Merge branch 'depends/rmk/memory_h' into next/cleanup2
[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 1
62 #define MAY_WRITE 2
63 #define MAY_READ 4
64 #define MAY_APPEND 8
65 #define MAY_ACCESS 16
66 #define MAY_OPEN 32
67 #define MAY_CHDIR 64
68 #define MAY_NOT_BLOCK 128 /* called from RCU mode, don't block */
69
70 /*
71 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
72 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
73 */
74
75 /* file is open for reading */
76 #define FMODE_READ ((__force fmode_t)0x1)
77 /* file is open for writing */
78 #define FMODE_WRITE ((__force fmode_t)0x2)
79 /* file is seekable */
80 #define FMODE_LSEEK ((__force fmode_t)0x4)
81 /* file can be accessed using pread */
82 #define FMODE_PREAD ((__force fmode_t)0x8)
83 /* file can be accessed using pwrite */
84 #define FMODE_PWRITE ((__force fmode_t)0x10)
85 /* File is opened for execution with sys_execve / sys_uselib */
86 #define FMODE_EXEC ((__force fmode_t)0x20)
87 /* File is opened with O_NDELAY (only set for block devices) */
88 #define FMODE_NDELAY ((__force fmode_t)0x40)
89 /* File is opened with O_EXCL (only set for block devices) */
90 #define FMODE_EXCL ((__force fmode_t)0x80)
91 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
92 (specialy hack for floppy.c) */
93 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
94
95 /*
96 * Don't update ctime and mtime.
97 *
98 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
99 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
100 */
101 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
102
103 /* Expect random access pattern */
104 #define FMODE_RANDOM ((__force fmode_t)0x1000)
105
106 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
107 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
108
109 /* File is opened with O_PATH; almost nothing can be done with it */
110 #define FMODE_PATH ((__force fmode_t)0x4000)
111
112 /* File was opened by fanotify and shouldn't generate fanotify events */
113 #define FMODE_NONOTIFY ((__force fmode_t)0x1000000)
114
115 /*
116 * The below are the various read and write types that we support. Some of
117 * them include behavioral modifiers that send information down to the
118 * block layer and IO scheduler. Terminology:
119 *
120 * The block layer uses device plugging to defer IO a little bit, in
121 * the hope that we will see more IO very shortly. This increases
122 * coalescing of adjacent IO and thus reduces the number of IOs we
123 * have to send to the device. It also allows for better queuing,
124 * if the IO isn't mergeable. If the caller is going to be waiting
125 * for the IO, then he must ensure that the device is unplugged so
126 * that the IO is dispatched to the driver.
127 *
128 * All IO is handled async in Linux. This is fine for background
129 * writes, but for reads or writes that someone waits for completion
130 * on, we want to notify the block layer and IO scheduler so that they
131 * know about it. That allows them to make better scheduling
132 * decisions. So when the below references 'sync' and 'async', it
133 * is referencing this priority hint.
134 *
135 * With that in mind, the available types are:
136 *
137 * READ A normal read operation. Device will be plugged.
138 * READ_SYNC A synchronous read. Device is not plugged, caller can
139 * immediately wait on this read without caring about
140 * unplugging.
141 * READA Used for read-ahead operations. Lower priority, and the
142 * block layer could (in theory) choose to ignore this
143 * request if it runs into resource problems.
144 * WRITE A normal async write. Device will be plugged.
145 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
146 * the hint that someone will be waiting on this IO
147 * shortly. The write equivalent of READ_SYNC.
148 * WRITE_ODIRECT Special case write for O_DIRECT only.
149 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
150 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
151 * non-volatile media on completion.
152 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
153 * by a cache flush and data is guaranteed to be on
154 * non-volatile media on completion.
155 *
156 */
157 #define RW_MASK REQ_WRITE
158 #define RWA_MASK REQ_RAHEAD
159
160 #define READ 0
161 #define WRITE RW_MASK
162 #define READA RWA_MASK
163
164 #define READ_SYNC (READ | REQ_SYNC)
165 #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
166 #define WRITE_ODIRECT (WRITE | REQ_SYNC)
167 #define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
168 #define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
169 #define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
170
171 #define SEL_IN 1
172 #define SEL_OUT 2
173 #define SEL_EX 4
174
175 /* public flags for file_system_type */
176 #define FS_REQUIRES_DEV 1
177 #define FS_BINARY_MOUNTDATA 2
178 #define FS_HAS_SUBTYPE 4
179 #define FS_REVAL_DOT 16384 /* Check the paths ".", ".." for staleness */
180 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move()
181 * during rename() internally.
182 */
183
184 /*
185 * These are the fs-independent mount-flags: up to 32 flags are supported
186 */
187 #define MS_RDONLY 1 /* Mount read-only */
188 #define MS_NOSUID 2 /* Ignore suid and sgid bits */
189 #define MS_NODEV 4 /* Disallow access to device special files */
190 #define MS_NOEXEC 8 /* Disallow program execution */
191 #define MS_SYNCHRONOUS 16 /* Writes are synced at once */
192 #define MS_REMOUNT 32 /* Alter flags of a mounted FS */
193 #define MS_MANDLOCK 64 /* Allow mandatory locks on an FS */
194 #define MS_DIRSYNC 128 /* Directory modifications are synchronous */
195 #define MS_NOATIME 1024 /* Do not update access times. */
196 #define MS_NODIRATIME 2048 /* Do not update directory access times */
197 #define MS_BIND 4096
198 #define MS_MOVE 8192
199 #define MS_REC 16384
200 #define MS_VERBOSE 32768 /* War is peace. Verbosity is silence.
201 MS_VERBOSE is deprecated. */
202 #define MS_SILENT 32768
203 #define MS_POSIXACL (1<<16) /* VFS does not apply the umask */
204 #define MS_UNBINDABLE (1<<17) /* change to unbindable */
205 #define MS_PRIVATE (1<<18) /* change to private */
206 #define MS_SLAVE (1<<19) /* change to slave */
207 #define MS_SHARED (1<<20) /* change to shared */
208 #define MS_RELATIME (1<<21) /* Update atime relative to mtime/ctime. */
209 #define MS_KERNMOUNT (1<<22) /* this is a kern_mount call */
210 #define MS_I_VERSION (1<<23) /* Update inode I_version field */
211 #define MS_STRICTATIME (1<<24) /* Always perform atime updates */
212 #define MS_NOSEC (1<<28)
213 #define MS_BORN (1<<29)
214 #define MS_ACTIVE (1<<30)
215 #define MS_NOUSER (1<<31)
216
217 /*
218 * Superblock flags that can be altered by MS_REMOUNT
219 */
220 #define MS_RMT_MASK (MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK|MS_I_VERSION)
221
222 /*
223 * Old magic mount flag and mask
224 */
225 #define MS_MGC_VAL 0xC0ED0000
226 #define MS_MGC_MSK 0xffff0000
227
228 /* Inode flags - they have nothing to superblock flags now */
229
230 #define S_SYNC 1 /* Writes are synced at once */
231 #define S_NOATIME 2 /* Do not update access times */
232 #define S_APPEND 4 /* Append-only file */
233 #define S_IMMUTABLE 8 /* Immutable file */
234 #define S_DEAD 16 /* removed, but still open directory */
235 #define S_NOQUOTA 32 /* Inode is not counted to quota */
236 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
237 #define S_NOCMTIME 128 /* Do not update file c/mtime */
238 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
239 #define S_PRIVATE 512 /* Inode is fs-internal */
240 #define S_IMA 1024 /* Inode has an associated IMA struct */
241 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
242 #define S_NOSEC 4096 /* no suid or xattr security attributes */
243
244 /*
245 * Note that nosuid etc flags are inode-specific: setting some file-system
246 * flags just means all the inodes inherit those flags by default. It might be
247 * possible to override it selectively if you really wanted to with some
248 * ioctl() that is not currently implemented.
249 *
250 * Exception: MS_RDONLY is always applied to the entire file system.
251 *
252 * Unfortunately, it is possible to change a filesystems flags with it mounted
253 * with files in use. This means that all of the inodes will not have their
254 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
255 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
256 */
257 #define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))
258
259 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
260 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
261 ((inode)->i_flags & S_SYNC))
262 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
263 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
264 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
265 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
266 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
267
268 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
269 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
270 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
271 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
272
273 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
274 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
275 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
276 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
277 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
278 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
279 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
280
281 /* the read-only stuff doesn't really belong here, but any other place is
282 probably as bad and I don't want to create yet another include file. */
283
284 #define BLKROSET _IO(0x12,93) /* set device read-only (0 = read-write) */
285 #define BLKROGET _IO(0x12,94) /* get read-only status (0 = read_write) */
286 #define BLKRRPART _IO(0x12,95) /* re-read partition table */
287 #define BLKGETSIZE _IO(0x12,96) /* return device size /512 (long *arg) */
288 #define BLKFLSBUF _IO(0x12,97) /* flush buffer cache */
289 #define BLKRASET _IO(0x12,98) /* set read ahead for block device */
290 #define BLKRAGET _IO(0x12,99) /* get current read ahead setting */
291 #define BLKFRASET _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
292 #define BLKFRAGET _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
293 #define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
294 #define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
295 #define BLKSSZGET _IO(0x12,104)/* get block device sector size */
296 #if 0
297 #define BLKPG _IO(0x12,105)/* See blkpg.h */
298
299 /* Some people are morons. Do not use sizeof! */
300
301 #define BLKELVGET _IOR(0x12,106,size_t)/* elevator get */
302 #define BLKELVSET _IOW(0x12,107,size_t)/* elevator set */
303 /* This was here just to show that the number is taken -
304 probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
305 #endif
306 /* A jump here: 108-111 have been used for various private purposes. */
307 #define BLKBSZGET _IOR(0x12,112,size_t)
308 #define BLKBSZSET _IOW(0x12,113,size_t)
309 #define BLKGETSIZE64 _IOR(0x12,114,size_t) /* return device size in bytes (u64 *arg) */
310 #define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup)
311 #define BLKTRACESTART _IO(0x12,116)
312 #define BLKTRACESTOP _IO(0x12,117)
313 #define BLKTRACETEARDOWN _IO(0x12,118)
314 #define BLKDISCARD _IO(0x12,119)
315 #define BLKIOMIN _IO(0x12,120)
316 #define BLKIOOPT _IO(0x12,121)
317 #define BLKALIGNOFF _IO(0x12,122)
318 #define BLKPBSZGET _IO(0x12,123)
319 #define BLKDISCARDZEROES _IO(0x12,124)
320 #define BLKSECDISCARD _IO(0x12,125)
321
322 #define BMAP_IOCTL 1 /* obsolete - kept for compatibility */
323 #define FIBMAP _IO(0x00,1) /* bmap access */
324 #define FIGETBSZ _IO(0x00,2) /* get the block size used for bmap */
325 #define FIFREEZE _IOWR('X', 119, int) /* Freeze */
326 #define FITHAW _IOWR('X', 120, int) /* Thaw */
327 #define FITRIM _IOWR('X', 121, struct fstrim_range) /* Trim */
328
329 #define FS_IOC_GETFLAGS _IOR('f', 1, long)
330 #define FS_IOC_SETFLAGS _IOW('f', 2, long)
331 #define FS_IOC_GETVERSION _IOR('v', 1, long)
332 #define FS_IOC_SETVERSION _IOW('v', 2, long)
333 #define FS_IOC_FIEMAP _IOWR('f', 11, struct fiemap)
334 #define FS_IOC32_GETFLAGS _IOR('f', 1, int)
335 #define FS_IOC32_SETFLAGS _IOW('f', 2, int)
336 #define FS_IOC32_GETVERSION _IOR('v', 1, int)
337 #define FS_IOC32_SETVERSION _IOW('v', 2, int)
338
339 /*
340 * Inode flags (FS_IOC_GETFLAGS / FS_IOC_SETFLAGS)
341 */
342 #define FS_SECRM_FL 0x00000001 /* Secure deletion */
343 #define FS_UNRM_FL 0x00000002 /* Undelete */
344 #define FS_COMPR_FL 0x00000004 /* Compress file */
345 #define FS_SYNC_FL 0x00000008 /* Synchronous updates */
346 #define FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
347 #define FS_APPEND_FL 0x00000020 /* writes to file may only append */
348 #define FS_NODUMP_FL 0x00000040 /* do not dump file */
349 #define FS_NOATIME_FL 0x00000080 /* do not update atime */
350 /* Reserved for compression usage... */
351 #define FS_DIRTY_FL 0x00000100
352 #define FS_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
353 #define FS_NOCOMP_FL 0x00000400 /* Don't compress */
354 #define FS_ECOMPR_FL 0x00000800 /* Compression error */
355 /* End compression flags --- maybe not all used */
356 #define FS_BTREE_FL 0x00001000 /* btree format dir */
357 #define FS_INDEX_FL 0x00001000 /* hash-indexed directory */
358 #define FS_IMAGIC_FL 0x00002000 /* AFS directory */
359 #define FS_JOURNAL_DATA_FL 0x00004000 /* Reserved for ext3 */
360 #define FS_NOTAIL_FL 0x00008000 /* file tail should not be merged */
361 #define FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
362 #define FS_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
363 #define FS_EXTENT_FL 0x00080000 /* Extents */
364 #define FS_DIRECTIO_FL 0x00100000 /* Use direct i/o */
365 #define FS_NOCOW_FL 0x00800000 /* Do not cow file */
366 #define FS_RESERVED_FL 0x80000000 /* reserved for ext2 lib */
367
368 #define FS_FL_USER_VISIBLE 0x0003DFFF /* User visible flags */
369 #define FS_FL_USER_MODIFIABLE 0x000380FF /* User modifiable flags */
370
371
372 #define SYNC_FILE_RANGE_WAIT_BEFORE 1
373 #define SYNC_FILE_RANGE_WRITE 2
374 #define SYNC_FILE_RANGE_WAIT_AFTER 4
375
376 #ifdef __KERNEL__
377
378 #include <linux/linkage.h>
379 #include <linux/wait.h>
380 #include <linux/kdev_t.h>
381 #include <linux/dcache.h>
382 #include <linux/path.h>
383 #include <linux/stat.h>
384 #include <linux/cache.h>
385 #include <linux/list.h>
386 #include <linux/radix-tree.h>
387 #include <linux/prio_tree.h>
388 #include <linux/init.h>
389 #include <linux/pid.h>
390 #include <linux/mutex.h>
391 #include <linux/capability.h>
392 #include <linux/semaphore.h>
393 #include <linux/fiemap.h>
394 #include <linux/rculist_bl.h>
395 #include <linux/shrinker.h>
396 #include <linux/atomic.h>
397
398 #include <asm/byteorder.h>
399
400 struct export_operations;
401 struct hd_geometry;
402 struct iovec;
403 struct nameidata;
404 struct kiocb;
405 struct kobject;
406 struct pipe_inode_info;
407 struct poll_table_struct;
408 struct kstatfs;
409 struct vm_area_struct;
410 struct vfsmount;
411 struct cred;
412
413 extern void __init inode_init(void);
414 extern void __init inode_init_early(void);
415 extern void __init files_init(unsigned long);
416
417 extern struct files_stat_struct files_stat;
418 extern unsigned long get_max_files(void);
419 extern int sysctl_nr_open;
420 extern struct inodes_stat_t inodes_stat;
421 extern int leases_enable, lease_break_time;
422
423 struct buffer_head;
424 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
425 struct buffer_head *bh_result, int create);
426 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
427 ssize_t bytes, void *private, int ret,
428 bool is_async);
429
430 /*
431 * Attribute flags. These should be or-ed together to figure out what
432 * has been changed!
433 */
434 #define ATTR_MODE (1 << 0)
435 #define ATTR_UID (1 << 1)
436 #define ATTR_GID (1 << 2)
437 #define ATTR_SIZE (1 << 3)
438 #define ATTR_ATIME (1 << 4)
439 #define ATTR_MTIME (1 << 5)
440 #define ATTR_CTIME (1 << 6)
441 #define ATTR_ATIME_SET (1 << 7)
442 #define ATTR_MTIME_SET (1 << 8)
443 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
444 #define ATTR_ATTR_FLAG (1 << 10)
445 #define ATTR_KILL_SUID (1 << 11)
446 #define ATTR_KILL_SGID (1 << 12)
447 #define ATTR_FILE (1 << 13)
448 #define ATTR_KILL_PRIV (1 << 14)
449 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
450 #define ATTR_TIMES_SET (1 << 16)
451
452 /*
453 * This is the Inode Attributes structure, used for notify_change(). It
454 * uses the above definitions as flags, to know which values have changed.
455 * Also, in this manner, a Filesystem can look at only the values it cares
456 * about. Basically, these are the attributes that the VFS layer can
457 * request to change from the FS layer.
458 *
459 * Derek Atkins <warlord@MIT.EDU> 94-10-20
460 */
461 struct iattr {
462 unsigned int ia_valid;
463 umode_t ia_mode;
464 uid_t ia_uid;
465 gid_t ia_gid;
466 loff_t ia_size;
467 struct timespec ia_atime;
468 struct timespec ia_mtime;
469 struct timespec ia_ctime;
470
471 /*
472 * Not an attribute, but an auxiliary info for filesystems wanting to
473 * implement an ftruncate() like method. NOTE: filesystem should
474 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
475 */
476 struct file *ia_file;
477 };
478
479 /*
480 * Includes for diskquotas.
481 */
482 #include <linux/quota.h>
483
484 /**
485 * enum positive_aop_returns - aop return codes with specific semantics
486 *
487 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
488 * completed, that the page is still locked, and
489 * should be considered active. The VM uses this hint
490 * to return the page to the active list -- it won't
491 * be a candidate for writeback again in the near
492 * future. Other callers must be careful to unlock
493 * the page if they get this return. Returned by
494 * writepage();
495 *
496 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
497 * unlocked it and the page might have been truncated.
498 * The caller should back up to acquiring a new page and
499 * trying again. The aop will be taking reasonable
500 * precautions not to livelock. If the caller held a page
501 * reference, it should drop it before retrying. Returned
502 * by readpage().
503 *
504 * address_space_operation functions return these large constants to indicate
505 * special semantics to the caller. These are much larger than the bytes in a
506 * page to allow for functions that return the number of bytes operated on in a
507 * given page.
508 */
509
510 enum positive_aop_returns {
511 AOP_WRITEPAGE_ACTIVATE = 0x80000,
512 AOP_TRUNCATED_PAGE = 0x80001,
513 };
514
515 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
516 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
517 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
518 * helper code (eg buffer layer)
519 * to clear GFP_FS from alloc */
520
521 /*
522 * oh the beauties of C type declarations.
523 */
524 struct page;
525 struct address_space;
526 struct writeback_control;
527
528 struct iov_iter {
529 const struct iovec *iov;
530 unsigned long nr_segs;
531 size_t iov_offset;
532 size_t count;
533 };
534
535 size_t iov_iter_copy_from_user_atomic(struct page *page,
536 struct iov_iter *i, unsigned long offset, size_t bytes);
537 size_t iov_iter_copy_from_user(struct page *page,
538 struct iov_iter *i, unsigned long offset, size_t bytes);
539 void iov_iter_advance(struct iov_iter *i, size_t bytes);
540 int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes);
541 size_t iov_iter_single_seg_count(struct iov_iter *i);
542
543 static inline void iov_iter_init(struct iov_iter *i,
544 const struct iovec *iov, unsigned long nr_segs,
545 size_t count, size_t written)
546 {
547 i->iov = iov;
548 i->nr_segs = nr_segs;
549 i->iov_offset = 0;
550 i->count = count + written;
551
552 iov_iter_advance(i, written);
553 }
554
555 static inline size_t iov_iter_count(struct iov_iter *i)
556 {
557 return i->count;
558 }
559
560 /*
561 * "descriptor" for what we're up to with a read.
562 * This allows us to use the same read code yet
563 * have multiple different users of the data that
564 * we read from a file.
565 *
566 * The simplest case just copies the data to user
567 * mode.
568 */
569 typedef struct {
570 size_t written;
571 size_t count;
572 union {
573 char __user *buf;
574 void *data;
575 } arg;
576 int error;
577 } read_descriptor_t;
578
579 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
580 unsigned long, unsigned long);
581
582 struct address_space_operations {
583 int (*writepage)(struct page *page, struct writeback_control *wbc);
584 int (*readpage)(struct file *, struct page *);
585
586 /* Write back some dirty pages from this mapping. */
587 int (*writepages)(struct address_space *, struct writeback_control *);
588
589 /* Set a page dirty. Return true if this dirtied it */
590 int (*set_page_dirty)(struct page *page);
591
592 int (*readpages)(struct file *filp, struct address_space *mapping,
593 struct list_head *pages, unsigned nr_pages);
594
595 int (*write_begin)(struct file *, struct address_space *mapping,
596 loff_t pos, unsigned len, unsigned flags,
597 struct page **pagep, void **fsdata);
598 int (*write_end)(struct file *, struct address_space *mapping,
599 loff_t pos, unsigned len, unsigned copied,
600 struct page *page, void *fsdata);
601
602 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
603 sector_t (*bmap)(struct address_space *, sector_t);
604 void (*invalidatepage) (struct page *, unsigned long);
605 int (*releasepage) (struct page *, gfp_t);
606 void (*freepage)(struct page *);
607 ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
608 loff_t offset, unsigned long nr_segs);
609 int (*get_xip_mem)(struct address_space *, pgoff_t, int,
610 void **, unsigned long *);
611 /* migrate the contents of a page to the specified target */
612 int (*migratepage) (struct address_space *,
613 struct page *, struct page *);
614 int (*launder_page) (struct page *);
615 int (*is_partially_uptodate) (struct page *, read_descriptor_t *,
616 unsigned long);
617 int (*error_remove_page)(struct address_space *, struct page *);
618 };
619
620 extern const struct address_space_operations empty_aops;
621
622 /*
623 * pagecache_write_begin/pagecache_write_end must be used by general code
624 * to write into the pagecache.
625 */
626 int pagecache_write_begin(struct file *, struct address_space *mapping,
627 loff_t pos, unsigned len, unsigned flags,
628 struct page **pagep, void **fsdata);
629
630 int pagecache_write_end(struct file *, struct address_space *mapping,
631 loff_t pos, unsigned len, unsigned copied,
632 struct page *page, void *fsdata);
633
634 struct backing_dev_info;
635 struct address_space {
636 struct inode *host; /* owner: inode, block_device */
637 struct radix_tree_root page_tree; /* radix tree of all pages */
638 spinlock_t tree_lock; /* and lock protecting it */
639 unsigned int i_mmap_writable;/* count VM_SHARED mappings */
640 struct prio_tree_root i_mmap; /* tree of private and shared mappings */
641 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
642 struct mutex i_mmap_mutex; /* protect tree, count, list */
643 /* Protected by tree_lock together with the radix tree */
644 unsigned long nrpages; /* number of total pages */
645 pgoff_t writeback_index;/* writeback starts here */
646 const struct address_space_operations *a_ops; /* methods */
647 unsigned long flags; /* error bits/gfp mask */
648 struct backing_dev_info *backing_dev_info; /* device readahead, etc */
649 spinlock_t private_lock; /* for use by the address_space */
650 struct list_head private_list; /* ditto */
651 struct address_space *assoc_mapping; /* ditto */
652 } __attribute__((aligned(sizeof(long))));
653 /*
654 * On most architectures that alignment is already the case; but
655 * must be enforced here for CRIS, to let the least significant bit
656 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
657 */
658
659 struct block_device {
660 dev_t bd_dev; /* not a kdev_t - it's a search key */
661 int bd_openers;
662 struct inode * bd_inode; /* will die */
663 struct super_block * bd_super;
664 struct mutex bd_mutex; /* open/close mutex */
665 struct list_head bd_inodes;
666 void * bd_claiming;
667 void * bd_holder;
668 int bd_holders;
669 bool bd_write_holder;
670 #ifdef CONFIG_SYSFS
671 struct list_head bd_holder_disks;
672 #endif
673 struct block_device * bd_contains;
674 unsigned bd_block_size;
675 struct hd_struct * bd_part;
676 /* number of times partitions within this device have been opened. */
677 unsigned bd_part_count;
678 int bd_invalidated;
679 struct gendisk * bd_disk;
680 struct list_head bd_list;
681 /*
682 * Private data. You must have bd_claim'ed the block_device
683 * to use this. NOTE: bd_claim allows an owner to claim
684 * the same device multiple times, the owner must take special
685 * care to not mess up bd_private for that case.
686 */
687 unsigned long bd_private;
688
689 /* The counter of freeze processes */
690 int bd_fsfreeze_count;
691 /* Mutex for freeze */
692 struct mutex bd_fsfreeze_mutex;
693 };
694
695 /*
696 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
697 * radix trees
698 */
699 #define PAGECACHE_TAG_DIRTY 0
700 #define PAGECACHE_TAG_WRITEBACK 1
701 #define PAGECACHE_TAG_TOWRITE 2
702
703 int mapping_tagged(struct address_space *mapping, int tag);
704
705 /*
706 * Might pages of this file be mapped into userspace?
707 */
708 static inline int mapping_mapped(struct address_space *mapping)
709 {
710 return !prio_tree_empty(&mapping->i_mmap) ||
711 !list_empty(&mapping->i_mmap_nonlinear);
712 }
713
714 /*
715 * Might pages of this file have been modified in userspace?
716 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
717 * marks vma as VM_SHARED if it is shared, and the file was opened for
718 * writing i.e. vma may be mprotected writable even if now readonly.
719 */
720 static inline int mapping_writably_mapped(struct address_space *mapping)
721 {
722 return mapping->i_mmap_writable != 0;
723 }
724
725 /*
726 * Use sequence counter to get consistent i_size on 32-bit processors.
727 */
728 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
729 #include <linux/seqlock.h>
730 #define __NEED_I_SIZE_ORDERED
731 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
732 #else
733 #define i_size_ordered_init(inode) do { } while (0)
734 #endif
735
736 struct posix_acl;
737 #define ACL_NOT_CACHED ((void *)(-1))
738
739 #define IOP_FASTPERM 0x0001
740 #define IOP_LOOKUP 0x0002
741 #define IOP_NOFOLLOW 0x0004
742
743 /*
744 * Keep mostly read-only and often accessed (especially for
745 * the RCU path lookup and 'stat' data) fields at the beginning
746 * of the 'struct inode'
747 */
748 struct inode {
749 umode_t i_mode;
750 unsigned short i_opflags;
751 uid_t i_uid;
752 gid_t i_gid;
753 unsigned int i_flags;
754
755 #ifdef CONFIG_FS_POSIX_ACL
756 struct posix_acl *i_acl;
757 struct posix_acl *i_default_acl;
758 #endif
759
760 const struct inode_operations *i_op;
761 struct super_block *i_sb;
762 struct address_space *i_mapping;
763
764 #ifdef CONFIG_SECURITY
765 void *i_security;
766 #endif
767
768 /* Stat data, not accessed from path walking */
769 unsigned long i_ino;
770 unsigned int i_nlink;
771 dev_t i_rdev;
772 loff_t i_size;
773 struct timespec i_atime;
774 struct timespec i_mtime;
775 struct timespec i_ctime;
776 unsigned int i_blkbits;
777 blkcnt_t i_blocks;
778
779 #ifdef __NEED_I_SIZE_ORDERED
780 seqcount_t i_size_seqcount;
781 #endif
782
783 /* Misc */
784 unsigned long i_state;
785 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
786 struct mutex i_mutex;
787
788 unsigned long dirtied_when; /* jiffies of first dirtying */
789
790 struct hlist_node i_hash;
791 struct list_head i_wb_list; /* backing dev IO list */
792 struct list_head i_lru; /* inode LRU list */
793 struct list_head i_sb_list;
794 union {
795 struct list_head i_dentry;
796 struct rcu_head i_rcu;
797 };
798 atomic_t i_count;
799 u64 i_version;
800 unsigned short i_bytes;
801 atomic_t i_dio_count;
802 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
803 struct file_lock *i_flock;
804 struct address_space i_data;
805 #ifdef CONFIG_QUOTA
806 struct dquot *i_dquot[MAXQUOTAS];
807 #endif
808 struct list_head i_devices;
809 union {
810 struct pipe_inode_info *i_pipe;
811 struct block_device *i_bdev;
812 struct cdev *i_cdev;
813 };
814
815 __u32 i_generation;
816
817 #ifdef CONFIG_FSNOTIFY
818 __u32 i_fsnotify_mask; /* all events this inode cares about */
819 struct hlist_head i_fsnotify_marks;
820 #endif
821
822 #ifdef CONFIG_IMA
823 atomic_t i_readcount; /* struct files open RO */
824 #endif
825 atomic_t i_writecount;
826 void *i_private; /* fs or device private pointer */
827 };
828
829 static inline int inode_unhashed(struct inode *inode)
830 {
831 return hlist_unhashed(&inode->i_hash);
832 }
833
834 /*
835 * inode->i_mutex nesting subclasses for the lock validator:
836 *
837 * 0: the object of the current VFS operation
838 * 1: parent
839 * 2: child/target
840 * 3: quota file
841 *
842 * The locking order between these classes is
843 * parent -> child -> normal -> xattr -> quota
844 */
845 enum inode_i_mutex_lock_class
846 {
847 I_MUTEX_NORMAL,
848 I_MUTEX_PARENT,
849 I_MUTEX_CHILD,
850 I_MUTEX_XATTR,
851 I_MUTEX_QUOTA
852 };
853
854 /*
855 * NOTE: in a 32bit arch with a preemptable kernel and
856 * an UP compile the i_size_read/write must be atomic
857 * with respect to the local cpu (unlike with preempt disabled),
858 * but they don't need to be atomic with respect to other cpus like in
859 * true SMP (so they need either to either locally disable irq around
860 * the read or for example on x86 they can be still implemented as a
861 * cmpxchg8b without the need of the lock prefix). For SMP compiles
862 * and 64bit archs it makes no difference if preempt is enabled or not.
863 */
864 static inline loff_t i_size_read(const struct inode *inode)
865 {
866 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
867 loff_t i_size;
868 unsigned int seq;
869
870 do {
871 seq = read_seqcount_begin(&inode->i_size_seqcount);
872 i_size = inode->i_size;
873 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
874 return i_size;
875 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
876 loff_t i_size;
877
878 preempt_disable();
879 i_size = inode->i_size;
880 preempt_enable();
881 return i_size;
882 #else
883 return inode->i_size;
884 #endif
885 }
886
887 /*
888 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
889 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
890 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
891 */
892 static inline void i_size_write(struct inode *inode, loff_t i_size)
893 {
894 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
895 write_seqcount_begin(&inode->i_size_seqcount);
896 inode->i_size = i_size;
897 write_seqcount_end(&inode->i_size_seqcount);
898 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
899 preempt_disable();
900 inode->i_size = i_size;
901 preempt_enable();
902 #else
903 inode->i_size = i_size;
904 #endif
905 }
906
907 static inline unsigned iminor(const struct inode *inode)
908 {
909 return MINOR(inode->i_rdev);
910 }
911
912 static inline unsigned imajor(const struct inode *inode)
913 {
914 return MAJOR(inode->i_rdev);
915 }
916
917 extern struct block_device *I_BDEV(struct inode *inode);
918
919 struct fown_struct {
920 rwlock_t lock; /* protects pid, uid, euid fields */
921 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
922 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
923 uid_t uid, euid; /* uid/euid of process setting the owner */
924 int signum; /* posix.1b rt signal to be delivered on IO */
925 };
926
927 /*
928 * Track a single file's readahead state
929 */
930 struct file_ra_state {
931 pgoff_t start; /* where readahead started */
932 unsigned int size; /* # of readahead pages */
933 unsigned int async_size; /* do asynchronous readahead when
934 there are only # of pages ahead */
935
936 unsigned int ra_pages; /* Maximum readahead window */
937 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
938 loff_t prev_pos; /* Cache last read() position */
939 };
940
941 /*
942 * Check if @index falls in the readahead windows.
943 */
944 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
945 {
946 return (index >= ra->start &&
947 index < ra->start + ra->size);
948 }
949
950 #define FILE_MNT_WRITE_TAKEN 1
951 #define FILE_MNT_WRITE_RELEASED 2
952
953 struct file {
954 /*
955 * fu_list becomes invalid after file_free is called and queued via
956 * fu_rcuhead for RCU freeing
957 */
958 union {
959 struct list_head fu_list;
960 struct rcu_head fu_rcuhead;
961 } f_u;
962 struct path f_path;
963 #define f_dentry f_path.dentry
964 #define f_vfsmnt f_path.mnt
965 const struct file_operations *f_op;
966 spinlock_t f_lock; /* f_ep_links, f_flags, no IRQ */
967 #ifdef CONFIG_SMP
968 int f_sb_list_cpu;
969 #endif
970 atomic_long_t f_count;
971 unsigned int f_flags;
972 fmode_t f_mode;
973 loff_t f_pos;
974 struct fown_struct f_owner;
975 const struct cred *f_cred;
976 struct file_ra_state f_ra;
977
978 u64 f_version;
979 #ifdef CONFIG_SECURITY
980 void *f_security;
981 #endif
982 /* needed for tty driver, and maybe others */
983 void *private_data;
984
985 #ifdef CONFIG_EPOLL
986 /* Used by fs/eventpoll.c to link all the hooks to this file */
987 struct list_head f_ep_links;
988 #endif /* #ifdef CONFIG_EPOLL */
989 struct address_space *f_mapping;
990 #ifdef CONFIG_DEBUG_WRITECOUNT
991 unsigned long f_mnt_write_state;
992 #endif
993 };
994
995 struct file_handle {
996 __u32 handle_bytes;
997 int handle_type;
998 /* file identifier */
999 unsigned char f_handle[0];
1000 };
1001
1002 #define get_file(x) atomic_long_inc(&(x)->f_count)
1003 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
1004 #define file_count(x) atomic_long_read(&(x)->f_count)
1005
1006 #ifdef CONFIG_DEBUG_WRITECOUNT
1007 static inline void file_take_write(struct file *f)
1008 {
1009 WARN_ON(f->f_mnt_write_state != 0);
1010 f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN;
1011 }
1012 static inline void file_release_write(struct file *f)
1013 {
1014 f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED;
1015 }
1016 static inline void file_reset_write(struct file *f)
1017 {
1018 f->f_mnt_write_state = 0;
1019 }
1020 static inline void file_check_state(struct file *f)
1021 {
1022 /*
1023 * At this point, either both or neither of these bits
1024 * should be set.
1025 */
1026 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN);
1027 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED);
1028 }
1029 static inline int file_check_writeable(struct file *f)
1030 {
1031 if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN)
1032 return 0;
1033 printk(KERN_WARNING "writeable file with no "
1034 "mnt_want_write()\n");
1035 WARN_ON(1);
1036 return -EINVAL;
1037 }
1038 #else /* !CONFIG_DEBUG_WRITECOUNT */
1039 static inline void file_take_write(struct file *filp) {}
1040 static inline void file_release_write(struct file *filp) {}
1041 static inline void file_reset_write(struct file *filp) {}
1042 static inline void file_check_state(struct file *filp) {}
1043 static inline int file_check_writeable(struct file *filp)
1044 {
1045 return 0;
1046 }
1047 #endif /* CONFIG_DEBUG_WRITECOUNT */
1048
1049 #define MAX_NON_LFS ((1UL<<31) - 1)
1050
1051 /* Page cache limit. The filesystems should put that into their s_maxbytes
1052 limits, otherwise bad things can happen in VM. */
1053 #if BITS_PER_LONG==32
1054 #define MAX_LFS_FILESIZE (((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
1055 #elif BITS_PER_LONG==64
1056 #define MAX_LFS_FILESIZE 0x7fffffffffffffffUL
1057 #endif
1058
1059 #define FL_POSIX 1
1060 #define FL_FLOCK 2
1061 #define FL_ACCESS 8 /* not trying to lock, just looking */
1062 #define FL_EXISTS 16 /* when unlocking, test for existence */
1063 #define FL_LEASE 32 /* lease held on this file */
1064 #define FL_CLOSE 64 /* unlock on close */
1065 #define FL_SLEEP 128 /* A blocking lock */
1066
1067 /*
1068 * Special return value from posix_lock_file() and vfs_lock_file() for
1069 * asynchronous locking.
1070 */
1071 #define FILE_LOCK_DEFERRED 1
1072
1073 /*
1074 * The POSIX file lock owner is determined by
1075 * the "struct files_struct" in the thread group
1076 * (or NULL for no owner - BSD locks).
1077 *
1078 * Lockd stuffs a "host" pointer into this.
1079 */
1080 typedef struct files_struct *fl_owner_t;
1081
1082 struct file_lock_operations {
1083 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1084 void (*fl_release_private)(struct file_lock *);
1085 };
1086
1087 struct lock_manager_operations {
1088 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
1089 void (*lm_notify)(struct file_lock *); /* unblock callback */
1090 int (*lm_grant)(struct file_lock *, struct file_lock *, int);
1091 void (*lm_release_private)(struct file_lock *);
1092 void (*lm_break)(struct file_lock *);
1093 int (*lm_change)(struct file_lock **, int);
1094 };
1095
1096 struct lock_manager {
1097 struct list_head list;
1098 };
1099
1100 void locks_start_grace(struct lock_manager *);
1101 void locks_end_grace(struct lock_manager *);
1102 int locks_in_grace(void);
1103
1104 /* that will die - we need it for nfs_lock_info */
1105 #include <linux/nfs_fs_i.h>
1106
1107 struct file_lock {
1108 struct file_lock *fl_next; /* singly linked list for this inode */
1109 struct list_head fl_link; /* doubly linked list of all locks */
1110 struct list_head fl_block; /* circular list of blocked processes */
1111 fl_owner_t fl_owner;
1112 unsigned char fl_flags;
1113 unsigned char fl_type;
1114 unsigned int fl_pid;
1115 struct pid *fl_nspid;
1116 wait_queue_head_t fl_wait;
1117 struct file *fl_file;
1118 loff_t fl_start;
1119 loff_t fl_end;
1120
1121 struct fasync_struct * fl_fasync; /* for lease break notifications */
1122 unsigned long fl_break_time; /* for nonblocking lease breaks */
1123
1124 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1125 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1126 union {
1127 struct nfs_lock_info nfs_fl;
1128 struct nfs4_lock_info nfs4_fl;
1129 struct {
1130 struct list_head link; /* link in AFS vnode's pending_locks list */
1131 int state; /* state of grant or error if -ve */
1132 } afs;
1133 } fl_u;
1134 };
1135
1136 /* The following constant reflects the upper bound of the file/locking space */
1137 #ifndef OFFSET_MAX
1138 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1139 #define OFFSET_MAX INT_LIMIT(loff_t)
1140 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1141 #endif
1142
1143 #include <linux/fcntl.h>
1144
1145 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1146
1147 #ifdef CONFIG_FILE_LOCKING
1148 extern int fcntl_getlk(struct file *, struct flock __user *);
1149 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1150 struct flock __user *);
1151
1152 #if BITS_PER_LONG == 32
1153 extern int fcntl_getlk64(struct file *, struct flock64 __user *);
1154 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1155 struct flock64 __user *);
1156 #endif
1157
1158 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1159 extern int fcntl_getlease(struct file *filp);
1160
1161 /* fs/locks.c */
1162 void locks_free_lock(struct file_lock *fl);
1163 extern void locks_init_lock(struct file_lock *);
1164 extern struct file_lock * locks_alloc_lock(void);
1165 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1166 extern void __locks_copy_lock(struct file_lock *, const struct file_lock *);
1167 extern void locks_remove_posix(struct file *, fl_owner_t);
1168 extern void locks_remove_flock(struct file *);
1169 extern void locks_release_private(struct file_lock *);
1170 extern void posix_test_lock(struct file *, struct file_lock *);
1171 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1172 extern int posix_lock_file_wait(struct file *, struct file_lock *);
1173 extern int posix_unblock_lock(struct file *, struct file_lock *);
1174 extern int vfs_test_lock(struct file *, struct file_lock *);
1175 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1176 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1177 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
1178 extern int __break_lease(struct inode *inode, unsigned int flags);
1179 extern void lease_get_mtime(struct inode *, struct timespec *time);
1180 extern int generic_setlease(struct file *, long, struct file_lock **);
1181 extern int vfs_setlease(struct file *, long, struct file_lock **);
1182 extern int lease_modify(struct file_lock **, int);
1183 extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
1184 extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
1185 extern void lock_flocks(void);
1186 extern void unlock_flocks(void);
1187 #else /* !CONFIG_FILE_LOCKING */
1188 static inline int fcntl_getlk(struct file *file, struct flock __user *user)
1189 {
1190 return -EINVAL;
1191 }
1192
1193 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1194 unsigned int cmd, struct flock __user *user)
1195 {
1196 return -EACCES;
1197 }
1198
1199 #if BITS_PER_LONG == 32
1200 static inline int fcntl_getlk64(struct file *file, struct flock64 __user *user)
1201 {
1202 return -EINVAL;
1203 }
1204
1205 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1206 unsigned int cmd, struct flock64 __user *user)
1207 {
1208 return -EACCES;
1209 }
1210 #endif
1211 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1212 {
1213 return 0;
1214 }
1215
1216 static inline int fcntl_getlease(struct file *filp)
1217 {
1218 return 0;
1219 }
1220
1221 static inline void locks_init_lock(struct file_lock *fl)
1222 {
1223 return;
1224 }
1225
1226 static inline void __locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1227 {
1228 return;
1229 }
1230
1231 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1232 {
1233 return;
1234 }
1235
1236 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1237 {
1238 return;
1239 }
1240
1241 static inline void locks_remove_flock(struct file *filp)
1242 {
1243 return;
1244 }
1245
1246 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1247 {
1248 return;
1249 }
1250
1251 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1252 struct file_lock *conflock)
1253 {
1254 return -ENOLCK;
1255 }
1256
1257 static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1258 {
1259 return -ENOLCK;
1260 }
1261
1262 static inline int posix_unblock_lock(struct file *filp,
1263 struct file_lock *waiter)
1264 {
1265 return -ENOENT;
1266 }
1267
1268 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1269 {
1270 return 0;
1271 }
1272
1273 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1274 struct file_lock *fl, struct file_lock *conf)
1275 {
1276 return -ENOLCK;
1277 }
1278
1279 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1280 {
1281 return 0;
1282 }
1283
1284 static inline int flock_lock_file_wait(struct file *filp,
1285 struct file_lock *request)
1286 {
1287 return -ENOLCK;
1288 }
1289
1290 static inline int __break_lease(struct inode *inode, unsigned int mode)
1291 {
1292 return 0;
1293 }
1294
1295 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1296 {
1297 return;
1298 }
1299
1300 static inline int generic_setlease(struct file *filp, long arg,
1301 struct file_lock **flp)
1302 {
1303 return -EINVAL;
1304 }
1305
1306 static inline int vfs_setlease(struct file *filp, long arg,
1307 struct file_lock **lease)
1308 {
1309 return -EINVAL;
1310 }
1311
1312 static inline int lease_modify(struct file_lock **before, int arg)
1313 {
1314 return -EINVAL;
1315 }
1316
1317 static inline int lock_may_read(struct inode *inode, loff_t start,
1318 unsigned long len)
1319 {
1320 return 1;
1321 }
1322
1323 static inline int lock_may_write(struct inode *inode, loff_t start,
1324 unsigned long len)
1325 {
1326 return 1;
1327 }
1328
1329 static inline void lock_flocks(void)
1330 {
1331 }
1332
1333 static inline void unlock_flocks(void)
1334 {
1335 }
1336
1337 #endif /* !CONFIG_FILE_LOCKING */
1338
1339
1340 struct fasync_struct {
1341 spinlock_t fa_lock;
1342 int magic;
1343 int fa_fd;
1344 struct fasync_struct *fa_next; /* singly linked list */
1345 struct file *fa_file;
1346 struct rcu_head fa_rcu;
1347 };
1348
1349 #define FASYNC_MAGIC 0x4601
1350
1351 /* SMP safe fasync helpers: */
1352 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1353 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1354 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1355 extern struct fasync_struct *fasync_alloc(void);
1356 extern void fasync_free(struct fasync_struct *);
1357
1358 /* can be called from interrupts */
1359 extern void kill_fasync(struct fasync_struct **, int, int);
1360
1361 extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1362 extern int f_setown(struct file *filp, unsigned long arg, int force);
1363 extern void f_delown(struct file *filp);
1364 extern pid_t f_getown(struct file *filp);
1365 extern int send_sigurg(struct fown_struct *fown);
1366
1367 /*
1368 * Umount options
1369 */
1370
1371 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1372 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1373 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1374 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1375 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1376
1377 extern struct list_head super_blocks;
1378 extern spinlock_t sb_lock;
1379
1380 struct super_block {
1381 struct list_head s_list; /* Keep this first */
1382 dev_t s_dev; /* search index; _not_ kdev_t */
1383 unsigned char s_dirt;
1384 unsigned char s_blocksize_bits;
1385 unsigned long s_blocksize;
1386 loff_t s_maxbytes; /* Max file size */
1387 struct file_system_type *s_type;
1388 const struct super_operations *s_op;
1389 const struct dquot_operations *dq_op;
1390 const struct quotactl_ops *s_qcop;
1391 const struct export_operations *s_export_op;
1392 unsigned long s_flags;
1393 unsigned long s_magic;
1394 struct dentry *s_root;
1395 struct rw_semaphore s_umount;
1396 struct mutex s_lock;
1397 int s_count;
1398 atomic_t s_active;
1399 #ifdef CONFIG_SECURITY
1400 void *s_security;
1401 #endif
1402 const struct xattr_handler **s_xattr;
1403
1404 struct list_head s_inodes; /* all inodes */
1405 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1406 #ifdef CONFIG_SMP
1407 struct list_head __percpu *s_files;
1408 #else
1409 struct list_head s_files;
1410 #endif
1411 /* s_dentry_lru, s_nr_dentry_unused protected by dcache.c lru locks */
1412 struct list_head s_dentry_lru; /* unused dentry lru */
1413 int s_nr_dentry_unused; /* # of dentry on lru */
1414
1415 /* s_inode_lru_lock protects s_inode_lru and s_nr_inodes_unused */
1416 spinlock_t s_inode_lru_lock ____cacheline_aligned_in_smp;
1417 struct list_head s_inode_lru; /* unused inode lru */
1418 int s_nr_inodes_unused; /* # of inodes on lru */
1419
1420 struct block_device *s_bdev;
1421 struct backing_dev_info *s_bdi;
1422 struct mtd_info *s_mtd;
1423 struct list_head s_instances;
1424 struct quota_info s_dquot; /* Diskquota specific options */
1425
1426 int s_frozen;
1427 wait_queue_head_t s_wait_unfrozen;
1428
1429 char s_id[32]; /* Informational name */
1430 u8 s_uuid[16]; /* UUID */
1431
1432 void *s_fs_info; /* Filesystem private info */
1433 fmode_t s_mode;
1434
1435 /* Granularity of c/m/atime in ns.
1436 Cannot be worse than a second */
1437 u32 s_time_gran;
1438
1439 /*
1440 * The next field is for VFS *only*. No filesystems have any business
1441 * even looking at it. You had been warned.
1442 */
1443 struct mutex s_vfs_rename_mutex; /* Kludge */
1444
1445 /*
1446 * Filesystem subtype. If non-empty the filesystem type field
1447 * in /proc/mounts will be "type.subtype"
1448 */
1449 char *s_subtype;
1450
1451 /*
1452 * Saved mount options for lazy filesystems using
1453 * generic_show_options()
1454 */
1455 char __rcu *s_options;
1456 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1457
1458 /*
1459 * Saved pool identifier for cleancache (-1 means none)
1460 */
1461 int cleancache_poolid;
1462
1463 struct shrinker s_shrink; /* per-sb shrinker handle */
1464 };
1465
1466 /* superblock cache pruning functions */
1467 extern void prune_icache_sb(struct super_block *sb, int nr_to_scan);
1468 extern void prune_dcache_sb(struct super_block *sb, int nr_to_scan);
1469
1470 extern struct timespec current_fs_time(struct super_block *sb);
1471
1472 /*
1473 * Snapshotting support.
1474 */
1475 enum {
1476 SB_UNFROZEN = 0,
1477 SB_FREEZE_WRITE = 1,
1478 SB_FREEZE_TRANS = 2,
1479 };
1480
1481 #define vfs_check_frozen(sb, level) \
1482 wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
1483
1484 /*
1485 * until VFS tracks user namespaces for inodes, just make all files
1486 * belong to init_user_ns
1487 */
1488 extern struct user_namespace init_user_ns;
1489 #define inode_userns(inode) (&init_user_ns)
1490 extern bool inode_owner_or_capable(const struct inode *inode);
1491
1492 /* not quite ready to be deprecated, but... */
1493 extern void lock_super(struct super_block *);
1494 extern void unlock_super(struct super_block *);
1495
1496 /*
1497 * VFS helper functions..
1498 */
1499 extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *);
1500 extern int vfs_mkdir(struct inode *, struct dentry *, int);
1501 extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t);
1502 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1503 extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
1504 extern int vfs_rmdir(struct inode *, struct dentry *);
1505 extern int vfs_unlink(struct inode *, struct dentry *);
1506 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1507
1508 /*
1509 * VFS dentry helper functions.
1510 */
1511 extern void dentry_unhash(struct dentry *dentry);
1512
1513 /*
1514 * VFS file helper functions.
1515 */
1516 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1517 mode_t mode);
1518 /*
1519 * VFS FS_IOC_FIEMAP helper definitions.
1520 */
1521 struct fiemap_extent_info {
1522 unsigned int fi_flags; /* Flags as passed from user */
1523 unsigned int fi_extents_mapped; /* Number of mapped extents */
1524 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1525 struct fiemap_extent __user *fi_extents_start; /* Start of
1526 fiemap_extent array */
1527 };
1528 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1529 u64 phys, u64 len, u32 flags);
1530 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1531
1532 /*
1533 * File types
1534 *
1535 * NOTE! These match bits 12..15 of stat.st_mode
1536 * (ie "(i_mode >> 12) & 15").
1537 */
1538 #define DT_UNKNOWN 0
1539 #define DT_FIFO 1
1540 #define DT_CHR 2
1541 #define DT_DIR 4
1542 #define DT_BLK 6
1543 #define DT_REG 8
1544 #define DT_LNK 10
1545 #define DT_SOCK 12
1546 #define DT_WHT 14
1547
1548 /*
1549 * This is the "filldir" function type, used by readdir() to let
1550 * the kernel specify what kind of dirent layout it wants to have.
1551 * This allows the kernel to read directories into kernel space or
1552 * to have different dirent layouts depending on the binary type.
1553 */
1554 typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
1555 struct block_device_operations;
1556
1557 /* These macros are for out of kernel modules to test that
1558 * the kernel supports the unlocked_ioctl and compat_ioctl
1559 * fields in struct file_operations. */
1560 #define HAVE_COMPAT_IOCTL 1
1561 #define HAVE_UNLOCKED_IOCTL 1
1562
1563 struct file_operations {
1564 struct module *owner;
1565 loff_t (*llseek) (struct file *, loff_t, int);
1566 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1567 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1568 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1569 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1570 int (*readdir) (struct file *, void *, filldir_t);
1571 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1572 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1573 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1574 int (*mmap) (struct file *, struct vm_area_struct *);
1575 int (*open) (struct inode *, struct file *);
1576 int (*flush) (struct file *, fl_owner_t id);
1577 int (*release) (struct inode *, struct file *);
1578 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1579 int (*aio_fsync) (struct kiocb *, int datasync);
1580 int (*fasync) (int, struct file *, int);
1581 int (*lock) (struct file *, int, struct file_lock *);
1582 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1583 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1584 int (*check_flags)(int);
1585 int (*flock) (struct file *, int, struct file_lock *);
1586 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1587 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1588 int (*setlease)(struct file *, long, struct file_lock **);
1589 long (*fallocate)(struct file *file, int mode, loff_t offset,
1590 loff_t len);
1591 };
1592
1593 struct inode_operations {
1594 struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
1595 void * (*follow_link) (struct dentry *, struct nameidata *);
1596 int (*permission) (struct inode *, int);
1597 struct posix_acl * (*get_acl)(struct inode *, int);
1598
1599 int (*readlink) (struct dentry *, char __user *,int);
1600 void (*put_link) (struct dentry *, struct nameidata *, void *);
1601
1602 int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
1603 int (*link) (struct dentry *,struct inode *,struct dentry *);
1604 int (*unlink) (struct inode *,struct dentry *);
1605 int (*symlink) (struct inode *,struct dentry *,const char *);
1606 int (*mkdir) (struct inode *,struct dentry *,int);
1607 int (*rmdir) (struct inode *,struct dentry *);
1608 int (*mknod) (struct inode *,struct dentry *,int,dev_t);
1609 int (*rename) (struct inode *, struct dentry *,
1610 struct inode *, struct dentry *);
1611 void (*truncate) (struct inode *);
1612 int (*setattr) (struct dentry *, struct iattr *);
1613 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1614 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1615 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1616 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1617 int (*removexattr) (struct dentry *, const char *);
1618 void (*truncate_range)(struct inode *, loff_t, loff_t);
1619 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1620 u64 len);
1621 } ____cacheline_aligned;
1622
1623 struct seq_file;
1624
1625 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1626 unsigned long nr_segs, unsigned long fast_segs,
1627 struct iovec *fast_pointer,
1628 struct iovec **ret_pointer);
1629
1630 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1631 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1632 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1633 unsigned long, loff_t *);
1634 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1635 unsigned long, loff_t *);
1636
1637 struct super_operations {
1638 struct inode *(*alloc_inode)(struct super_block *sb);
1639 void (*destroy_inode)(struct inode *);
1640
1641 void (*dirty_inode) (struct inode *, int flags);
1642 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1643 int (*drop_inode) (struct inode *);
1644 void (*evict_inode) (struct inode *);
1645 void (*put_super) (struct super_block *);
1646 void (*write_super) (struct super_block *);
1647 int (*sync_fs)(struct super_block *sb, int wait);
1648 int (*freeze_fs) (struct super_block *);
1649 int (*unfreeze_fs) (struct super_block *);
1650 int (*statfs) (struct dentry *, struct kstatfs *);
1651 int (*remount_fs) (struct super_block *, int *, char *);
1652 void (*umount_begin) (struct super_block *);
1653
1654 int (*show_options)(struct seq_file *, struct vfsmount *);
1655 int (*show_devname)(struct seq_file *, struct vfsmount *);
1656 int (*show_path)(struct seq_file *, struct vfsmount *);
1657 int (*show_stats)(struct seq_file *, struct vfsmount *);
1658 #ifdef CONFIG_QUOTA
1659 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1660 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1661 #endif
1662 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1663 int (*nr_cached_objects)(struct super_block *);
1664 void (*free_cached_objects)(struct super_block *, int);
1665 };
1666
1667 /*
1668 * Inode state bits. Protected by inode->i_lock
1669 *
1670 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1671 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1672 *
1673 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1674 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1675 * various stages of removing an inode.
1676 *
1677 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1678 *
1679 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1680 * fdatasync(). i_atime is the usual cause.
1681 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1682 * these changes separately from I_DIRTY_SYNC so that we
1683 * don't have to write inode on fdatasync() when only
1684 * mtime has changed in it.
1685 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1686 * I_NEW Serves as both a mutex and completion notification.
1687 * New inodes set I_NEW. If two processes both create
1688 * the same inode, one of them will release its inode and
1689 * wait for I_NEW to be released before returning.
1690 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1691 * also cause waiting on I_NEW, without I_NEW actually
1692 * being set. find_inode() uses this to prevent returning
1693 * nearly-dead inodes.
1694 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1695 * is zero. I_FREEING must be set when I_WILL_FREE is
1696 * cleared.
1697 * I_FREEING Set when inode is about to be freed but still has dirty
1698 * pages or buffers attached or the inode itself is still
1699 * dirty.
1700 * I_CLEAR Added by end_writeback(). In this state the inode is clean
1701 * and can be destroyed. Inode keeps I_FREEING.
1702 *
1703 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1704 * prohibited for many purposes. iget() must wait for
1705 * the inode to be completely released, then create it
1706 * anew. Other functions will just ignore such inodes,
1707 * if appropriate. I_NEW is used for waiting.
1708 *
1709 * I_SYNC Synchonized write of dirty inode data. The bits is
1710 * set during data writeback, and cleared with a wakeup
1711 * on the bit address once it is done.
1712 *
1713 * I_REFERENCED Marks the inode as recently references on the LRU list.
1714 *
1715 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1716 *
1717 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1718 */
1719 #define I_DIRTY_SYNC (1 << 0)
1720 #define I_DIRTY_DATASYNC (1 << 1)
1721 #define I_DIRTY_PAGES (1 << 2)
1722 #define __I_NEW 3
1723 #define I_NEW (1 << __I_NEW)
1724 #define I_WILL_FREE (1 << 4)
1725 #define I_FREEING (1 << 5)
1726 #define I_CLEAR (1 << 6)
1727 #define __I_SYNC 7
1728 #define I_SYNC (1 << __I_SYNC)
1729 #define I_REFERENCED (1 << 8)
1730 #define __I_DIO_WAKEUP 9
1731 #define I_DIO_WAKEUP (1 << I_DIO_WAKEUP)
1732
1733 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1734
1735 extern void __mark_inode_dirty(struct inode *, int);
1736 static inline void mark_inode_dirty(struct inode *inode)
1737 {
1738 __mark_inode_dirty(inode, I_DIRTY);
1739 }
1740
1741 static inline void mark_inode_dirty_sync(struct inode *inode)
1742 {
1743 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1744 }
1745
1746 /**
1747 * inc_nlink - directly increment an inode's link count
1748 * @inode: inode
1749 *
1750 * This is a low-level filesystem helper to replace any
1751 * direct filesystem manipulation of i_nlink. Currently,
1752 * it is only here for parity with dec_nlink().
1753 */
1754 static inline void inc_nlink(struct inode *inode)
1755 {
1756 inode->i_nlink++;
1757 }
1758
1759 static inline void inode_inc_link_count(struct inode *inode)
1760 {
1761 inc_nlink(inode);
1762 mark_inode_dirty(inode);
1763 }
1764
1765 /**
1766 * drop_nlink - directly drop an inode's link count
1767 * @inode: inode
1768 *
1769 * This is a low-level filesystem helper to replace any
1770 * direct filesystem manipulation of i_nlink. In cases
1771 * where we are attempting to track writes to the
1772 * filesystem, a decrement to zero means an imminent
1773 * write when the file is truncated and actually unlinked
1774 * on the filesystem.
1775 */
1776 static inline void drop_nlink(struct inode *inode)
1777 {
1778 inode->i_nlink--;
1779 }
1780
1781 /**
1782 * clear_nlink - directly zero an inode's link count
1783 * @inode: inode
1784 *
1785 * This is a low-level filesystem helper to replace any
1786 * direct filesystem manipulation of i_nlink. See
1787 * drop_nlink() for why we care about i_nlink hitting zero.
1788 */
1789 static inline void clear_nlink(struct inode *inode)
1790 {
1791 inode->i_nlink = 0;
1792 }
1793
1794 static inline void inode_dec_link_count(struct inode *inode)
1795 {
1796 drop_nlink(inode);
1797 mark_inode_dirty(inode);
1798 }
1799
1800 /**
1801 * inode_inc_iversion - increments i_version
1802 * @inode: inode that need to be updated
1803 *
1804 * Every time the inode is modified, the i_version field will be incremented.
1805 * The filesystem has to be mounted with i_version flag
1806 */
1807
1808 static inline void inode_inc_iversion(struct inode *inode)
1809 {
1810 spin_lock(&inode->i_lock);
1811 inode->i_version++;
1812 spin_unlock(&inode->i_lock);
1813 }
1814
1815 extern void touch_atime(struct vfsmount *mnt, struct dentry *dentry);
1816 static inline void file_accessed(struct file *file)
1817 {
1818 if (!(file->f_flags & O_NOATIME))
1819 touch_atime(file->f_path.mnt, file->f_path.dentry);
1820 }
1821
1822 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1823 int sync_inode_metadata(struct inode *inode, int wait);
1824
1825 struct file_system_type {
1826 const char *name;
1827 int fs_flags;
1828 struct dentry *(*mount) (struct file_system_type *, int,
1829 const char *, void *);
1830 void (*kill_sb) (struct super_block *);
1831 struct module *owner;
1832 struct file_system_type * next;
1833 struct list_head fs_supers;
1834
1835 struct lock_class_key s_lock_key;
1836 struct lock_class_key s_umount_key;
1837 struct lock_class_key s_vfs_rename_key;
1838
1839 struct lock_class_key i_lock_key;
1840 struct lock_class_key i_mutex_key;
1841 struct lock_class_key i_mutex_dir_key;
1842 };
1843
1844 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1845 void *data, int (*fill_super)(struct super_block *, void *, int));
1846 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1847 int flags, const char *dev_name, void *data,
1848 int (*fill_super)(struct super_block *, void *, int));
1849 extern struct dentry *mount_single(struct file_system_type *fs_type,
1850 int flags, void *data,
1851 int (*fill_super)(struct super_block *, void *, int));
1852 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1853 int flags, void *data,
1854 int (*fill_super)(struct super_block *, void *, int));
1855 void generic_shutdown_super(struct super_block *sb);
1856 void kill_block_super(struct super_block *sb);
1857 void kill_anon_super(struct super_block *sb);
1858 void kill_litter_super(struct super_block *sb);
1859 void deactivate_super(struct super_block *sb);
1860 void deactivate_locked_super(struct super_block *sb);
1861 int set_anon_super(struct super_block *s, void *data);
1862 int get_anon_bdev(dev_t *);
1863 void free_anon_bdev(dev_t);
1864 struct super_block *sget(struct file_system_type *type,
1865 int (*test)(struct super_block *,void *),
1866 int (*set)(struct super_block *,void *),
1867 void *data);
1868 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
1869 const struct super_operations *ops,
1870 const struct dentry_operations *dops,
1871 unsigned long);
1872
1873 static inline void sb_mark_dirty(struct super_block *sb)
1874 {
1875 sb->s_dirt = 1;
1876 }
1877 static inline void sb_mark_clean(struct super_block *sb)
1878 {
1879 sb->s_dirt = 0;
1880 }
1881 static inline int sb_is_dirty(struct super_block *sb)
1882 {
1883 return sb->s_dirt;
1884 }
1885
1886 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1887 #define fops_get(fops) \
1888 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1889 #define fops_put(fops) \
1890 do { if (fops) module_put((fops)->owner); } while(0)
1891
1892 extern int register_filesystem(struct file_system_type *);
1893 extern int unregister_filesystem(struct file_system_type *);
1894 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1895 #define kern_mount(type) kern_mount_data(type, NULL)
1896 extern void kern_unmount(struct vfsmount *mnt);
1897 extern int may_umount_tree(struct vfsmount *);
1898 extern int may_umount(struct vfsmount *);
1899 extern long do_mount(char *, char *, char *, unsigned long, void *);
1900 extern struct vfsmount *collect_mounts(struct path *);
1901 extern void drop_collected_mounts(struct vfsmount *);
1902 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1903 struct vfsmount *);
1904 extern int vfs_statfs(struct path *, struct kstatfs *);
1905 extern int user_statfs(const char __user *, struct kstatfs *);
1906 extern int fd_statfs(int, struct kstatfs *);
1907 extern int statfs_by_dentry(struct dentry *, struct kstatfs *);
1908 extern int freeze_super(struct super_block *super);
1909 extern int thaw_super(struct super_block *super);
1910
1911 extern int current_umask(void);
1912
1913 /* /sys/fs */
1914 extern struct kobject *fs_kobj;
1915
1916 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
1917 extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1918
1919 #define FLOCK_VERIFY_READ 1
1920 #define FLOCK_VERIFY_WRITE 2
1921
1922 #ifdef CONFIG_FILE_LOCKING
1923 extern int locks_mandatory_locked(struct inode *);
1924 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1925
1926 /*
1927 * Candidates for mandatory locking have the setgid bit set
1928 * but no group execute bit - an otherwise meaningless combination.
1929 */
1930
1931 static inline int __mandatory_lock(struct inode *ino)
1932 {
1933 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1934 }
1935
1936 /*
1937 * ... and these candidates should be on MS_MANDLOCK mounted fs,
1938 * otherwise these will be advisory locks
1939 */
1940
1941 static inline int mandatory_lock(struct inode *ino)
1942 {
1943 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1944 }
1945
1946 static inline int locks_verify_locked(struct inode *inode)
1947 {
1948 if (mandatory_lock(inode))
1949 return locks_mandatory_locked(inode);
1950 return 0;
1951 }
1952
1953 static inline int locks_verify_truncate(struct inode *inode,
1954 struct file *filp,
1955 loff_t size)
1956 {
1957 if (inode->i_flock && mandatory_lock(inode))
1958 return locks_mandatory_area(
1959 FLOCK_VERIFY_WRITE, inode, filp,
1960 size < inode->i_size ? size : inode->i_size,
1961 (size < inode->i_size ? inode->i_size - size
1962 : size - inode->i_size)
1963 );
1964 return 0;
1965 }
1966
1967 static inline int break_lease(struct inode *inode, unsigned int mode)
1968 {
1969 if (inode->i_flock)
1970 return __break_lease(inode, mode);
1971 return 0;
1972 }
1973 #else /* !CONFIG_FILE_LOCKING */
1974 static inline int locks_mandatory_locked(struct inode *inode)
1975 {
1976 return 0;
1977 }
1978
1979 static inline int locks_mandatory_area(int rw, struct inode *inode,
1980 struct file *filp, loff_t offset,
1981 size_t count)
1982 {
1983 return 0;
1984 }
1985
1986 static inline int __mandatory_lock(struct inode *inode)
1987 {
1988 return 0;
1989 }
1990
1991 static inline int mandatory_lock(struct inode *inode)
1992 {
1993 return 0;
1994 }
1995
1996 static inline int locks_verify_locked(struct inode *inode)
1997 {
1998 return 0;
1999 }
2000
2001 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2002 size_t size)
2003 {
2004 return 0;
2005 }
2006
2007 static inline int break_lease(struct inode *inode, unsigned int mode)
2008 {
2009 return 0;
2010 }
2011
2012 #endif /* CONFIG_FILE_LOCKING */
2013
2014 /* fs/open.c */
2015
2016 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2017 struct file *filp);
2018 extern int do_fallocate(struct file *file, int mode, loff_t offset,
2019 loff_t len);
2020 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2021 int mode);
2022 extern struct file *filp_open(const char *, int, int);
2023 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2024 const char *, int);
2025 extern struct file * dentry_open(struct dentry *, struct vfsmount *, int,
2026 const struct cred *);
2027 extern int filp_close(struct file *, fl_owner_t id);
2028 extern char * getname(const char __user *);
2029
2030 /* fs/ioctl.c */
2031
2032 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2033
2034 /* fs/dcache.c */
2035 extern void __init vfs_caches_init_early(void);
2036 extern void __init vfs_caches_init(unsigned long);
2037
2038 extern struct kmem_cache *names_cachep;
2039
2040 #define __getname_gfp(gfp) kmem_cache_alloc(names_cachep, (gfp))
2041 #define __getname() __getname_gfp(GFP_KERNEL)
2042 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2043 #ifndef CONFIG_AUDITSYSCALL
2044 #define putname(name) __putname(name)
2045 #else
2046 extern void putname(const char *name);
2047 #endif
2048
2049 #ifdef CONFIG_BLOCK
2050 extern int register_blkdev(unsigned int, const char *);
2051 extern void unregister_blkdev(unsigned int, const char *);
2052 extern struct block_device *bdget(dev_t);
2053 extern struct block_device *bdgrab(struct block_device *bdev);
2054 extern void bd_set_size(struct block_device *, loff_t size);
2055 extern void bd_forget(struct inode *inode);
2056 extern void bdput(struct block_device *);
2057 extern void invalidate_bdev(struct block_device *);
2058 extern int sync_blockdev(struct block_device *bdev);
2059 extern struct super_block *freeze_bdev(struct block_device *);
2060 extern void emergency_thaw_all(void);
2061 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2062 extern int fsync_bdev(struct block_device *);
2063 #else
2064 static inline void bd_forget(struct inode *inode) {}
2065 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2066 static inline void invalidate_bdev(struct block_device *bdev) {}
2067
2068 static inline struct super_block *freeze_bdev(struct block_device *sb)
2069 {
2070 return NULL;
2071 }
2072
2073 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2074 {
2075 return 0;
2076 }
2077 #endif
2078 extern int sync_filesystem(struct super_block *);
2079 extern const struct file_operations def_blk_fops;
2080 extern const struct file_operations def_chr_fops;
2081 extern const struct file_operations bad_sock_fops;
2082 extern const struct file_operations def_fifo_fops;
2083 #ifdef CONFIG_BLOCK
2084 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2085 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2086 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2087 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2088 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2089 void *holder);
2090 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2091 void *holder);
2092 extern int blkdev_put(struct block_device *bdev, fmode_t mode);
2093 #ifdef CONFIG_SYSFS
2094 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2095 extern void bd_unlink_disk_holder(struct block_device *bdev,
2096 struct gendisk *disk);
2097 #else
2098 static inline int bd_link_disk_holder(struct block_device *bdev,
2099 struct gendisk *disk)
2100 {
2101 return 0;
2102 }
2103 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2104 struct gendisk *disk)
2105 {
2106 }
2107 #endif
2108 #endif
2109
2110 /* fs/char_dev.c */
2111 #define CHRDEV_MAJOR_HASH_SIZE 255
2112 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2113 extern int register_chrdev_region(dev_t, unsigned, const char *);
2114 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2115 unsigned int count, const char *name,
2116 const struct file_operations *fops);
2117 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2118 unsigned int count, const char *name);
2119 extern void unregister_chrdev_region(dev_t, unsigned);
2120 extern void chrdev_show(struct seq_file *,off_t);
2121
2122 static inline int register_chrdev(unsigned int major, const char *name,
2123 const struct file_operations *fops)
2124 {
2125 return __register_chrdev(major, 0, 256, name, fops);
2126 }
2127
2128 static inline void unregister_chrdev(unsigned int major, const char *name)
2129 {
2130 __unregister_chrdev(major, 0, 256, name);
2131 }
2132
2133 /* fs/block_dev.c */
2134 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2135 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2136
2137 #ifdef CONFIG_BLOCK
2138 #define BLKDEV_MAJOR_HASH_SIZE 255
2139 extern const char *__bdevname(dev_t, char *buffer);
2140 extern const char *bdevname(struct block_device *bdev, char *buffer);
2141 extern struct block_device *lookup_bdev(const char *);
2142 extern void blkdev_show(struct seq_file *,off_t);
2143
2144 #else
2145 #define BLKDEV_MAJOR_HASH_SIZE 0
2146 #endif
2147
2148 extern void init_special_inode(struct inode *, umode_t, dev_t);
2149
2150 /* Invalid inode operations -- fs/bad_inode.c */
2151 extern void make_bad_inode(struct inode *);
2152 extern int is_bad_inode(struct inode *);
2153
2154 extern const struct file_operations read_pipefifo_fops;
2155 extern const struct file_operations write_pipefifo_fops;
2156 extern const struct file_operations rdwr_pipefifo_fops;
2157
2158 extern int fs_may_remount_ro(struct super_block *);
2159
2160 #ifdef CONFIG_BLOCK
2161 /*
2162 * return READ, READA, or WRITE
2163 */
2164 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2165
2166 /*
2167 * return data direction, READ or WRITE
2168 */
2169 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2170
2171 extern void check_disk_size_change(struct gendisk *disk,
2172 struct block_device *bdev);
2173 extern int revalidate_disk(struct gendisk *);
2174 extern int check_disk_change(struct block_device *);
2175 extern int __invalidate_device(struct block_device *, bool);
2176 extern int invalidate_partition(struct gendisk *, int);
2177 #endif
2178 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2179 pgoff_t start, pgoff_t end);
2180
2181 static inline void invalidate_remote_inode(struct inode *inode)
2182 {
2183 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2184 S_ISLNK(inode->i_mode))
2185 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2186 }
2187 extern int invalidate_inode_pages2(struct address_space *mapping);
2188 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2189 pgoff_t start, pgoff_t end);
2190 extern int write_inode_now(struct inode *, int);
2191 extern int filemap_fdatawrite(struct address_space *);
2192 extern int filemap_flush(struct address_space *);
2193 extern int filemap_fdatawait(struct address_space *);
2194 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2195 loff_t lend);
2196 extern int filemap_write_and_wait(struct address_space *mapping);
2197 extern int filemap_write_and_wait_range(struct address_space *mapping,
2198 loff_t lstart, loff_t lend);
2199 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2200 loff_t start, loff_t end, int sync_mode);
2201 extern int filemap_fdatawrite_range(struct address_space *mapping,
2202 loff_t start, loff_t end);
2203
2204 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2205 int datasync);
2206 extern int vfs_fsync(struct file *file, int datasync);
2207 extern int generic_write_sync(struct file *file, loff_t pos, loff_t count);
2208 extern void sync_supers(void);
2209 extern void emergency_sync(void);
2210 extern void emergency_remount(void);
2211 #ifdef CONFIG_BLOCK
2212 extern sector_t bmap(struct inode *, sector_t);
2213 #endif
2214 extern int notify_change(struct dentry *, struct iattr *);
2215 extern int inode_permission(struct inode *, int);
2216 extern int generic_permission(struct inode *, int);
2217
2218 static inline bool execute_ok(struct inode *inode)
2219 {
2220 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2221 }
2222
2223 /*
2224 * get_write_access() gets write permission for a file.
2225 * put_write_access() releases this write permission.
2226 * This is used for regular files.
2227 * We cannot support write (and maybe mmap read-write shared) accesses and
2228 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2229 * can have the following values:
2230 * 0: no writers, no VM_DENYWRITE mappings
2231 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2232 * > 0: (i_writecount) users are writing to the file.
2233 *
2234 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2235 * except for the cases where we don't hold i_writecount yet. Then we need to
2236 * use {get,deny}_write_access() - these functions check the sign and refuse
2237 * to do the change if sign is wrong.
2238 */
2239 static inline int get_write_access(struct inode *inode)
2240 {
2241 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2242 }
2243 static inline int deny_write_access(struct file *file)
2244 {
2245 struct inode *inode = file->f_path.dentry->d_inode;
2246 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2247 }
2248 static inline void put_write_access(struct inode * inode)
2249 {
2250 atomic_dec(&inode->i_writecount);
2251 }
2252 static inline void allow_write_access(struct file *file)
2253 {
2254 if (file)
2255 atomic_inc(&file->f_path.dentry->d_inode->i_writecount);
2256 }
2257 #ifdef CONFIG_IMA
2258 static inline void i_readcount_dec(struct inode *inode)
2259 {
2260 BUG_ON(!atomic_read(&inode->i_readcount));
2261 atomic_dec(&inode->i_readcount);
2262 }
2263 static inline void i_readcount_inc(struct inode *inode)
2264 {
2265 atomic_inc(&inode->i_readcount);
2266 }
2267 #else
2268 static inline void i_readcount_dec(struct inode *inode)
2269 {
2270 return;
2271 }
2272 static inline void i_readcount_inc(struct inode *inode)
2273 {
2274 return;
2275 }
2276 #endif
2277 extern int do_pipe_flags(int *, int);
2278 extern struct file *create_read_pipe(struct file *f, int flags);
2279 extern struct file *create_write_pipe(int flags);
2280 extern void free_write_pipe(struct file *);
2281
2282 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2283 extern struct file * open_exec(const char *);
2284
2285 /* fs/dcache.c -- generic fs support functions */
2286 extern int is_subdir(struct dentry *, struct dentry *);
2287 extern int path_is_under(struct path *, struct path *);
2288 extern ino_t find_inode_number(struct dentry *, struct qstr *);
2289
2290 #include <linux/err.h>
2291
2292 /* needed for stackable file system support */
2293 extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
2294
2295 extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
2296
2297 extern int inode_init_always(struct super_block *, struct inode *);
2298 extern void inode_init_once(struct inode *);
2299 extern void address_space_init_once(struct address_space *mapping);
2300 extern void ihold(struct inode * inode);
2301 extern void iput(struct inode *);
2302 extern struct inode * igrab(struct inode *);
2303 extern ino_t iunique(struct super_block *, ino_t);
2304 extern int inode_needs_sync(struct inode *inode);
2305 extern int generic_delete_inode(struct inode *inode);
2306 extern int generic_drop_inode(struct inode *inode);
2307
2308 extern struct inode *ilookup5_nowait(struct super_block *sb,
2309 unsigned long hashval, int (*test)(struct inode *, void *),
2310 void *data);
2311 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2312 int (*test)(struct inode *, void *), void *data);
2313 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2314
2315 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2316 extern struct inode * iget_locked(struct super_block *, unsigned long);
2317 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2318 extern int insert_inode_locked(struct inode *);
2319 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2320 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2321 #else
2322 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2323 #endif
2324 extern void unlock_new_inode(struct inode *);
2325 extern unsigned int get_next_ino(void);
2326
2327 extern void __iget(struct inode * inode);
2328 extern void iget_failed(struct inode *);
2329 extern void end_writeback(struct inode *);
2330 extern void __destroy_inode(struct inode *);
2331 extern struct inode *new_inode_pseudo(struct super_block *sb);
2332 extern struct inode *new_inode(struct super_block *sb);
2333 extern void free_inode_nonrcu(struct inode *inode);
2334 extern int should_remove_suid(struct dentry *);
2335 extern int file_remove_suid(struct file *);
2336
2337 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2338 static inline void insert_inode_hash(struct inode *inode)
2339 {
2340 __insert_inode_hash(inode, inode->i_ino);
2341 }
2342
2343 extern void __remove_inode_hash(struct inode *);
2344 static inline void remove_inode_hash(struct inode *inode)
2345 {
2346 if (!inode_unhashed(inode))
2347 __remove_inode_hash(inode);
2348 }
2349
2350 extern void inode_sb_list_add(struct inode *inode);
2351
2352 #ifdef CONFIG_BLOCK
2353 extern void submit_bio(int, struct bio *);
2354 extern int bdev_read_only(struct block_device *);
2355 #endif
2356 extern int set_blocksize(struct block_device *, int);
2357 extern int sb_set_blocksize(struct super_block *, int);
2358 extern int sb_min_blocksize(struct super_block *, int);
2359
2360 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2361 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2362 extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
2363 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
2364 extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2365 extern ssize_t __generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long,
2366 loff_t *);
2367 extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2368 extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
2369 unsigned long *, loff_t, loff_t *, size_t, size_t);
2370 extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
2371 unsigned long, loff_t, loff_t *, size_t, ssize_t);
2372 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2373 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2374 extern int generic_segment_checks(const struct iovec *iov,
2375 unsigned long *nr_segs, size_t *count, int access_flags);
2376
2377 /* fs/block_dev.c */
2378 extern ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov,
2379 unsigned long nr_segs, loff_t pos);
2380 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2381 int datasync);
2382
2383 /* fs/splice.c */
2384 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2385 struct pipe_inode_info *, size_t, unsigned int);
2386 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2387 struct pipe_inode_info *, size_t, unsigned int);
2388 extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
2389 struct file *, loff_t *, size_t, unsigned int);
2390 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2391 struct file *out, loff_t *, size_t len, unsigned int flags);
2392 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2393 size_t len, unsigned int flags);
2394
2395 extern void
2396 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2397 extern loff_t noop_llseek(struct file *file, loff_t offset, int origin);
2398 extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
2399 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
2400 extern loff_t generic_file_llseek_unlocked(struct file *file, loff_t offset,
2401 int origin);
2402 extern int generic_file_open(struct inode * inode, struct file * filp);
2403 extern int nonseekable_open(struct inode * inode, struct file * filp);
2404
2405 #ifdef CONFIG_FS_XIP
2406 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2407 loff_t *ppos);
2408 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
2409 extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2410 size_t len, loff_t *ppos);
2411 extern int xip_truncate_page(struct address_space *mapping, loff_t from);
2412 #else
2413 static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2414 {
2415 return 0;
2416 }
2417 #endif
2418
2419 #ifdef CONFIG_BLOCK
2420 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2421 loff_t file_offset);
2422
2423 enum {
2424 /* need locking between buffered and direct access */
2425 DIO_LOCKING = 0x01,
2426
2427 /* filesystem does not support filling holes */
2428 DIO_SKIP_HOLES = 0x02,
2429 };
2430
2431 void dio_end_io(struct bio *bio, int error);
2432 void inode_dio_wait(struct inode *inode);
2433 void inode_dio_done(struct inode *inode);
2434
2435 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2436 struct block_device *bdev, const struct iovec *iov, loff_t offset,
2437 unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
2438 dio_submit_t submit_io, int flags);
2439
2440 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
2441 struct inode *inode, const struct iovec *iov, loff_t offset,
2442 unsigned long nr_segs, get_block_t get_block)
2443 {
2444 return __blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov,
2445 offset, nr_segs, get_block, NULL, NULL,
2446 DIO_LOCKING | DIO_SKIP_HOLES);
2447 }
2448 #else
2449 static inline void inode_dio_wait(struct inode *inode)
2450 {
2451 }
2452 #endif
2453
2454 extern const struct file_operations generic_ro_fops;
2455
2456 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2457
2458 extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
2459 extern int vfs_follow_link(struct nameidata *, const char *);
2460 extern int page_readlink(struct dentry *, char __user *, int);
2461 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2462 extern void page_put_link(struct dentry *, struct nameidata *, void *);
2463 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2464 int nofs);
2465 extern int page_symlink(struct inode *inode, const char *symname, int len);
2466 extern const struct inode_operations page_symlink_inode_operations;
2467 extern int generic_readlink(struct dentry *, char __user *, int);
2468 extern void generic_fillattr(struct inode *, struct kstat *);
2469 extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2470 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2471 void inode_add_bytes(struct inode *inode, loff_t bytes);
2472 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2473 loff_t inode_get_bytes(struct inode *inode);
2474 void inode_set_bytes(struct inode *inode, loff_t bytes);
2475
2476 extern int vfs_readdir(struct file *, filldir_t, void *);
2477
2478 extern int vfs_stat(const char __user *, struct kstat *);
2479 extern int vfs_lstat(const char __user *, struct kstat *);
2480 extern int vfs_fstat(unsigned int, struct kstat *);
2481 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2482
2483 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2484 unsigned long arg);
2485 extern int __generic_block_fiemap(struct inode *inode,
2486 struct fiemap_extent_info *fieinfo,
2487 loff_t start, loff_t len,
2488 get_block_t *get_block);
2489 extern int generic_block_fiemap(struct inode *inode,
2490 struct fiemap_extent_info *fieinfo, u64 start,
2491 u64 len, get_block_t *get_block);
2492
2493 extern void get_filesystem(struct file_system_type *fs);
2494 extern void put_filesystem(struct file_system_type *fs);
2495 extern struct file_system_type *get_fs_type(const char *name);
2496 extern struct super_block *get_super(struct block_device *);
2497 extern struct super_block *get_active_super(struct block_device *bdev);
2498 extern struct super_block *user_get_super(dev_t);
2499 extern void drop_super(struct super_block *sb);
2500 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2501 extern void iterate_supers_type(struct file_system_type *,
2502 void (*)(struct super_block *, void *), void *);
2503
2504 extern int dcache_dir_open(struct inode *, struct file *);
2505 extern int dcache_dir_close(struct inode *, struct file *);
2506 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2507 extern int dcache_readdir(struct file *, void *, filldir_t);
2508 extern int simple_setattr(struct dentry *, struct iattr *);
2509 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2510 extern int simple_statfs(struct dentry *, struct kstatfs *);
2511 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2512 extern int simple_unlink(struct inode *, struct dentry *);
2513 extern int simple_rmdir(struct inode *, struct dentry *);
2514 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2515 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2516 extern int simple_empty(struct dentry *);
2517 extern int simple_readpage(struct file *file, struct page *page);
2518 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2519 loff_t pos, unsigned len, unsigned flags,
2520 struct page **pagep, void **fsdata);
2521 extern int simple_write_end(struct file *file, struct address_space *mapping,
2522 loff_t pos, unsigned len, unsigned copied,
2523 struct page *page, void *fsdata);
2524
2525 extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
2526 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2527 extern const struct file_operations simple_dir_operations;
2528 extern const struct inode_operations simple_dir_inode_operations;
2529 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2530 struct dentry *d_alloc_name(struct dentry *, const char *);
2531 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2532 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2533 extern void simple_release_fs(struct vfsmount **mount, int *count);
2534
2535 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2536 loff_t *ppos, const void *from, size_t available);
2537 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2538 const void __user *from, size_t count);
2539
2540 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2541
2542 extern int generic_check_addressable(unsigned, u64);
2543
2544 #ifdef CONFIG_MIGRATION
2545 extern int buffer_migrate_page(struct address_space *,
2546 struct page *, struct page *);
2547 #else
2548 #define buffer_migrate_page NULL
2549 #endif
2550
2551 extern int inode_change_ok(const struct inode *, struct iattr *);
2552 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2553 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2554
2555 extern void file_update_time(struct file *file);
2556
2557 extern int generic_show_options(struct seq_file *m, struct vfsmount *mnt);
2558 extern void save_mount_options(struct super_block *sb, char *options);
2559 extern void replace_mount_options(struct super_block *sb, char *options);
2560
2561 static inline ino_t parent_ino(struct dentry *dentry)
2562 {
2563 ino_t res;
2564
2565 /*
2566 * Don't strictly need d_lock here? If the parent ino could change
2567 * then surely we'd have a deeper race in the caller?
2568 */
2569 spin_lock(&dentry->d_lock);
2570 res = dentry->d_parent->d_inode->i_ino;
2571 spin_unlock(&dentry->d_lock);
2572 return res;
2573 }
2574
2575 /* Transaction based IO helpers */
2576
2577 /*
2578 * An argresp is stored in an allocated page and holds the
2579 * size of the argument or response, along with its content
2580 */
2581 struct simple_transaction_argresp {
2582 ssize_t size;
2583 char data[0];
2584 };
2585
2586 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2587
2588 char *simple_transaction_get(struct file *file, const char __user *buf,
2589 size_t size);
2590 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2591 size_t size, loff_t *pos);
2592 int simple_transaction_release(struct inode *inode, struct file *file);
2593
2594 void simple_transaction_set(struct file *file, size_t n);
2595
2596 /*
2597 * simple attribute files
2598 *
2599 * These attributes behave similar to those in sysfs:
2600 *
2601 * Writing to an attribute immediately sets a value, an open file can be
2602 * written to multiple times.
2603 *
2604 * Reading from an attribute creates a buffer from the value that might get
2605 * read with multiple read calls. When the attribute has been read
2606 * completely, no further read calls are possible until the file is opened
2607 * again.
2608 *
2609 * All attributes contain a text representation of a numeric value
2610 * that are accessed with the get() and set() functions.
2611 */
2612 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2613 static int __fops ## _open(struct inode *inode, struct file *file) \
2614 { \
2615 __simple_attr_check_format(__fmt, 0ull); \
2616 return simple_attr_open(inode, file, __get, __set, __fmt); \
2617 } \
2618 static const struct file_operations __fops = { \
2619 .owner = THIS_MODULE, \
2620 .open = __fops ## _open, \
2621 .release = simple_attr_release, \
2622 .read = simple_attr_read, \
2623 .write = simple_attr_write, \
2624 .llseek = generic_file_llseek, \
2625 };
2626
2627 static inline void __attribute__((format(printf, 1, 2)))
2628 __simple_attr_check_format(const char *fmt, ...)
2629 {
2630 /* don't do anything, just let the compiler check the arguments; */
2631 }
2632
2633 int simple_attr_open(struct inode *inode, struct file *file,
2634 int (*get)(void *, u64 *), int (*set)(void *, u64),
2635 const char *fmt);
2636 int simple_attr_release(struct inode *inode, struct file *file);
2637 ssize_t simple_attr_read(struct file *file, char __user *buf,
2638 size_t len, loff_t *ppos);
2639 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2640 size_t len, loff_t *ppos);
2641
2642 struct ctl_table;
2643 int proc_nr_files(struct ctl_table *table, int write,
2644 void __user *buffer, size_t *lenp, loff_t *ppos);
2645 int proc_nr_dentry(struct ctl_table *table, int write,
2646 void __user *buffer, size_t *lenp, loff_t *ppos);
2647 int proc_nr_inodes(struct ctl_table *table, int write,
2648 void __user *buffer, size_t *lenp, loff_t *ppos);
2649 int __init get_filesystem_list(char *buf);
2650
2651 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
2652 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
2653
2654 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2655 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2656 (flag & __FMODE_NONOTIFY)))
2657
2658 static inline int is_sxid(mode_t mode)
2659 {
2660 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
2661 }
2662
2663 static inline void inode_has_no_xattr(struct inode *inode)
2664 {
2665 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
2666 inode->i_flags |= S_NOSEC;
2667 }
2668
2669 #endif /* __KERNEL__ */
2670 #endif /* _LINUX_FS_H */