
2524 extern int vfs_readdir(struct file *, filldir_t, void *);
2525
2526 extern int vfs_stat(const char __user *, struct kstat *);
2527 extern int vfs_lstat(const char __user *, struct kstat *);
2528 extern int vfs_fstat(unsigned int, struct kstat *);
2529 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2530
2531 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2532 unsigned long arg);
2533 extern int __generic_block_fiemap(struct inode *inode,
2534 struct fiemap_extent_info *fieinfo,
2535 loff_t start, loff_t len,
2536 get_block_t *get_block);
2537 extern int generic_block_fiemap(struct inode *inode,
2538 struct fiemap_extent_info *fieinfo, u64 start,
2539 u64 len, get_block_t *get_block);
2540
2541 extern void get_filesystem(struct file_system_type *fs);
2542 extern void put_filesystem(struct file_system_type *fs);
2543 extern struct file_system_type *get_fs_type(const char *name);
2544 extern struct super_block *get_super(struct block_device *);
2545 extern struct super_block *get_super_thawed(struct block_device *);
2546 extern struct super_block *get_active_super(struct block_device *bdev);
2547 extern void drop_super(struct super_block *sb);
2548 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2549 extern void iterate_supers_type(struct file_system_type *,
2550 void (*)(struct super_block *, void *), void *);
2551
2552 extern int dcache_dir_open(struct inode *, struct file *);
2553 extern int dcache_dir_close(struct inode *, struct file *);
2554 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2555 extern int dcache_readdir(struct file *, void *, filldir_t);
2556 extern int simple_setattr(struct dentry *, struct iattr *);
2557 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2558 extern int simple_statfs(struct dentry *, struct kstatfs *);
2559 extern int simple_open(struct inode *inode, struct file *file);
2560 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2561 extern int simple_unlink(struct inode *, struct dentry *);
2562 extern int simple_rmdir(struct inode *, struct dentry *);
2563 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2564 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2565 extern int simple_empty(struct dentry *);
2566 extern int simple_readpage(struct file *file, struct page *page);
2567 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2568 loff_t pos, unsigned len, unsigned flags,
2569 struct page **pagep, void **fsdata);
2570 extern int simple_write_end(struct file *file, struct address_space *mapping,
2571 loff_t pos, unsigned len, unsigned copied,
2572 struct page *page, void *fsdata);
2573
2574 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
2575 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2576 extern const struct file_operations simple_dir_operations;
2577 extern const struct inode_operations simple_dir_inode_operations;
2578 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2579 struct dentry *d_alloc_name(struct dentry *, const char *);
2580 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2581 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2582 extern void simple_release_fs(struct vfsmount **mount, int *count);
2583
2584 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2585 loff_t *ppos, const void *from, size_t available);
2586 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2587 const void __user *from, size_t count);
2588
2589 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2590
2591 extern int generic_check_addressable(unsigned, u64);
2592
2593 #ifdef CONFIG_MIGRATION
2594 extern int buffer_migrate_page(struct address_space *,
2595 struct page *, struct page *,
2596 enum migrate_mode);
2597 #else
2598 #define buffer_migrate_page NULL
2599 #endif
2600
2601 extern int inode_change_ok(const struct inode *, struct iattr *);
2602 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2603 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2604
2605 extern int file_update_time(struct file *file);
2606
2607 extern int generic_show_options(struct seq_file *m, struct dentry *root);
2608 extern void save_mount_options(struct super_block *sb, char *options);
2609 extern void replace_mount_options(struct super_block *sb, char *options);
2610
2611 static inline ino_t parent_ino(struct dentry *dentry)
2612 {
2613 ino_t res;
2614
2615 /*
2616 * Don't strictly need d_lock here? If the parent ino could change
2617 * then surely we'd have a deeper race in the caller?
2618 */
2619 spin_lock(&dentry->d_lock);
2620 res = dentry->d_parent->d_inode->i_ino;
2621 spin_unlock(&dentry->d_lock);
2622 return res;
2623 }
2624
2625 /* Transaction based IO helpers */
2626
2627 /*
2628 * An argresp is stored in an allocated page and holds the
2629 * size of the argument or response, along with its content
2630 */
2631 struct simple_transaction_argresp {
2632 ssize_t size;
2633 char data[0];
2634 };
2635
2636 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2637
2638 char *simple_transaction_get(struct file *file, const char __user *buf,
2639 size_t size);
2640 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2641 size_t size, loff_t *pos);
2642 int simple_transaction_release(struct inode *inode, struct file *file);
2643
2644 void simple_transaction_set(struct file *file, size_t n);
2645
2646 /*
2647 * simple attribute files
2648 *
2649 * These attributes behave similar to those in sysfs:
2650 *
2651 * Writing to an attribute immediately sets a value, an open file can be
2652 * written to multiple times.
2653 *
2654 * Reading from an attribute creates a buffer from the value that might get
2655 * read with multiple read calls. When the attribute has been read
2656 * completely, no further read calls are possible until the file is opened
2657 * again.
2658 *
2659 * All attributes contain a text representation of a numeric value
2660 * that are accessed with the get() and set() functions.
2661 */
2662 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2663 static int __fops ## _open(struct inode *inode, struct file *file) \
2664 { \
2665 __simple_attr_check_format(__fmt, 0ull); \
2666 return simple_attr_open(inode, file, __get, __set, __fmt); \
2667 } \
2668 static const struct file_operations __fops = { \
2669 .owner = THIS_MODULE, \
2670 .open = __fops ## _open, \
2671 .release = simple_attr_release, \
2672 .read = simple_attr_read, \
2673 .write = simple_attr_write, \
2674 .llseek = generic_file_llseek, \
2675 };
2676
2677 static inline __printf(1, 2)
2678 void __simple_attr_check_format(const char *fmt, ...)
2679 {
2680 /* don't do anything, just let the compiler check the arguments; */
2681 }
2682
2683 int simple_attr_open(struct inode *inode, struct file *file,
2684 int (*get)(void *, u64 *), int (*set)(void *, u64),
2685 const char *fmt);
2686 int simple_attr_release(struct inode *inode, struct file *file);
2687 ssize_t simple_attr_read(struct file *file, char __user *buf,
2688 size_t len, loff_t *ppos);
2689 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2690 size_t len, loff_t *ppos);
2691
2692 struct ctl_table;
2693 int proc_nr_files(struct ctl_table *table, int write,
2694 void __user *buffer, size_t *lenp, loff_t *ppos);
2695 int proc_nr_dentry(struct ctl_table *table, int write,
2696 void __user *buffer, size_t *lenp, loff_t *ppos);
2697 int proc_nr_inodes(struct ctl_table *table, int write,
2698 void __user *buffer, size_t *lenp, loff_t *ppos);
2699 int __init get_filesystem_list(char *buf);
2700
2701 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
2702 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
2703
2704 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2705 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2706 (flag & __FMODE_NONOTIFY)))
2707
2708 static inline int is_sxid(umode_t mode)
2709 {
2710 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
2711 }
2712
2713 static inline void inode_has_no_xattr(struct inode *inode)
2714 {
2715 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
2716 inode->i_flags |= S_NOSEC;
2717 }
2718
2719 #endif /* __KERNEL__ */
2720 #endif /* _LINUX_FS_H */