Merge branch 'for_3.2/gpio-cleanup' of git://gitorious.org/khilman/linux-omap-pm...
[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 READ_META (READ | REQ_META)
166 #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
167 #define WRITE_ODIRECT (WRITE | REQ_SYNC)
168 #define WRITE_META (WRITE | REQ_META)
169 #define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
170 #define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
171 #define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
172
173 #define SEL_IN 1
174 #define SEL_OUT 2
175 #define SEL_EX 4
176
177 /* public flags for file_system_type */
178 #define FS_REQUIRES_DEV 1
179 #define FS_BINARY_MOUNTDATA 2
180 #define FS_HAS_SUBTYPE 4
181 #define FS_REVAL_DOT 16384 /* Check the paths ".", ".." for staleness */
182 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move()
183 * during rename() internally.
184 */
185
186 /*
187 * These are the fs-independent mount-flags: up to 32 flags are supported
188 */
189 #define MS_RDONLY 1 /* Mount read-only */
190 #define MS_NOSUID 2 /* Ignore suid and sgid bits */
191 #define MS_NODEV 4 /* Disallow access to device special files */
192 #define MS_NOEXEC 8 /* Disallow program execution */
193 #define MS_SYNCHRONOUS 16 /* Writes are synced at once */
194 #define MS_REMOUNT 32 /* Alter flags of a mounted FS */
195 #define MS_MANDLOCK 64 /* Allow mandatory locks on an FS */
196 #define MS_DIRSYNC 128 /* Directory modifications are synchronous */
197 #define MS_NOATIME 1024 /* Do not update access times. */
198 #define MS_NODIRATIME 2048 /* Do not update directory access times */
199 #define MS_BIND 4096
200 #define MS_MOVE 8192
201 #define MS_REC 16384
202 #define MS_VERBOSE 32768 /* War is peace. Verbosity is silence.
203 MS_VERBOSE is deprecated. */
204 #define MS_SILENT 32768
205 #define MS_POSIXACL (1<<16) /* VFS does not apply the umask */
206 #define MS_UNBINDABLE (1<<17) /* change to unbindable */
207 #define MS_PRIVATE (1<<18) /* change to private */
208 #define MS_SLAVE (1<<19) /* change to slave */
209 #define MS_SHARED (1<<20) /* change to shared */
210 #define MS_RELATIME (1<<21) /* Update atime relative to mtime/ctime. */
211 #define MS_KERNMOUNT (1<<22) /* this is a kern_mount call */
212 #define MS_I_VERSION (1<<23) /* Update inode I_version field */
213 #define MS_STRICTATIME (1<<24) /* Always perform atime updates */
214 #define MS_NOSEC (1<<28)
215 #define MS_BORN (1<<29)
216 #define MS_ACTIVE (1<<30)
217 #define MS_NOUSER (1<<31)
218
219 /*
220 * Superblock flags that can be altered by MS_REMOUNT
221 */
222 #define MS_RMT_MASK (MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK|MS_I_VERSION)
223
224 /*
225 * Old magic mount flag and mask
226 */
227 #define MS_MGC_VAL 0xC0ED0000
228 #define MS_MGC_MSK 0xffff0000
229
230 /* Inode flags - they have nothing to superblock flags now */
231
232 #define S_SYNC 1 /* Writes are synced at once */
233 #define S_NOATIME 2 /* Do not update access times */
234 #define S_APPEND 4 /* Append-only file */
235 #define S_IMMUTABLE 8 /* Immutable file */
236 #define S_DEAD 16 /* removed, but still open directory */
237 #define S_NOQUOTA 32 /* Inode is not counted to quota */
238 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
239 #define S_NOCMTIME 128 /* Do not update file c/mtime */
240 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
241 #define S_PRIVATE 512 /* Inode is fs-internal */
242 #define S_IMA 1024 /* Inode has an associated IMA struct */
243 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
244 #define S_NOSEC 4096 /* no suid or xattr security attributes */
245
246 /*
247 * Note that nosuid etc flags are inode-specific: setting some file-system
248 * flags just means all the inodes inherit those flags by default. It might be
249 * possible to override it selectively if you really wanted to with some
250 * ioctl() that is not currently implemented.
251 *
252 * Exception: MS_RDONLY is always applied to the entire file system.
253 *
254 * Unfortunately, it is possible to change a filesystems flags with it mounted
255 * with files in use. This means that all of the inodes will not have their
256 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
257 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
258 */
259 #define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))
260
261 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
262 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
263 ((inode)->i_flags & S_SYNC))
264 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
265 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
266 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
267 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
268 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
269
270 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
271 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
272 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
273 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
274
275 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
276 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
277 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
278 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
279 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
280 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
281 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
282
283 /* the read-only stuff doesn't really belong here, but any other place is
284 probably as bad and I don't want to create yet another include file. */
285
286 #define BLKROSET _IO(0x12,93) /* set device read-only (0 = read-write) */
287 #define BLKROGET _IO(0x12,94) /* get read-only status (0 = read_write) */
288 #define BLKRRPART _IO(0x12,95) /* re-read partition table */
289 #define BLKGETSIZE _IO(0x12,96) /* return device size /512 (long *arg) */
290 #define BLKFLSBUF _IO(0x12,97) /* flush buffer cache */
291 #define BLKRASET _IO(0x12,98) /* set read ahead for block device */
292 #define BLKRAGET _IO(0x12,99) /* get current read ahead setting */
293 #define BLKFRASET _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
294 #define BLKFRAGET _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
295 #define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
296 #define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
297 #define BLKSSZGET _IO(0x12,104)/* get block device sector size */
298 #if 0
299 #define BLKPG _IO(0x12,105)/* See blkpg.h */
300
301 /* Some people are morons. Do not use sizeof! */
302
303 #define BLKELVGET _IOR(0x12,106,size_t)/* elevator get */
304 #define BLKELVSET _IOW(0x12,107,size_t)/* elevator set */
305 /* This was here just to show that the number is taken -
306 probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
307 #endif
308 /* A jump here: 108-111 have been used for various private purposes. */
309 #define BLKBSZGET _IOR(0x12,112,size_t)
310 #define BLKBSZSET _IOW(0x12,113,size_t)
311 #define BLKGETSIZE64 _IOR(0x12,114,size_t) /* return device size in bytes (u64 *arg) */
312 #define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup)
313 #define BLKTRACESTART _IO(0x12,116)
314 #define BLKTRACESTOP _IO(0x12,117)
315 #define BLKTRACETEARDOWN _IO(0x12,118)
316 #define BLKDISCARD _IO(0x12,119)
317 #define BLKIOMIN _IO(0x12,120)
318 #define BLKIOOPT _IO(0x12,121)
319 #define BLKALIGNOFF _IO(0x12,122)
320 #define BLKPBSZGET _IO(0x12,123)
321 #define BLKDISCARDZEROES _IO(0x12,124)
322 #define BLKSECDISCARD _IO(0x12,125)
323
324 #define BMAP_IOCTL 1 /* obsolete - kept for compatibility */
325 #define FIBMAP _IO(0x00,1) /* bmap access */
326 #define FIGETBSZ _IO(0x00,2) /* get the block size used for bmap */
327 #define FIFREEZE _IOWR('X', 119, int) /* Freeze */
328 #define FITHAW _IOWR('X', 120, int) /* Thaw */
329 #define FITRIM _IOWR('X', 121, struct fstrim_range) /* Trim */
330
331 #define FS_IOC_GETFLAGS _IOR('f', 1, long)
332 #define FS_IOC_SETFLAGS _IOW('f', 2, long)
333 #define FS_IOC_GETVERSION _IOR('v', 1, long)
334 #define FS_IOC_SETVERSION _IOW('v', 2, long)
335 #define FS_IOC_FIEMAP _IOWR('f', 11, struct fiemap)
336 #define FS_IOC32_GETFLAGS _IOR('f', 1, int)
337 #define FS_IOC32_SETFLAGS _IOW('f', 2, int)
338 #define FS_IOC32_GETVERSION _IOR('v', 1, int)
339 #define FS_IOC32_SETVERSION _IOW('v', 2, int)
340
341 /*
342 * Inode flags (FS_IOC_GETFLAGS / FS_IOC_SETFLAGS)
343 */
344 #define FS_SECRM_FL 0x00000001 /* Secure deletion */
345 #define FS_UNRM_FL 0x00000002 /* Undelete */
346 #define FS_COMPR_FL 0x00000004 /* Compress file */
347 #define FS_SYNC_FL 0x00000008 /* Synchronous updates */
348 #define FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
349 #define FS_APPEND_FL 0x00000020 /* writes to file may only append */
350 #define FS_NODUMP_FL 0x00000040 /* do not dump file */
351 #define FS_NOATIME_FL 0x00000080 /* do not update atime */
352 /* Reserved for compression usage... */
353 #define FS_DIRTY_FL 0x00000100
354 #define FS_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
355 #define FS_NOCOMP_FL 0x00000400 /* Don't compress */
356 #define FS_ECOMPR_FL 0x00000800 /* Compression error */
357 /* End compression flags --- maybe not all used */
358 #define FS_BTREE_FL 0x00001000 /* btree format dir */
359 #define FS_INDEX_FL 0x00001000 /* hash-indexed directory */
360 #define FS_IMAGIC_FL 0x00002000 /* AFS directory */
361 #define FS_JOURNAL_DATA_FL 0x00004000 /* Reserved for ext3 */
362 #define FS_NOTAIL_FL 0x00008000 /* file tail should not be merged */
363 #define FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
364 #define FS_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
365 #define FS_EXTENT_FL 0x00080000 /* Extents */
366 #define FS_DIRECTIO_FL 0x00100000 /* Use direct i/o */
367 #define FS_NOCOW_FL 0x00800000 /* Do not cow file */
368 #define FS_RESERVED_FL 0x80000000 /* reserved for ext2 lib */
369
370 #define FS_FL_USER_VISIBLE 0x0003DFFF /* User visible flags */
371 #define FS_FL_USER_MODIFIABLE 0x000380FF /* User modifiable flags */
372
373
374 #define SYNC_FILE_RANGE_WAIT_BEFORE 1
375 #define SYNC_FILE_RANGE_WRITE 2
376 #define SYNC_FILE_RANGE_WAIT_AFTER 4
377
378 #ifdef __KERNEL__
379
380 #include <linux/linkage.h>
381 #include <linux/wait.h>
382 #include <linux/kdev_t.h>
383 #include <linux/dcache.h>
384 #include <linux/path.h>
385 #include <linux/stat.h>
386 #include <linux/cache.h>
387 #include <linux/list.h>
388 #include <linux/radix-tree.h>
389 #include <linux/prio_tree.h>
390 #include <linux/init.h>
391 #include <linux/pid.h>
392 #include <linux/mutex.h>
393 #include <linux/capability.h>
394 #include <linux/semaphore.h>
395 #include <linux/fiemap.h>
396 #include <linux/rculist_bl.h>
397 #include <linux/shrinker.h>
398 #include <linux/atomic.h>
399
400 #include <asm/byteorder.h>
401
402 struct export_operations;
403 struct hd_geometry;
404 struct iovec;
405 struct nameidata;
406 struct kiocb;
407 struct kobject;
408 struct pipe_inode_info;
409 struct poll_table_struct;
410 struct kstatfs;
411 struct vm_area_struct;
412 struct vfsmount;
413 struct cred;
414
415 extern void __init inode_init(void);
416 extern void __init inode_init_early(void);
417 extern void __init files_init(unsigned long);
418
419 extern struct files_stat_struct files_stat;
420 extern unsigned long get_max_files(void);
421 extern int sysctl_nr_open;
422 extern struct inodes_stat_t inodes_stat;
423 extern int leases_enable, lease_break_time;
424
425 struct buffer_head;
426 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
427 struct buffer_head *bh_result, int create);
428 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
429 ssize_t bytes, void *private, int ret,
430 bool is_async);
431
432 /*
433 * Attribute flags. These should be or-ed together to figure out what
434 * has been changed!
435 */
436 #define ATTR_MODE (1 << 0)
437 #define ATTR_UID (1 << 1)
438 #define ATTR_GID (1 << 2)
439 #define ATTR_SIZE (1 << 3)
440 #define ATTR_ATIME (1 << 4)
441 #define ATTR_MTIME (1 << 5)
442 #define ATTR_CTIME (1 << 6)
443 #define ATTR_ATIME_SET (1 << 7)
444 #define ATTR_MTIME_SET (1 << 8)
445 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
446 #define ATTR_ATTR_FLAG (1 << 10)
447 #define ATTR_KILL_SUID (1 << 11)
448 #define ATTR_KILL_SGID (1 << 12)
449 #define ATTR_FILE (1 << 13)
450 #define ATTR_KILL_PRIV (1 << 14)
451 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
452 #define ATTR_TIMES_SET (1 << 16)
453
454 /*
455 * This is the Inode Attributes structure, used for notify_change(). It
456 * uses the above definitions as flags, to know which values have changed.
457 * Also, in this manner, a Filesystem can look at only the values it cares
458 * about. Basically, these are the attributes that the VFS layer can
459 * request to change from the FS layer.
460 *
461 * Derek Atkins <warlord@MIT.EDU> 94-10-20
462 */
463 struct iattr {
464 unsigned int ia_valid;
465 umode_t ia_mode;
466 uid_t ia_uid;
467 gid_t ia_gid;
468 loff_t ia_size;
469 struct timespec ia_atime;
470 struct timespec ia_mtime;
471 struct timespec ia_ctime;
472
473 /*
474 * Not an attribute, but an auxiliary info for filesystems wanting to
475 * implement an ftruncate() like method. NOTE: filesystem should
476 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
477 */
478 struct file *ia_file;
479 };
480
481 /*
482 * Includes for diskquotas.
483 */
484 #include <linux/quota.h>
485
486 /**
487 * enum positive_aop_returns - aop return codes with specific semantics
488 *
489 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
490 * completed, that the page is still locked, and
491 * should be considered active. The VM uses this hint
492 * to return the page to the active list -- it won't
493 * be a candidate for writeback again in the near
494 * future. Other callers must be careful to unlock
495 * the page if they get this return. Returned by
496 * writepage();
497 *
498 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
499 * unlocked it and the page might have been truncated.
500 * The caller should back up to acquiring a new page and
501 * trying again. The aop will be taking reasonable
502 * precautions not to livelock. If the caller held a page
503 * reference, it should drop it before retrying. Returned
504 * by readpage().
505 *
506 * address_space_operation functions return these large constants to indicate
507 * special semantics to the caller. These are much larger than the bytes in a
508 * page to allow for functions that return the number of bytes operated on in a
509 * given page.
510 */
511
512 enum positive_aop_returns {
513 AOP_WRITEPAGE_ACTIVATE = 0x80000,
514 AOP_TRUNCATED_PAGE = 0x80001,
515 };
516
517 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
518 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
519 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
520 * helper code (eg buffer layer)
521 * to clear GFP_FS from alloc */
522
523 /*
524 * oh the beauties of C type declarations.
525 */
526 struct page;
527 struct address_space;
528 struct writeback_control;
529
530 struct iov_iter {
531 const struct iovec *iov;
532 unsigned long nr_segs;
533 size_t iov_offset;
534 size_t count;
535 };
536
537 size_t iov_iter_copy_from_user_atomic(struct page *page,
538 struct iov_iter *i, unsigned long offset, size_t bytes);
539 size_t iov_iter_copy_from_user(struct page *page,
540 struct iov_iter *i, unsigned long offset, size_t bytes);
541 void iov_iter_advance(struct iov_iter *i, size_t bytes);
542 int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes);
543 size_t iov_iter_single_seg_count(struct iov_iter *i);
544
545 static inline void iov_iter_init(struct iov_iter *i,
546 const struct iovec *iov, unsigned long nr_segs,
547 size_t count, size_t written)
548 {
549 i->iov = iov;
550 i->nr_segs = nr_segs;
551 i->iov_offset = 0;
552 i->count = count + written;
553
554 iov_iter_advance(i, written);
555 }
556
557 static inline size_t iov_iter_count(struct iov_iter *i)
558 {
559 return i->count;
560 }
561
562 /*
563 * "descriptor" for what we're up to with a read.
564 * This allows us to use the same read code yet
565 * have multiple different users of the data that
566 * we read from a file.
567 *
568 * The simplest case just copies the data to user
569 * mode.
570 */
571 typedef struct {
572 size_t written;
573 size_t count;
574 union {
575 char __user *buf;
576 void *data;
577 } arg;
578 int error;
579 } read_descriptor_t;
580
581 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
582 unsigned long, unsigned long);
583
584 struct address_space_operations {
585 int (*writepage)(struct page *page, struct writeback_control *wbc);
586 int (*readpage)(struct file *, struct page *);
587
588 /* Write back some dirty pages from this mapping. */
589 int (*writepages)(struct address_space *, struct writeback_control *);
590
591 /* Set a page dirty. Return true if this dirtied it */
592 int (*set_page_dirty)(struct page *page);
593
594 int (*readpages)(struct file *filp, struct address_space *mapping,
595 struct list_head *pages, unsigned nr_pages);
596
597 int (*write_begin)(struct file *, struct address_space *mapping,
598 loff_t pos, unsigned len, unsigned flags,
599 struct page **pagep, void **fsdata);
600 int (*write_end)(struct file *, struct address_space *mapping,
601 loff_t pos, unsigned len, unsigned copied,
602 struct page *page, void *fsdata);
603
604 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
605 sector_t (*bmap)(struct address_space *, sector_t);
606 void (*invalidatepage) (struct page *, unsigned long);
607 int (*releasepage) (struct page *, gfp_t);
608 void (*freepage)(struct page *);
609 ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
610 loff_t offset, unsigned long nr_segs);
611 int (*get_xip_mem)(struct address_space *, pgoff_t, int,
612 void **, unsigned long *);
613 /* migrate the contents of a page to the specified target */
614 int (*migratepage) (struct address_space *,
615 struct page *, struct page *);
616 int (*launder_page) (struct page *);
617 int (*is_partially_uptodate) (struct page *, read_descriptor_t *,
618 unsigned long);
619 int (*error_remove_page)(struct address_space *, struct page *);
620 };
621
622 extern const struct address_space_operations empty_aops;
623
624 /*
625 * pagecache_write_begin/pagecache_write_end must be used by general code
626 * to write into the pagecache.
627 */
628 int pagecache_write_begin(struct file *, struct address_space *mapping,
629 loff_t pos, unsigned len, unsigned flags,
630 struct page **pagep, void **fsdata);
631
632 int pagecache_write_end(struct file *, struct address_space *mapping,
633 loff_t pos, unsigned len, unsigned copied,
634 struct page *page, void *fsdata);
635
636 struct backing_dev_info;
637 struct address_space {
638 struct inode *host; /* owner: inode, block_device */
639 struct radix_tree_root page_tree; /* radix tree of all pages */
640 spinlock_t tree_lock; /* and lock protecting it */
641 unsigned int i_mmap_writable;/* count VM_SHARED mappings */
642 struct prio_tree_root i_mmap; /* tree of private and shared mappings */
643 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
644 struct mutex i_mmap_mutex; /* protect tree, count, list */
645 /* Protected by tree_lock together with the radix tree */
646 unsigned long nrpages; /* number of total pages */
647 pgoff_t writeback_index;/* writeback starts here */
648 const struct address_space_operations *a_ops; /* methods */
649 unsigned long flags; /* error bits/gfp mask */
650 struct backing_dev_info *backing_dev_info; /* device readahead, etc */
651 spinlock_t private_lock; /* for use by the address_space */
652 struct list_head private_list; /* ditto */
653 struct address_space *assoc_mapping; /* ditto */
654 } __attribute__((aligned(sizeof(long))));
655 /*
656 * On most architectures that alignment is already the case; but
657 * must be enforced here for CRIS, to let the least significant bit
658 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
659 */
660
661 struct block_device {
662 dev_t bd_dev; /* not a kdev_t - it's a search key */
663 int bd_openers;
664 struct inode * bd_inode; /* will die */
665 struct super_block * bd_super;
666 struct mutex bd_mutex; /* open/close mutex */
667 struct list_head bd_inodes;
668 void * bd_claiming;
669 void * bd_holder;
670 int bd_holders;
671 bool bd_write_holder;
672 #ifdef CONFIG_SYSFS
673 struct list_head bd_holder_disks;
674 #endif
675 struct block_device * bd_contains;
676 unsigned bd_block_size;
677 struct hd_struct * bd_part;
678 /* number of times partitions within this device have been opened. */
679 unsigned bd_part_count;
680 int bd_invalidated;
681 struct gendisk * bd_disk;
682 struct list_head bd_list;
683 /*
684 * Private data. You must have bd_claim'ed the block_device
685 * to use this. NOTE: bd_claim allows an owner to claim
686 * the same device multiple times, the owner must take special
687 * care to not mess up bd_private for that case.
688 */
689 unsigned long bd_private;
690
691 /* The counter of freeze processes */
692 int bd_fsfreeze_count;
693 /* Mutex for freeze */
694 struct mutex bd_fsfreeze_mutex;
695 };
696
697 /*
698 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
699 * radix trees
700 */
701 #define PAGECACHE_TAG_DIRTY 0
702 #define PAGECACHE_TAG_WRITEBACK 1
703 #define PAGECACHE_TAG_TOWRITE 2
704
705 int mapping_tagged(struct address_space *mapping, int tag);
706
707 /*
708 * Might pages of this file be mapped into userspace?
709 */
710 static inline int mapping_mapped(struct address_space *mapping)
711 {
712 return !prio_tree_empty(&mapping->i_mmap) ||
713 !list_empty(&mapping->i_mmap_nonlinear);
714 }
715
716 /*
717 * Might pages of this file have been modified in userspace?
718 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
719 * marks vma as VM_SHARED if it is shared, and the file was opened for
720 * writing i.e. vma may be mprotected writable even if now readonly.
721 */
722 static inline int mapping_writably_mapped(struct address_space *mapping)
723 {
724 return mapping->i_mmap_writable != 0;
725 }
726
727 /*
728 * Use sequence counter to get consistent i_size on 32-bit processors.
729 */
730 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
731 #include <linux/seqlock.h>
732 #define __NEED_I_SIZE_ORDERED
733 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
734 #else
735 #define i_size_ordered_init(inode) do { } while (0)
736 #endif
737
738 struct posix_acl;
739 #define ACL_NOT_CACHED ((void *)(-1))
740
741 #define IOP_FASTPERM 0x0001
742 #define IOP_LOOKUP 0x0002
743 #define IOP_NOFOLLOW 0x0004
744
745 /*
746 * Keep mostly read-only and often accessed (especially for
747 * the RCU path lookup and 'stat' data) fields at the beginning
748 * of the 'struct inode'
749 */
750 struct inode {
751 umode_t i_mode;
752 unsigned short i_opflags;
753 uid_t i_uid;
754 gid_t i_gid;
755 unsigned int i_flags;
756
757 #ifdef CONFIG_FS_POSIX_ACL
758 struct posix_acl *i_acl;
759 struct posix_acl *i_default_acl;
760 #endif
761
762 const struct inode_operations *i_op;
763 struct super_block *i_sb;
764 struct address_space *i_mapping;
765
766 #ifdef CONFIG_SECURITY
767 void *i_security;
768 #endif
769
770 /* Stat data, not accessed from path walking */
771 unsigned long i_ino;
772 unsigned int i_nlink;
773 dev_t i_rdev;
774 loff_t i_size;
775 struct timespec i_atime;
776 struct timespec i_mtime;
777 struct timespec i_ctime;
778 unsigned int i_blkbits;
779 blkcnt_t i_blocks;
780
781 #ifdef __NEED_I_SIZE_ORDERED
782 seqcount_t i_size_seqcount;
783 #endif
784
785 /* Misc */
786 unsigned long i_state;
787 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
788 struct mutex i_mutex;
789
790 unsigned long dirtied_when; /* jiffies of first dirtying */
791
792 struct hlist_node i_hash;
793 struct list_head i_wb_list; /* backing dev IO list */
794 struct list_head i_lru; /* inode LRU list */
795 struct list_head i_sb_list;
796 union {
797 struct list_head i_dentry;
798 struct rcu_head i_rcu;
799 };
800 atomic_t i_count;
801 u64 i_version;
802 unsigned short i_bytes;
803 atomic_t i_dio_count;
804 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
805 struct file_lock *i_flock;
806 struct address_space i_data;
807 #ifdef CONFIG_QUOTA
808 struct dquot *i_dquot[MAXQUOTAS];
809 #endif
810 struct list_head i_devices;
811 union {
812 struct pipe_inode_info *i_pipe;
813 struct block_device *i_bdev;
814 struct cdev *i_cdev;
815 };
816
817 __u32 i_generation;
818
819 #ifdef CONFIG_FSNOTIFY
820 __u32 i_fsnotify_mask; /* all events this inode cares about */
821 struct hlist_head i_fsnotify_marks;
822 #endif
823
824 #ifdef CONFIG_IMA
825 atomic_t i_readcount; /* struct files open RO */
826 #endif
827 atomic_t i_writecount;
828 void *i_private; /* fs or device private pointer */
829 };
830
831 static inline int inode_unhashed(struct inode *inode)
832 {
833 return hlist_unhashed(&inode->i_hash);
834 }
835
836 /*
837 * inode->i_mutex nesting subclasses for the lock validator:
838 *
839 * 0: the object of the current VFS operation
840 * 1: parent
841 * 2: child/target
842 * 3: quota file
843 *
844 * The locking order between these classes is
845 * parent -> child -> normal -> xattr -> quota
846 */
847 enum inode_i_mutex_lock_class
848 {
849 I_MUTEX_NORMAL,
850 I_MUTEX_PARENT,
851 I_MUTEX_CHILD,
852 I_MUTEX_XATTR,
853 I_MUTEX_QUOTA
854 };
855
856 /*
857 * NOTE: in a 32bit arch with a preemptable kernel and
858 * an UP compile the i_size_read/write must be atomic
859 * with respect to the local cpu (unlike with preempt disabled),
860 * but they don't need to be atomic with respect to other cpus like in
861 * true SMP (so they need either to either locally disable irq around
862 * the read or for example on x86 they can be still implemented as a
863 * cmpxchg8b without the need of the lock prefix). For SMP compiles
864 * and 64bit archs it makes no difference if preempt is enabled or not.
865 */
866 static inline loff_t i_size_read(const struct inode *inode)
867 {
868 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
869 loff_t i_size;
870 unsigned int seq;
871
872 do {
873 seq = read_seqcount_begin(&inode->i_size_seqcount);
874 i_size = inode->i_size;
875 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
876 return i_size;
877 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
878 loff_t i_size;
879
880 preempt_disable();
881 i_size = inode->i_size;
882 preempt_enable();
883 return i_size;
884 #else
885 return inode->i_size;
886 #endif
887 }
888
889 /*
890 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
891 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
892 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
893 */
894 static inline void i_size_write(struct inode *inode, loff_t i_size)
895 {
896 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
897 write_seqcount_begin(&inode->i_size_seqcount);
898 inode->i_size = i_size;
899 write_seqcount_end(&inode->i_size_seqcount);
900 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
901 preempt_disable();
902 inode->i_size = i_size;
903 preempt_enable();
904 #else
905 inode->i_size = i_size;
906 #endif
907 }
908
909 static inline unsigned iminor(const struct inode *inode)
910 {
911 return MINOR(inode->i_rdev);
912 }
913
914 static inline unsigned imajor(const struct inode *inode)
915 {
916 return MAJOR(inode->i_rdev);
917 }
918
919 extern struct block_device *I_BDEV(struct inode *inode);
920
921 struct fown_struct {
922 rwlock_t lock; /* protects pid, uid, euid fields */
923 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
924 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
925 uid_t uid, euid; /* uid/euid of process setting the owner */
926 int signum; /* posix.1b rt signal to be delivered on IO */
927 };
928
929 /*
930 * Track a single file's readahead state
931 */
932 struct file_ra_state {
933 pgoff_t start; /* where readahead started */
934 unsigned int size; /* # of readahead pages */
935 unsigned int async_size; /* do asynchronous readahead when
936 there are only # of pages ahead */
937
938 unsigned int ra_pages; /* Maximum readahead window */
939 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
940 loff_t prev_pos; /* Cache last read() position */
941 };
942
943 /*
944 * Check if @index falls in the readahead windows.
945 */
946 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
947 {
948 return (index >= ra->start &&
949 index < ra->start + ra->size);
950 }
951
952 #define FILE_MNT_WRITE_TAKEN 1
953 #define FILE_MNT_WRITE_RELEASED 2
954
955 struct file {
956 /*
957 * fu_list becomes invalid after file_free is called and queued via
958 * fu_rcuhead for RCU freeing
959 */
960 union {
961 struct list_head fu_list;
962 struct rcu_head fu_rcuhead;
963 } f_u;
964 struct path f_path;
965 #define f_dentry f_path.dentry
966 #define f_vfsmnt f_path.mnt
967 const struct file_operations *f_op;
968 spinlock_t f_lock; /* f_ep_links, f_flags, no IRQ */
969 #ifdef CONFIG_SMP
970 int f_sb_list_cpu;
971 #endif
972 atomic_long_t f_count;
973 unsigned int f_flags;
974 fmode_t f_mode;
975 loff_t f_pos;
976 struct fown_struct f_owner;
977 const struct cred *f_cred;
978 struct file_ra_state f_ra;
979
980 u64 f_version;
981 #ifdef CONFIG_SECURITY
982 void *f_security;
983 #endif
984 /* needed for tty driver, and maybe others */
985 void *private_data;
986
987 #ifdef CONFIG_EPOLL
988 /* Used by fs/eventpoll.c to link all the hooks to this file */
989 struct list_head f_ep_links;
990 #endif /* #ifdef CONFIG_EPOLL */
991 struct address_space *f_mapping;
992 #ifdef CONFIG_DEBUG_WRITECOUNT
993 unsigned long f_mnt_write_state;
994 #endif
995 };
996
997 struct file_handle {
998 __u32 handle_bytes;
999 int handle_type;
1000 /* file identifier */
1001 unsigned char f_handle[0];
1002 };
1003
1004 #define get_file(x) atomic_long_inc(&(x)->f_count)
1005 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
1006 #define file_count(x) atomic_long_read(&(x)->f_count)
1007
1008 #ifdef CONFIG_DEBUG_WRITECOUNT
1009 static inline void file_take_write(struct file *f)
1010 {
1011 WARN_ON(f->f_mnt_write_state != 0);
1012 f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN;
1013 }
1014 static inline void file_release_write(struct file *f)
1015 {
1016 f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED;
1017 }
1018 static inline void file_reset_write(struct file *f)
1019 {
1020 f->f_mnt_write_state = 0;
1021 }
1022 static inline void file_check_state(struct file *f)
1023 {
1024 /*
1025 * At this point, either both or neither of these bits
1026 * should be set.
1027 */
1028 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN);
1029 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED);
1030 }
1031 static inline int file_check_writeable(struct file *f)
1032 {
1033 if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN)
1034 return 0;
1035 printk(KERN_WARNING "writeable file with no "
1036 "mnt_want_write()\n");
1037 WARN_ON(1);
1038 return -EINVAL;
1039 }
1040 #else /* !CONFIG_DEBUG_WRITECOUNT */
1041 static inline void file_take_write(struct file *filp) {}
1042 static inline void file_release_write(struct file *filp) {}
1043 static inline void file_reset_write(struct file *filp) {}
1044 static inline void file_check_state(struct file *filp) {}
1045 static inline int file_check_writeable(struct file *filp)
1046 {
1047 return 0;
1048 }
1049 #endif /* CONFIG_DEBUG_WRITECOUNT */
1050
1051 #define MAX_NON_LFS ((1UL<<31) - 1)
1052
1053 /* Page cache limit. The filesystems should put that into their s_maxbytes
1054 limits, otherwise bad things can happen in VM. */
1055 #if BITS_PER_LONG==32
1056 #define MAX_LFS_FILESIZE (((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
1057 #elif BITS_PER_LONG==64
1058 #define MAX_LFS_FILESIZE 0x7fffffffffffffffUL
1059 #endif
1060
1061 #define FL_POSIX 1
1062 #define FL_FLOCK 2
1063 #define FL_ACCESS 8 /* not trying to lock, just looking */
1064 #define FL_EXISTS 16 /* when unlocking, test for existence */
1065 #define FL_LEASE 32 /* lease held on this file */
1066 #define FL_CLOSE 64 /* unlock on close */
1067 #define FL_SLEEP 128 /* A blocking lock */
1068
1069 /*
1070 * Special return value from posix_lock_file() and vfs_lock_file() for
1071 * asynchronous locking.
1072 */
1073 #define FILE_LOCK_DEFERRED 1
1074
1075 /*
1076 * The POSIX file lock owner is determined by
1077 * the "struct files_struct" in the thread group
1078 * (or NULL for no owner - BSD locks).
1079 *
1080 * Lockd stuffs a "host" pointer into this.
1081 */
1082 typedef struct files_struct *fl_owner_t;
1083
1084 struct file_lock_operations {
1085 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1086 void (*fl_release_private)(struct file_lock *);
1087 };
1088
1089 struct lock_manager_operations {
1090 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
1091 void (*lm_notify)(struct file_lock *); /* unblock callback */
1092 int (*lm_grant)(struct file_lock *, struct file_lock *, int);
1093 void (*lm_release_private)(struct file_lock *);
1094 void (*lm_break)(struct file_lock *);
1095 int (*lm_change)(struct file_lock **, int);
1096 };
1097
1098 struct lock_manager {
1099 struct list_head list;
1100 };
1101
1102 void locks_start_grace(struct lock_manager *);
1103 void locks_end_grace(struct lock_manager *);
1104 int locks_in_grace(void);
1105
1106 /* that will die - we need it for nfs_lock_info */
1107 #include <linux/nfs_fs_i.h>
1108
1109 struct file_lock {
1110 struct file_lock *fl_next; /* singly linked list for this inode */
1111 struct list_head fl_link; /* doubly linked list of all locks */
1112 struct list_head fl_block; /* circular list of blocked processes */
1113 fl_owner_t fl_owner;
1114 unsigned char fl_flags;
1115 unsigned char fl_type;
1116 unsigned int fl_pid;
1117 struct pid *fl_nspid;
1118 wait_queue_head_t fl_wait;
1119 struct file *fl_file;
1120 loff_t fl_start;
1121 loff_t fl_end;
1122
1123 struct fasync_struct * fl_fasync; /* for lease break notifications */
1124 unsigned long fl_break_time; /* for nonblocking lease breaks */
1125
1126 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1127 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1128 union {
1129 struct nfs_lock_info nfs_fl;
1130 struct nfs4_lock_info nfs4_fl;
1131 struct {
1132 struct list_head link; /* link in AFS vnode's pending_locks list */
1133 int state; /* state of grant or error if -ve */
1134 } afs;
1135 } fl_u;
1136 };
1137
1138 /* The following constant reflects the upper bound of the file/locking space */
1139 #ifndef OFFSET_MAX
1140 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1141 #define OFFSET_MAX INT_LIMIT(loff_t)
1142 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1143 #endif
1144
1145 #include <linux/fcntl.h>
1146
1147 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1148
1149 #ifdef CONFIG_FILE_LOCKING
1150 extern int fcntl_getlk(struct file *, struct flock __user *);
1151 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1152 struct flock __user *);
1153
1154 #if BITS_PER_LONG == 32
1155 extern int fcntl_getlk64(struct file *, struct flock64 __user *);
1156 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1157 struct flock64 __user *);
1158 #endif
1159
1160 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1161 extern int fcntl_getlease(struct file *filp);
1162
1163 /* fs/locks.c */
1164 void locks_free_lock(struct file_lock *fl);
1165 extern void locks_init_lock(struct file_lock *);
1166 extern struct file_lock * locks_alloc_lock(void);
1167 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1168 extern void __locks_copy_lock(struct file_lock *, const struct file_lock *);
1169 extern void locks_remove_posix(struct file *, fl_owner_t);
1170 extern void locks_remove_flock(struct file *);
1171 extern void locks_release_private(struct file_lock *);
1172 extern void posix_test_lock(struct file *, struct file_lock *);
1173 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1174 extern int posix_lock_file_wait(struct file *, struct file_lock *);
1175 extern int posix_unblock_lock(struct file *, struct file_lock *);
1176 extern int vfs_test_lock(struct file *, struct file_lock *);
1177 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1178 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1179 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
1180 extern int __break_lease(struct inode *inode, unsigned int flags);
1181 extern void lease_get_mtime(struct inode *, struct timespec *time);
1182 extern int generic_setlease(struct file *, long, struct file_lock **);
1183 extern int vfs_setlease(struct file *, long, struct file_lock **);
1184 extern int lease_modify(struct file_lock **, int);
1185 extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
1186 extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
1187 extern void lock_flocks(void);
1188 extern void unlock_flocks(void);
1189 #else /* !CONFIG_FILE_LOCKING */
1190 static inline int fcntl_getlk(struct file *file, struct flock __user *user)
1191 {
1192 return -EINVAL;
1193 }
1194
1195 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1196 unsigned int cmd, struct flock __user *user)
1197 {
1198 return -EACCES;
1199 }
1200
1201 #if BITS_PER_LONG == 32
1202 static inline int fcntl_getlk64(struct file *file, struct flock64 __user *user)
1203 {
1204 return -EINVAL;
1205 }
1206
1207 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1208 unsigned int cmd, struct flock64 __user *user)
1209 {
1210 return -EACCES;
1211 }
1212 #endif
1213 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1214 {
1215 return 0;
1216 }
1217
1218 static inline int fcntl_getlease(struct file *filp)
1219 {
1220 return 0;
1221 }
1222
1223 static inline void locks_init_lock(struct file_lock *fl)
1224 {
1225 return;
1226 }
1227
1228 static inline void __locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1229 {
1230 return;
1231 }
1232
1233 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1234 {
1235 return;
1236 }
1237
1238 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1239 {
1240 return;
1241 }
1242
1243 static inline void locks_remove_flock(struct file *filp)
1244 {
1245 return;
1246 }
1247
1248 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1249 {
1250 return;
1251 }
1252
1253 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1254 struct file_lock *conflock)
1255 {
1256 return -ENOLCK;
1257 }
1258
1259 static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1260 {
1261 return -ENOLCK;
1262 }
1263
1264 static inline int posix_unblock_lock(struct file *filp,
1265 struct file_lock *waiter)
1266 {
1267 return -ENOENT;
1268 }
1269
1270 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1271 {
1272 return 0;
1273 }
1274
1275 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1276 struct file_lock *fl, struct file_lock *conf)
1277 {
1278 return -ENOLCK;
1279 }
1280
1281 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1282 {
1283 return 0;
1284 }
1285
1286 static inline int flock_lock_file_wait(struct file *filp,
1287 struct file_lock *request)
1288 {
1289 return -ENOLCK;
1290 }
1291
1292 static inline int __break_lease(struct inode *inode, unsigned int mode)
1293 {
1294 return 0;
1295 }
1296
1297 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1298 {
1299 return;
1300 }
1301
1302 static inline int generic_setlease(struct file *filp, long arg,
1303 struct file_lock **flp)
1304 {
1305 return -EINVAL;
1306 }
1307
1308 static inline int vfs_setlease(struct file *filp, long arg,
1309 struct file_lock **lease)
1310 {
1311 return -EINVAL;
1312 }
1313
1314 static inline int lease_modify(struct file_lock **before, int arg)
1315 {
1316 return -EINVAL;
1317 }
1318
1319 static inline int lock_may_read(struct inode *inode, loff_t start,
1320 unsigned long len)
1321 {
1322 return 1;
1323 }
1324
1325 static inline int lock_may_write(struct inode *inode, loff_t start,
1326 unsigned long len)
1327 {
1328 return 1;
1329 }
1330
1331 static inline void lock_flocks(void)
1332 {
1333 }
1334
1335 static inline void unlock_flocks(void)
1336 {
1337 }
1338
1339 #endif /* !CONFIG_FILE_LOCKING */
1340
1341
1342 struct fasync_struct {
1343 spinlock_t fa_lock;
1344 int magic;
1345 int fa_fd;
1346 struct fasync_struct *fa_next; /* singly linked list */
1347 struct file *fa_file;
1348 struct rcu_head fa_rcu;
1349 };
1350
1351 #define FASYNC_MAGIC 0x4601
1352
1353 /* SMP safe fasync helpers: */
1354 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1355 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1356 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1357 extern struct fasync_struct *fasync_alloc(void);
1358 extern void fasync_free(struct fasync_struct *);
1359
1360 /* can be called from interrupts */
1361 extern void kill_fasync(struct fasync_struct **, int, int);
1362
1363 extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1364 extern int f_setown(struct file *filp, unsigned long arg, int force);
1365 extern void f_delown(struct file *filp);
1366 extern pid_t f_getown(struct file *filp);
1367 extern int send_sigurg(struct fown_struct *fown);
1368
1369 /*
1370 * Umount options
1371 */
1372
1373 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1374 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1375 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1376 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1377 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1378
1379 extern struct list_head super_blocks;
1380 extern spinlock_t sb_lock;
1381
1382 struct super_block {
1383 struct list_head s_list; /* Keep this first */
1384 dev_t s_dev; /* search index; _not_ kdev_t */
1385 unsigned char s_dirt;
1386 unsigned char s_blocksize_bits;
1387 unsigned long s_blocksize;
1388 loff_t s_maxbytes; /* Max file size */
1389 struct file_system_type *s_type;
1390 const struct super_operations *s_op;
1391 const struct dquot_operations *dq_op;
1392 const struct quotactl_ops *s_qcop;
1393 const struct export_operations *s_export_op;
1394 unsigned long s_flags;
1395 unsigned long s_magic;
1396 struct dentry *s_root;
1397 struct rw_semaphore s_umount;
1398 struct mutex s_lock;
1399 int s_count;
1400 atomic_t s_active;
1401 #ifdef CONFIG_SECURITY
1402 void *s_security;
1403 #endif
1404 const struct xattr_handler **s_xattr;
1405
1406 struct list_head s_inodes; /* all inodes */
1407 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1408 #ifdef CONFIG_SMP
1409 struct list_head __percpu *s_files;
1410 #else
1411 struct list_head s_files;
1412 #endif
1413 /* s_dentry_lru, s_nr_dentry_unused protected by dcache.c lru locks */
1414 struct list_head s_dentry_lru; /* unused dentry lru */
1415 int s_nr_dentry_unused; /* # of dentry on lru */
1416
1417 /* s_inode_lru_lock protects s_inode_lru and s_nr_inodes_unused */
1418 spinlock_t s_inode_lru_lock ____cacheline_aligned_in_smp;
1419 struct list_head s_inode_lru; /* unused inode lru */
1420 int s_nr_inodes_unused; /* # of inodes on lru */
1421
1422 struct block_device *s_bdev;
1423 struct backing_dev_info *s_bdi;
1424 struct mtd_info *s_mtd;
1425 struct list_head s_instances;
1426 struct quota_info s_dquot; /* Diskquota specific options */
1427
1428 int s_frozen;
1429 wait_queue_head_t s_wait_unfrozen;
1430
1431 char s_id[32]; /* Informational name */
1432 u8 s_uuid[16]; /* UUID */
1433
1434 void *s_fs_info; /* Filesystem private info */
1435 fmode_t s_mode;
1436
1437 /* Granularity of c/m/atime in ns.
1438 Cannot be worse than a second */
1439 u32 s_time_gran;
1440
1441 /*
1442 * The next field is for VFS *only*. No filesystems have any business
1443 * even looking at it. You had been warned.
1444 */
1445 struct mutex s_vfs_rename_mutex; /* Kludge */
1446
1447 /*
1448 * Filesystem subtype. If non-empty the filesystem type field
1449 * in /proc/mounts will be "type.subtype"
1450 */
1451 char *s_subtype;
1452
1453 /*
1454 * Saved mount options for lazy filesystems using
1455 * generic_show_options()
1456 */
1457 char __rcu *s_options;
1458 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1459
1460 /*
1461 * Saved pool identifier for cleancache (-1 means none)
1462 */
1463 int cleancache_poolid;
1464
1465 struct shrinker s_shrink; /* per-sb shrinker handle */
1466 };
1467
1468 /* superblock cache pruning functions */
1469 extern void prune_icache_sb(struct super_block *sb, int nr_to_scan);
1470 extern void prune_dcache_sb(struct super_block *sb, int nr_to_scan);
1471
1472 extern struct timespec current_fs_time(struct super_block *sb);
1473
1474 /*
1475 * Snapshotting support.
1476 */
1477 enum {
1478 SB_UNFROZEN = 0,
1479 SB_FREEZE_WRITE = 1,
1480 SB_FREEZE_TRANS = 2,
1481 };
1482
1483 #define vfs_check_frozen(sb, level) \
1484 wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
1485
1486 /*
1487 * until VFS tracks user namespaces for inodes, just make all files
1488 * belong to init_user_ns
1489 */
1490 extern struct user_namespace init_user_ns;
1491 #define inode_userns(inode) (&init_user_ns)
1492 extern bool inode_owner_or_capable(const struct inode *inode);
1493
1494 /* not quite ready to be deprecated, but... */
1495 extern void lock_super(struct super_block *);
1496 extern void unlock_super(struct super_block *);
1497
1498 /*
1499 * VFS helper functions..
1500 */
1501 extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *);
1502 extern int vfs_mkdir(struct inode *, struct dentry *, int);
1503 extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t);
1504 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1505 extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
1506 extern int vfs_rmdir(struct inode *, struct dentry *);
1507 extern int vfs_unlink(struct inode *, struct dentry *);
1508 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1509
1510 /*
1511 * VFS dentry helper functions.
1512 */
1513 extern void dentry_unhash(struct dentry *dentry);
1514
1515 /*
1516 * VFS file helper functions.
1517 */
1518 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1519 mode_t mode);
1520 /*
1521 * VFS FS_IOC_FIEMAP helper definitions.
1522 */
1523 struct fiemap_extent_info {
1524 unsigned int fi_flags; /* Flags as passed from user */
1525 unsigned int fi_extents_mapped; /* Number of mapped extents */
1526 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1527 struct fiemap_extent __user *fi_extents_start; /* Start of
1528 fiemap_extent array */
1529 };
1530 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1531 u64 phys, u64 len, u32 flags);
1532 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1533
1534 /*
1535 * File types
1536 *
1537 * NOTE! These match bits 12..15 of stat.st_mode
1538 * (ie "(i_mode >> 12) & 15").
1539 */
1540 #define DT_UNKNOWN 0
1541 #define DT_FIFO 1
1542 #define DT_CHR 2
1543 #define DT_DIR 4
1544 #define DT_BLK 6
1545 #define DT_REG 8
1546 #define DT_LNK 10
1547 #define DT_SOCK 12
1548 #define DT_WHT 14
1549
1550 /*
1551 * This is the "filldir" function type, used by readdir() to let
1552 * the kernel specify what kind of dirent layout it wants to have.
1553 * This allows the kernel to read directories into kernel space or
1554 * to have different dirent layouts depending on the binary type.
1555 */
1556 typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
1557 struct block_device_operations;
1558
1559 /* These macros are for out of kernel modules to test that
1560 * the kernel supports the unlocked_ioctl and compat_ioctl
1561 * fields in struct file_operations. */
1562 #define HAVE_COMPAT_IOCTL 1
1563 #define HAVE_UNLOCKED_IOCTL 1
1564
1565 struct file_operations {
1566 struct module *owner;
1567 loff_t (*llseek) (struct file *, loff_t, int);
1568 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1569 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1570 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1571 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1572 int (*readdir) (struct file *, void *, filldir_t);
1573 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1574 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1575 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1576 int (*mmap) (struct file *, struct vm_area_struct *);
1577 int (*open) (struct inode *, struct file *);
1578 int (*flush) (struct file *, fl_owner_t id);
1579 int (*release) (struct inode *, struct file *);
1580 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1581 int (*aio_fsync) (struct kiocb *, int datasync);
1582 int (*fasync) (int, struct file *, int);
1583 int (*lock) (struct file *, int, struct file_lock *);
1584 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1585 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1586 int (*check_flags)(int);
1587 int (*flock) (struct file *, int, struct file_lock *);
1588 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1589 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1590 int (*setlease)(struct file *, long, struct file_lock **);
1591 long (*fallocate)(struct file *file, int mode, loff_t offset,
1592 loff_t len);
1593 };
1594
1595 struct inode_operations {
1596 struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
1597 void * (*follow_link) (struct dentry *, struct nameidata *);
1598 int (*permission) (struct inode *, int);
1599 struct posix_acl * (*get_acl)(struct inode *, int);
1600
1601 int (*readlink) (struct dentry *, char __user *,int);
1602 void (*put_link) (struct dentry *, struct nameidata *, void *);
1603
1604 int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
1605 int (*link) (struct dentry *,struct inode *,struct dentry *);
1606 int (*unlink) (struct inode *,struct dentry *);
1607 int (*symlink) (struct inode *,struct dentry *,const char *);
1608 int (*mkdir) (struct inode *,struct dentry *,int);
1609 int (*rmdir) (struct inode *,struct dentry *);
1610 int (*mknod) (struct inode *,struct dentry *,int,dev_t);
1611 int (*rename) (struct inode *, struct dentry *,
1612 struct inode *, struct dentry *);
1613 void (*truncate) (struct inode *);
1614 int (*setattr) (struct dentry *, struct iattr *);
1615 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1616 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1617 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1618 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1619 int (*removexattr) (struct dentry *, const char *);
1620 void (*truncate_range)(struct inode *, loff_t, loff_t);
1621 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1622 u64 len);
1623 } ____cacheline_aligned;
1624
1625 struct seq_file;
1626
1627 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1628 unsigned long nr_segs, unsigned long fast_segs,
1629 struct iovec *fast_pointer,
1630 struct iovec **ret_pointer);
1631
1632 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1633 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1634 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1635 unsigned long, loff_t *);
1636 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1637 unsigned long, loff_t *);
1638
1639 struct super_operations {
1640 struct inode *(*alloc_inode)(struct super_block *sb);
1641 void (*destroy_inode)(struct inode *);
1642
1643 void (*dirty_inode) (struct inode *, int flags);
1644 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1645 int (*drop_inode) (struct inode *);
1646 void (*evict_inode) (struct inode *);
1647 void (*put_super) (struct super_block *);
1648 void (*write_super) (struct super_block *);
1649 int (*sync_fs)(struct super_block *sb, int wait);
1650 int (*freeze_fs) (struct super_block *);
1651 int (*unfreeze_fs) (struct super_block *);
1652 int (*statfs) (struct dentry *, struct kstatfs *);
1653 int (*remount_fs) (struct super_block *, int *, char *);
1654 void (*umount_begin) (struct super_block *);
1655
1656 int (*show_options)(struct seq_file *, struct vfsmount *);
1657 int (*show_devname)(struct seq_file *, struct vfsmount *);
1658 int (*show_path)(struct seq_file *, struct vfsmount *);
1659 int (*show_stats)(struct seq_file *, struct vfsmount *);
1660 #ifdef CONFIG_QUOTA
1661 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1662 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1663 #endif
1664 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1665 int (*nr_cached_objects)(struct super_block *);
1666 void (*free_cached_objects)(struct super_block *, int);
1667 };
1668
1669 /*
1670 * Inode state bits. Protected by inode->i_lock
1671 *
1672 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1673 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1674 *
1675 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1676 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1677 * various stages of removing an inode.
1678 *
1679 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1680 *
1681 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1682 * fdatasync(). i_atime is the usual cause.
1683 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1684 * these changes separately from I_DIRTY_SYNC so that we
1685 * don't have to write inode on fdatasync() when only
1686 * mtime has changed in it.
1687 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1688 * I_NEW Serves as both a mutex and completion notification.
1689 * New inodes set I_NEW. If two processes both create
1690 * the same inode, one of them will release its inode and
1691 * wait for I_NEW to be released before returning.
1692 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1693 * also cause waiting on I_NEW, without I_NEW actually
1694 * being set. find_inode() uses this to prevent returning
1695 * nearly-dead inodes.
1696 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1697 * is zero. I_FREEING must be set when I_WILL_FREE is
1698 * cleared.
1699 * I_FREEING Set when inode is about to be freed but still has dirty
1700 * pages or buffers attached or the inode itself is still
1701 * dirty.
1702 * I_CLEAR Added by end_writeback(). In this state the inode is clean
1703 * and can be destroyed. Inode keeps I_FREEING.
1704 *
1705 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1706 * prohibited for many purposes. iget() must wait for
1707 * the inode to be completely released, then create it
1708 * anew. Other functions will just ignore such inodes,
1709 * if appropriate. I_NEW is used for waiting.
1710 *
1711 * I_SYNC Synchonized write of dirty inode data. The bits is
1712 * set during data writeback, and cleared with a wakeup
1713 * on the bit address once it is done.
1714 *
1715 * I_REFERENCED Marks the inode as recently references on the LRU list.
1716 *
1717 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1718 *
1719 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1720 */
1721 #define I_DIRTY_SYNC (1 << 0)
1722 #define I_DIRTY_DATASYNC (1 << 1)
1723 #define I_DIRTY_PAGES (1 << 2)
1724 #define __I_NEW 3
1725 #define I_NEW (1 << __I_NEW)
1726 #define I_WILL_FREE (1 << 4)
1727 #define I_FREEING (1 << 5)
1728 #define I_CLEAR (1 << 6)
1729 #define __I_SYNC 7
1730 #define I_SYNC (1 << __I_SYNC)
1731 #define I_REFERENCED (1 << 8)
1732 #define __I_DIO_WAKEUP 9
1733 #define I_DIO_WAKEUP (1 << I_DIO_WAKEUP)
1734
1735 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1736
1737 extern void __mark_inode_dirty(struct inode *, int);
1738 static inline void mark_inode_dirty(struct inode *inode)
1739 {
1740 __mark_inode_dirty(inode, I_DIRTY);
1741 }
1742
1743 static inline void mark_inode_dirty_sync(struct inode *inode)
1744 {
1745 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1746 }
1747
1748 /**
1749 * inc_nlink - directly increment an inode's link count
1750 * @inode: inode
1751 *
1752 * This is a low-level filesystem helper to replace any
1753 * direct filesystem manipulation of i_nlink. Currently,
1754 * it is only here for parity with dec_nlink().
1755 */
1756 static inline void inc_nlink(struct inode *inode)
1757 {
1758 inode->i_nlink++;
1759 }
1760
1761 static inline void inode_inc_link_count(struct inode *inode)
1762 {
1763 inc_nlink(inode);
1764 mark_inode_dirty(inode);
1765 }
1766
1767 /**
1768 * drop_nlink - directly drop an inode's link count
1769 * @inode: inode
1770 *
1771 * This is a low-level filesystem helper to replace any
1772 * direct filesystem manipulation of i_nlink. In cases
1773 * where we are attempting to track writes to the
1774 * filesystem, a decrement to zero means an imminent
1775 * write when the file is truncated and actually unlinked
1776 * on the filesystem.
1777 */
1778 static inline void drop_nlink(struct inode *inode)
1779 {
1780 inode->i_nlink--;
1781 }
1782
1783 /**
1784 * clear_nlink - directly zero an inode's link count
1785 * @inode: inode
1786 *
1787 * This is a low-level filesystem helper to replace any
1788 * direct filesystem manipulation of i_nlink. See
1789 * drop_nlink() for why we care about i_nlink hitting zero.
1790 */
1791 static inline void clear_nlink(struct inode *inode)
1792 {
1793 inode->i_nlink = 0;
1794 }
1795
1796 static inline void inode_dec_link_count(struct inode *inode)
1797 {
1798 drop_nlink(inode);
1799 mark_inode_dirty(inode);
1800 }
1801
1802 /**
1803 * inode_inc_iversion - increments i_version
1804 * @inode: inode that need to be updated
1805 *
1806 * Every time the inode is modified, the i_version field will be incremented.
1807 * The filesystem has to be mounted with i_version flag
1808 */
1809
1810 static inline void inode_inc_iversion(struct inode *inode)
1811 {
1812 spin_lock(&inode->i_lock);
1813 inode->i_version++;
1814 spin_unlock(&inode->i_lock);
1815 }
1816
1817 extern void touch_atime(struct vfsmount *mnt, struct dentry *dentry);
1818 static inline void file_accessed(struct file *file)
1819 {
1820 if (!(file->f_flags & O_NOATIME))
1821 touch_atime(file->f_path.mnt, file->f_path.dentry);
1822 }
1823
1824 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1825 int sync_inode_metadata(struct inode *inode, int wait);
1826
1827 struct file_system_type {
1828 const char *name;
1829 int fs_flags;
1830 struct dentry *(*mount) (struct file_system_type *, int,
1831 const char *, void *);
1832 void (*kill_sb) (struct super_block *);
1833 struct module *owner;
1834 struct file_system_type * next;
1835 struct list_head fs_supers;
1836
1837 struct lock_class_key s_lock_key;
1838 struct lock_class_key s_umount_key;
1839 struct lock_class_key s_vfs_rename_key;
1840
1841 struct lock_class_key i_lock_key;
1842 struct lock_class_key i_mutex_key;
1843 struct lock_class_key i_mutex_dir_key;
1844 };
1845
1846 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1847 void *data, int (*fill_super)(struct super_block *, void *, int));
1848 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1849 int flags, const char *dev_name, void *data,
1850 int (*fill_super)(struct super_block *, void *, int));
1851 extern struct dentry *mount_single(struct file_system_type *fs_type,
1852 int flags, void *data,
1853 int (*fill_super)(struct super_block *, void *, int));
1854 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1855 int flags, void *data,
1856 int (*fill_super)(struct super_block *, void *, int));
1857 void generic_shutdown_super(struct super_block *sb);
1858 void kill_block_super(struct super_block *sb);
1859 void kill_anon_super(struct super_block *sb);
1860 void kill_litter_super(struct super_block *sb);
1861 void deactivate_super(struct super_block *sb);
1862 void deactivate_locked_super(struct super_block *sb);
1863 int set_anon_super(struct super_block *s, void *data);
1864 int get_anon_bdev(dev_t *);
1865 void free_anon_bdev(dev_t);
1866 struct super_block *sget(struct file_system_type *type,
1867 int (*test)(struct super_block *,void *),
1868 int (*set)(struct super_block *,void *),
1869 void *data);
1870 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
1871 const struct super_operations *ops,
1872 const struct dentry_operations *dops,
1873 unsigned long);
1874
1875 static inline void sb_mark_dirty(struct super_block *sb)
1876 {
1877 sb->s_dirt = 1;
1878 }
1879 static inline void sb_mark_clean(struct super_block *sb)
1880 {
1881 sb->s_dirt = 0;
1882 }
1883 static inline int sb_is_dirty(struct super_block *sb)
1884 {
1885 return sb->s_dirt;
1886 }
1887
1888 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1889 #define fops_get(fops) \
1890 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1891 #define fops_put(fops) \
1892 do { if (fops) module_put((fops)->owner); } while(0)
1893
1894 extern int register_filesystem(struct file_system_type *);
1895 extern int unregister_filesystem(struct file_system_type *);
1896 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1897 #define kern_mount(type) kern_mount_data(type, NULL)
1898 extern void kern_unmount(struct vfsmount *mnt);
1899 extern int may_umount_tree(struct vfsmount *);
1900 extern int may_umount(struct vfsmount *);
1901 extern long do_mount(char *, char *, char *, unsigned long, void *);
1902 extern struct vfsmount *collect_mounts(struct path *);
1903 extern void drop_collected_mounts(struct vfsmount *);
1904 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1905 struct vfsmount *);
1906 extern int vfs_statfs(struct path *, struct kstatfs *);
1907 extern int user_statfs(const char __user *, struct kstatfs *);
1908 extern int fd_statfs(int, struct kstatfs *);
1909 extern int statfs_by_dentry(struct dentry *, struct kstatfs *);
1910 extern int freeze_super(struct super_block *super);
1911 extern int thaw_super(struct super_block *super);
1912
1913 extern int current_umask(void);
1914
1915 /* /sys/fs */
1916 extern struct kobject *fs_kobj;
1917
1918 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
1919 extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1920
1921 #define FLOCK_VERIFY_READ 1
1922 #define FLOCK_VERIFY_WRITE 2
1923
1924 #ifdef CONFIG_FILE_LOCKING
1925 extern int locks_mandatory_locked(struct inode *);
1926 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1927
1928 /*
1929 * Candidates for mandatory locking have the setgid bit set
1930 * but no group execute bit - an otherwise meaningless combination.
1931 */
1932
1933 static inline int __mandatory_lock(struct inode *ino)
1934 {
1935 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1936 }
1937
1938 /*
1939 * ... and these candidates should be on MS_MANDLOCK mounted fs,
1940 * otherwise these will be advisory locks
1941 */
1942
1943 static inline int mandatory_lock(struct inode *ino)
1944 {
1945 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1946 }
1947
1948 static inline int locks_verify_locked(struct inode *inode)
1949 {
1950 if (mandatory_lock(inode))
1951 return locks_mandatory_locked(inode);
1952 return 0;
1953 }
1954
1955 static inline int locks_verify_truncate(struct inode *inode,
1956 struct file *filp,
1957 loff_t size)
1958 {
1959 if (inode->i_flock && mandatory_lock(inode))
1960 return locks_mandatory_area(
1961 FLOCK_VERIFY_WRITE, inode, filp,
1962 size < inode->i_size ? size : inode->i_size,
1963 (size < inode->i_size ? inode->i_size - size
1964 : size - inode->i_size)
1965 );
1966 return 0;
1967 }
1968
1969 static inline int break_lease(struct inode *inode, unsigned int mode)
1970 {
1971 if (inode->i_flock)
1972 return __break_lease(inode, mode);
1973 return 0;
1974 }
1975 #else /* !CONFIG_FILE_LOCKING */
1976 static inline int locks_mandatory_locked(struct inode *inode)
1977 {
1978 return 0;
1979 }
1980
1981 static inline int locks_mandatory_area(int rw, struct inode *inode,
1982 struct file *filp, loff_t offset,
1983 size_t count)
1984 {
1985 return 0;
1986 }
1987
1988 static inline int __mandatory_lock(struct inode *inode)
1989 {
1990 return 0;
1991 }
1992
1993 static inline int mandatory_lock(struct inode *inode)
1994 {
1995 return 0;
1996 }
1997
1998 static inline int locks_verify_locked(struct inode *inode)
1999 {
2000 return 0;
2001 }
2002
2003 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2004 size_t size)
2005 {
2006 return 0;
2007 }
2008
2009 static inline int break_lease(struct inode *inode, unsigned int mode)
2010 {
2011 return 0;
2012 }
2013
2014 #endif /* CONFIG_FILE_LOCKING */
2015
2016 /* fs/open.c */
2017
2018 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2019 struct file *filp);
2020 extern int do_fallocate(struct file *file, int mode, loff_t offset,
2021 loff_t len);
2022 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2023 int mode);
2024 extern struct file *filp_open(const char *, int, int);
2025 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2026 const char *, int);
2027 extern struct file * dentry_open(struct dentry *, struct vfsmount *, int,
2028 const struct cred *);
2029 extern int filp_close(struct file *, fl_owner_t id);
2030 extern char * getname(const char __user *);
2031
2032 /* fs/ioctl.c */
2033
2034 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2035
2036 /* fs/dcache.c */
2037 extern void __init vfs_caches_init_early(void);
2038 extern void __init vfs_caches_init(unsigned long);
2039
2040 extern struct kmem_cache *names_cachep;
2041
2042 #define __getname_gfp(gfp) kmem_cache_alloc(names_cachep, (gfp))
2043 #define __getname() __getname_gfp(GFP_KERNEL)
2044 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2045 #ifndef CONFIG_AUDITSYSCALL
2046 #define putname(name) __putname(name)
2047 #else
2048 extern void putname(const char *name);
2049 #endif
2050
2051 #ifdef CONFIG_BLOCK
2052 extern int register_blkdev(unsigned int, const char *);
2053 extern void unregister_blkdev(unsigned int, const char *);
2054 extern struct block_device *bdget(dev_t);
2055 extern struct block_device *bdgrab(struct block_device *bdev);
2056 extern void bd_set_size(struct block_device *, loff_t size);
2057 extern void bd_forget(struct inode *inode);
2058 extern void bdput(struct block_device *);
2059 extern void invalidate_bdev(struct block_device *);
2060 extern int sync_blockdev(struct block_device *bdev);
2061 extern struct super_block *freeze_bdev(struct block_device *);
2062 extern void emergency_thaw_all(void);
2063 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2064 extern int fsync_bdev(struct block_device *);
2065 #else
2066 static inline void bd_forget(struct inode *inode) {}
2067 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2068 static inline void invalidate_bdev(struct block_device *bdev) {}
2069
2070 static inline struct super_block *freeze_bdev(struct block_device *sb)
2071 {
2072 return NULL;
2073 }
2074
2075 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2076 {
2077 return 0;
2078 }
2079 #endif
2080 extern int sync_filesystem(struct super_block *);
2081 extern const struct file_operations def_blk_fops;
2082 extern const struct file_operations def_chr_fops;
2083 extern const struct file_operations bad_sock_fops;
2084 extern const struct file_operations def_fifo_fops;
2085 #ifdef CONFIG_BLOCK
2086 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2087 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2088 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2089 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2090 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2091 void *holder);
2092 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2093 void *holder);
2094 extern int blkdev_put(struct block_device *bdev, fmode_t mode);
2095 #ifdef CONFIG_SYSFS
2096 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2097 extern void bd_unlink_disk_holder(struct block_device *bdev,
2098 struct gendisk *disk);
2099 #else
2100 static inline int bd_link_disk_holder(struct block_device *bdev,
2101 struct gendisk *disk)
2102 {
2103 return 0;
2104 }
2105 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2106 struct gendisk *disk)
2107 {
2108 }
2109 #endif
2110 #endif
2111
2112 /* fs/char_dev.c */
2113 #define CHRDEV_MAJOR_HASH_SIZE 255
2114 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2115 extern int register_chrdev_region(dev_t, unsigned, const char *);
2116 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2117 unsigned int count, const char *name,
2118 const struct file_operations *fops);
2119 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2120 unsigned int count, const char *name);
2121 extern void unregister_chrdev_region(dev_t, unsigned);
2122 extern void chrdev_show(struct seq_file *,off_t);
2123
2124 static inline int register_chrdev(unsigned int major, const char *name,
2125 const struct file_operations *fops)
2126 {
2127 return __register_chrdev(major, 0, 256, name, fops);
2128 }
2129
2130 static inline void unregister_chrdev(unsigned int major, const char *name)
2131 {
2132 __unregister_chrdev(major, 0, 256, name);
2133 }
2134
2135 /* fs/block_dev.c */
2136 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2137 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2138
2139 #ifdef CONFIG_BLOCK
2140 #define BLKDEV_MAJOR_HASH_SIZE 255
2141 extern const char *__bdevname(dev_t, char *buffer);
2142 extern const char *bdevname(struct block_device *bdev, char *buffer);
2143 extern struct block_device *lookup_bdev(const char *);
2144 extern void blkdev_show(struct seq_file *,off_t);
2145
2146 #else
2147 #define BLKDEV_MAJOR_HASH_SIZE 0
2148 #endif
2149
2150 extern void init_special_inode(struct inode *, umode_t, dev_t);
2151
2152 /* Invalid inode operations -- fs/bad_inode.c */
2153 extern void make_bad_inode(struct inode *);
2154 extern int is_bad_inode(struct inode *);
2155
2156 extern const struct file_operations read_pipefifo_fops;
2157 extern const struct file_operations write_pipefifo_fops;
2158 extern const struct file_operations rdwr_pipefifo_fops;
2159
2160 extern int fs_may_remount_ro(struct super_block *);
2161
2162 #ifdef CONFIG_BLOCK
2163 /*
2164 * return READ, READA, or WRITE
2165 */
2166 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2167
2168 /*
2169 * return data direction, READ or WRITE
2170 */
2171 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2172
2173 extern void check_disk_size_change(struct gendisk *disk,
2174 struct block_device *bdev);
2175 extern int revalidate_disk(struct gendisk *);
2176 extern int check_disk_change(struct block_device *);
2177 extern int __invalidate_device(struct block_device *, bool);
2178 extern int invalidate_partition(struct gendisk *, int);
2179 #endif
2180 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2181 pgoff_t start, pgoff_t end);
2182
2183 static inline void invalidate_remote_inode(struct inode *inode)
2184 {
2185 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2186 S_ISLNK(inode->i_mode))
2187 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2188 }
2189 extern int invalidate_inode_pages2(struct address_space *mapping);
2190 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2191 pgoff_t start, pgoff_t end);
2192 extern int write_inode_now(struct inode *, int);
2193 extern int filemap_fdatawrite(struct address_space *);
2194 extern int filemap_flush(struct address_space *);
2195 extern int filemap_fdatawait(struct address_space *);
2196 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2197 loff_t lend);
2198 extern int filemap_write_and_wait(struct address_space *mapping);
2199 extern int filemap_write_and_wait_range(struct address_space *mapping,
2200 loff_t lstart, loff_t lend);
2201 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2202 loff_t start, loff_t end, int sync_mode);
2203 extern int filemap_fdatawrite_range(struct address_space *mapping,
2204 loff_t start, loff_t end);
2205
2206 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2207 int datasync);
2208 extern int vfs_fsync(struct file *file, int datasync);
2209 extern int generic_write_sync(struct file *file, loff_t pos, loff_t count);
2210 extern void sync_supers(void);
2211 extern void emergency_sync(void);
2212 extern void emergency_remount(void);
2213 #ifdef CONFIG_BLOCK
2214 extern sector_t bmap(struct inode *, sector_t);
2215 #endif
2216 extern int notify_change(struct dentry *, struct iattr *);
2217 extern int inode_permission(struct inode *, int);
2218 extern int generic_permission(struct inode *, int);
2219
2220 static inline bool execute_ok(struct inode *inode)
2221 {
2222 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2223 }
2224
2225 /*
2226 * get_write_access() gets write permission for a file.
2227 * put_write_access() releases this write permission.
2228 * This is used for regular files.
2229 * We cannot support write (and maybe mmap read-write shared) accesses and
2230 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2231 * can have the following values:
2232 * 0: no writers, no VM_DENYWRITE mappings
2233 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2234 * > 0: (i_writecount) users are writing to the file.
2235 *
2236 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2237 * except for the cases where we don't hold i_writecount yet. Then we need to
2238 * use {get,deny}_write_access() - these functions check the sign and refuse
2239 * to do the change if sign is wrong.
2240 */
2241 static inline int get_write_access(struct inode *inode)
2242 {
2243 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2244 }
2245 static inline int deny_write_access(struct file *file)
2246 {
2247 struct inode *inode = file->f_path.dentry->d_inode;
2248 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2249 }
2250 static inline void put_write_access(struct inode * inode)
2251 {
2252 atomic_dec(&inode->i_writecount);
2253 }
2254 static inline void allow_write_access(struct file *file)
2255 {
2256 if (file)
2257 atomic_inc(&file->f_path.dentry->d_inode->i_writecount);
2258 }
2259 #ifdef CONFIG_IMA
2260 static inline void i_readcount_dec(struct inode *inode)
2261 {
2262 BUG_ON(!atomic_read(&inode->i_readcount));
2263 atomic_dec(&inode->i_readcount);
2264 }
2265 static inline void i_readcount_inc(struct inode *inode)
2266 {
2267 atomic_inc(&inode->i_readcount);
2268 }
2269 #else
2270 static inline void i_readcount_dec(struct inode *inode)
2271 {
2272 return;
2273 }
2274 static inline void i_readcount_inc(struct inode *inode)
2275 {
2276 return;
2277 }
2278 #endif
2279 extern int do_pipe_flags(int *, int);
2280 extern struct file *create_read_pipe(struct file *f, int flags);
2281 extern struct file *create_write_pipe(int flags);
2282 extern void free_write_pipe(struct file *);
2283
2284 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2285 extern struct file * open_exec(const char *);
2286
2287 /* fs/dcache.c -- generic fs support functions */
2288 extern int is_subdir(struct dentry *, struct dentry *);
2289 extern int path_is_under(struct path *, struct path *);
2290 extern ino_t find_inode_number(struct dentry *, struct qstr *);
2291
2292 #include <linux/err.h>
2293
2294 /* needed for stackable file system support */
2295 extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
2296
2297 extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
2298
2299 extern int inode_init_always(struct super_block *, struct inode *);
2300 extern void inode_init_once(struct inode *);
2301 extern void address_space_init_once(struct address_space *mapping);
2302 extern void ihold(struct inode * inode);
2303 extern void iput(struct inode *);
2304 extern struct inode * igrab(struct inode *);
2305 extern ino_t iunique(struct super_block *, ino_t);
2306 extern int inode_needs_sync(struct inode *inode);
2307 extern int generic_delete_inode(struct inode *inode);
2308 extern int generic_drop_inode(struct inode *inode);
2309
2310 extern struct inode *ilookup5_nowait(struct super_block *sb,
2311 unsigned long hashval, int (*test)(struct inode *, void *),
2312 void *data);
2313 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2314 int (*test)(struct inode *, void *), void *data);
2315 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2316
2317 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2318 extern struct inode * iget_locked(struct super_block *, unsigned long);
2319 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2320 extern int insert_inode_locked(struct inode *);
2321 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2322 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2323 #else
2324 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2325 #endif
2326 extern void unlock_new_inode(struct inode *);
2327 extern unsigned int get_next_ino(void);
2328
2329 extern void __iget(struct inode * inode);
2330 extern void iget_failed(struct inode *);
2331 extern void end_writeback(struct inode *);
2332 extern void __destroy_inode(struct inode *);
2333 extern struct inode *new_inode_pseudo(struct super_block *sb);
2334 extern struct inode *new_inode(struct super_block *sb);
2335 extern void free_inode_nonrcu(struct inode *inode);
2336 extern int should_remove_suid(struct dentry *);
2337 extern int file_remove_suid(struct file *);
2338
2339 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2340 static inline void insert_inode_hash(struct inode *inode)
2341 {
2342 __insert_inode_hash(inode, inode->i_ino);
2343 }
2344
2345 extern void __remove_inode_hash(struct inode *);
2346 static inline void remove_inode_hash(struct inode *inode)
2347 {
2348 if (!inode_unhashed(inode))
2349 __remove_inode_hash(inode);
2350 }
2351
2352 extern void inode_sb_list_add(struct inode *inode);
2353
2354 #ifdef CONFIG_BLOCK
2355 extern void submit_bio(int, struct bio *);
2356 extern int bdev_read_only(struct block_device *);
2357 #endif
2358 extern int set_blocksize(struct block_device *, int);
2359 extern int sb_set_blocksize(struct super_block *, int);
2360 extern int sb_min_blocksize(struct super_block *, int);
2361
2362 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2363 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2364 extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
2365 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
2366 extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2367 extern ssize_t __generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long,
2368 loff_t *);
2369 extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2370 extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
2371 unsigned long *, loff_t, loff_t *, size_t, size_t);
2372 extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
2373 unsigned long, loff_t, loff_t *, size_t, ssize_t);
2374 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2375 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2376 extern int generic_segment_checks(const struct iovec *iov,
2377 unsigned long *nr_segs, size_t *count, int access_flags);
2378
2379 /* fs/block_dev.c */
2380 extern ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov,
2381 unsigned long nr_segs, loff_t pos);
2382 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2383 int datasync);
2384
2385 /* fs/splice.c */
2386 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2387 struct pipe_inode_info *, size_t, unsigned int);
2388 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2389 struct pipe_inode_info *, size_t, unsigned int);
2390 extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
2391 struct file *, loff_t *, size_t, unsigned int);
2392 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2393 struct file *out, loff_t *, size_t len, unsigned int flags);
2394 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2395 size_t len, unsigned int flags);
2396
2397 extern void
2398 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2399 extern loff_t noop_llseek(struct file *file, loff_t offset, int origin);
2400 extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
2401 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
2402 extern loff_t generic_file_llseek_unlocked(struct file *file, loff_t offset,
2403 int origin);
2404 extern int generic_file_open(struct inode * inode, struct file * filp);
2405 extern int nonseekable_open(struct inode * inode, struct file * filp);
2406
2407 #ifdef CONFIG_FS_XIP
2408 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2409 loff_t *ppos);
2410 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
2411 extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2412 size_t len, loff_t *ppos);
2413 extern int xip_truncate_page(struct address_space *mapping, loff_t from);
2414 #else
2415 static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2416 {
2417 return 0;
2418 }
2419 #endif
2420
2421 #ifdef CONFIG_BLOCK
2422 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2423 loff_t file_offset);
2424
2425 enum {
2426 /* need locking between buffered and direct access */
2427 DIO_LOCKING = 0x01,
2428
2429 /* filesystem does not support filling holes */
2430 DIO_SKIP_HOLES = 0x02,
2431 };
2432
2433 void dio_end_io(struct bio *bio, int error);
2434 void inode_dio_wait(struct inode *inode);
2435 void inode_dio_done(struct inode *inode);
2436
2437 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2438 struct block_device *bdev, const struct iovec *iov, loff_t offset,
2439 unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
2440 dio_submit_t submit_io, int flags);
2441
2442 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
2443 struct inode *inode, const struct iovec *iov, loff_t offset,
2444 unsigned long nr_segs, get_block_t get_block)
2445 {
2446 return __blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov,
2447 offset, nr_segs, get_block, NULL, NULL,
2448 DIO_LOCKING | DIO_SKIP_HOLES);
2449 }
2450 #else
2451 static inline void inode_dio_wait(struct inode *inode)
2452 {
2453 }
2454 #endif
2455
2456 extern const struct file_operations generic_ro_fops;
2457
2458 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2459
2460 extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
2461 extern int vfs_follow_link(struct nameidata *, const char *);
2462 extern int page_readlink(struct dentry *, char __user *, int);
2463 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2464 extern void page_put_link(struct dentry *, struct nameidata *, void *);
2465 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2466 int nofs);
2467 extern int page_symlink(struct inode *inode, const char *symname, int len);
2468 extern const struct inode_operations page_symlink_inode_operations;
2469 extern int generic_readlink(struct dentry *, char __user *, int);
2470 extern void generic_fillattr(struct inode *, struct kstat *);
2471 extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2472 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2473 void inode_add_bytes(struct inode *inode, loff_t bytes);
2474 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2475 loff_t inode_get_bytes(struct inode *inode);
2476 void inode_set_bytes(struct inode *inode, loff_t bytes);
2477
2478 extern int vfs_readdir(struct file *, filldir_t, void *);
2479
2480 extern int vfs_stat(const char __user *, struct kstat *);
2481 extern int vfs_lstat(const char __user *, struct kstat *);
2482 extern int vfs_fstat(unsigned int, struct kstat *);
2483 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2484
2485 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2486 unsigned long arg);
2487 extern int __generic_block_fiemap(struct inode *inode,
2488 struct fiemap_extent_info *fieinfo,
2489 loff_t start, loff_t len,
2490 get_block_t *get_block);
2491 extern int generic_block_fiemap(struct inode *inode,
2492 struct fiemap_extent_info *fieinfo, u64 start,
2493 u64 len, get_block_t *get_block);
2494
2495 extern void get_filesystem(struct file_system_type *fs);
2496 extern void put_filesystem(struct file_system_type *fs);
2497 extern struct file_system_type *get_fs_type(const char *name);
2498 extern struct super_block *get_super(struct block_device *);
2499 extern struct super_block *get_active_super(struct block_device *bdev);
2500 extern struct super_block *user_get_super(dev_t);
2501 extern void drop_super(struct super_block *sb);
2502 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2503 extern void iterate_supers_type(struct file_system_type *,
2504 void (*)(struct super_block *, void *), void *);
2505
2506 extern int dcache_dir_open(struct inode *, struct file *);
2507 extern int dcache_dir_close(struct inode *, struct file *);
2508 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2509 extern int dcache_readdir(struct file *, void *, filldir_t);
2510 extern int simple_setattr(struct dentry *, struct iattr *);
2511 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2512 extern int simple_statfs(struct dentry *, struct kstatfs *);
2513 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2514 extern int simple_unlink(struct inode *, struct dentry *);
2515 extern int simple_rmdir(struct inode *, struct dentry *);
2516 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2517 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2518 extern int simple_empty(struct dentry *);
2519 extern int simple_readpage(struct file *file, struct page *page);
2520 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2521 loff_t pos, unsigned len, unsigned flags,
2522 struct page **pagep, void **fsdata);
2523 extern int simple_write_end(struct file *file, struct address_space *mapping,
2524 loff_t pos, unsigned len, unsigned copied,
2525 struct page *page, void *fsdata);
2526
2527 extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
2528 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2529 extern const struct file_operations simple_dir_operations;
2530 extern const struct inode_operations simple_dir_inode_operations;
2531 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2532 struct dentry *d_alloc_name(struct dentry *, const char *);
2533 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2534 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2535 extern void simple_release_fs(struct vfsmount **mount, int *count);
2536
2537 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2538 loff_t *ppos, const void *from, size_t available);
2539 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2540 const void __user *from, size_t count);
2541
2542 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2543
2544 extern int generic_check_addressable(unsigned, u64);
2545
2546 #ifdef CONFIG_MIGRATION
2547 extern int buffer_migrate_page(struct address_space *,
2548 struct page *, struct page *);
2549 #else
2550 #define buffer_migrate_page NULL
2551 #endif
2552
2553 extern int inode_change_ok(const struct inode *, struct iattr *);
2554 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2555 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2556
2557 extern void file_update_time(struct file *file);
2558
2559 extern int generic_show_options(struct seq_file *m, struct vfsmount *mnt);
2560 extern void save_mount_options(struct super_block *sb, char *options);
2561 extern void replace_mount_options(struct super_block *sb, char *options);
2562
2563 static inline ino_t parent_ino(struct dentry *dentry)
2564 {
2565 ino_t res;
2566
2567 /*
2568 * Don't strictly need d_lock here? If the parent ino could change
2569 * then surely we'd have a deeper race in the caller?
2570 */
2571 spin_lock(&dentry->d_lock);
2572 res = dentry->d_parent->d_inode->i_ino;
2573 spin_unlock(&dentry->d_lock);
2574 return res;
2575 }
2576
2577 /* Transaction based IO helpers */
2578
2579 /*
2580 * An argresp is stored in an allocated page and holds the
2581 * size of the argument or response, along with its content
2582 */
2583 struct simple_transaction_argresp {
2584 ssize_t size;
2585 char data[0];
2586 };
2587
2588 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2589
2590 char *simple_transaction_get(struct file *file, const char __user *buf,
2591 size_t size);
2592 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2593 size_t size, loff_t *pos);
2594 int simple_transaction_release(struct inode *inode, struct file *file);
2595
2596 void simple_transaction_set(struct file *file, size_t n);
2597
2598 /*
2599 * simple attribute files
2600 *
2601 * These attributes behave similar to those in sysfs:
2602 *
2603 * Writing to an attribute immediately sets a value, an open file can be
2604 * written to multiple times.
2605 *
2606 * Reading from an attribute creates a buffer from the value that might get
2607 * read with multiple read calls. When the attribute has been read
2608 * completely, no further read calls are possible until the file is opened
2609 * again.
2610 *
2611 * All attributes contain a text representation of a numeric value
2612 * that are accessed with the get() and set() functions.
2613 */
2614 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2615 static int __fops ## _open(struct inode *inode, struct file *file) \
2616 { \
2617 __simple_attr_check_format(__fmt, 0ull); \
2618 return simple_attr_open(inode, file, __get, __set, __fmt); \
2619 } \
2620 static const struct file_operations __fops = { \
2621 .owner = THIS_MODULE, \
2622 .open = __fops ## _open, \
2623 .release = simple_attr_release, \
2624 .read = simple_attr_read, \
2625 .write = simple_attr_write, \
2626 .llseek = generic_file_llseek, \
2627 };
2628
2629 static inline void __attribute__((format(printf, 1, 2)))
2630 __simple_attr_check_format(const char *fmt, ...)
2631 {
2632 /* don't do anything, just let the compiler check the arguments; */
2633 }
2634
2635 int simple_attr_open(struct inode *inode, struct file *file,
2636 int (*get)(void *, u64 *), int (*set)(void *, u64),
2637 const char *fmt);
2638 int simple_attr_release(struct inode *inode, struct file *file);
2639 ssize_t simple_attr_read(struct file *file, char __user *buf,
2640 size_t len, loff_t *ppos);
2641 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2642 size_t len, loff_t *ppos);
2643
2644 struct ctl_table;
2645 int proc_nr_files(struct ctl_table *table, int write,
2646 void __user *buffer, size_t *lenp, loff_t *ppos);
2647 int proc_nr_dentry(struct ctl_table *table, int write,
2648 void __user *buffer, size_t *lenp, loff_t *ppos);
2649 int proc_nr_inodes(struct ctl_table *table, int write,
2650 void __user *buffer, size_t *lenp, loff_t *ppos);
2651 int __init get_filesystem_list(char *buf);
2652
2653 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
2654 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
2655
2656 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2657 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2658 (flag & __FMODE_NONOTIFY)))
2659
2660 static inline int is_sxid(mode_t mode)
2661 {
2662 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
2663 }
2664
2665 static inline void inode_has_no_xattr(struct inode *inode)
2666 {
2667 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
2668 inode->i_flags |= S_NOSEC;
2669 }
2670
2671 #endif /* __KERNEL__ */
2672 #endif /* _LINUX_FS_H */