Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt...
[deliverable/linux.git] / include / linux / fs.h
1 #ifndef _LINUX_FS_H
2 #define _LINUX_FS_H
3
4 /*
5 * This file has definitions for some important file table
6 * structures etc.
7 */
8
9 #include <linux/limits.h>
10 #include <linux/ioctl.h>
11 #include <linux/blk_types.h>
12 #include <linux/types.h>
13
14 /*
15 * It's silly to have NR_OPEN bigger than NR_FILE, but you can change
16 * the file limit at runtime and only root can increase the per-process
17 * nr_file rlimit, so it's safe to set up a ridiculously high absolute
18 * upper limit on files-per-process.
19 *
20 * Some programs (notably those using select()) may have to be
21 * recompiled to take full advantage of the new limits..
22 */
23
24 /* Fixed constants first: */
25 #undef NR_OPEN
26 #define INR_OPEN_CUR 1024 /* Initial setting for nfile rlimits */
27 #define INR_OPEN_MAX 4096 /* Hard limit for nfile rlimits */
28
29 #define BLOCK_SIZE_BITS 10
30 #define BLOCK_SIZE (1<<BLOCK_SIZE_BITS)
31
32 #define SEEK_SET 0 /* seek relative to beginning of file */
33 #define SEEK_CUR 1 /* seek relative to current file position */
34 #define SEEK_END 2 /* seek relative to end of file */
35 #define SEEK_DATA 3 /* seek to the next data */
36 #define SEEK_HOLE 4 /* seek to the next hole */
37 #define SEEK_MAX SEEK_HOLE
38
39 struct fstrim_range {
40 __u64 start;
41 __u64 len;
42 __u64 minlen;
43 };
44
45 /* And dynamically-tunable limits and defaults: */
46 struct files_stat_struct {
47 unsigned long nr_files; /* read only */
48 unsigned long nr_free_files; /* read only */
49 unsigned long max_files; /* tunable */
50 };
51
52 struct inodes_stat_t {
53 int nr_inodes;
54 int nr_unused;
55 int dummy[5]; /* padding for sysctl ABI compatibility */
56 };
57
58
59 #define NR_FILE 8192 /* this can well be larger on a larger system */
60
61 #define MAY_EXEC 0x00000001
62 #define MAY_WRITE 0x00000002
63 #define MAY_READ 0x00000004
64 #define MAY_APPEND 0x00000008
65 #define MAY_ACCESS 0x00000010
66 #define MAY_OPEN 0x00000020
67 #define MAY_CHDIR 0x00000040
68 /* called from RCU mode, don't block */
69 #define MAY_NOT_BLOCK 0x00000080
70
71 /*
72 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
73 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
74 */
75
76 /* file is open for reading */
77 #define FMODE_READ ((__force fmode_t)0x1)
78 /* file is open for writing */
79 #define FMODE_WRITE ((__force fmode_t)0x2)
80 /* file is seekable */
81 #define FMODE_LSEEK ((__force fmode_t)0x4)
82 /* file can be accessed using pread */
83 #define FMODE_PREAD ((__force fmode_t)0x8)
84 /* file can be accessed using pwrite */
85 #define FMODE_PWRITE ((__force fmode_t)0x10)
86 /* File is opened for execution with sys_execve / sys_uselib */
87 #define FMODE_EXEC ((__force fmode_t)0x20)
88 /* File is opened with O_NDELAY (only set for block devices) */
89 #define FMODE_NDELAY ((__force fmode_t)0x40)
90 /* File is opened with O_EXCL (only set for block devices) */
91 #define FMODE_EXCL ((__force fmode_t)0x80)
92 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
93 (specialy hack for floppy.c) */
94 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
95
96 /*
97 * Don't update ctime and mtime.
98 *
99 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
100 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
101 */
102 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
103
104 /* Expect random access pattern */
105 #define FMODE_RANDOM ((__force fmode_t)0x1000)
106
107 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
108 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
109
110 /* File is opened with O_PATH; almost nothing can be done with it */
111 #define FMODE_PATH ((__force fmode_t)0x4000)
112
113 /* File was opened by fanotify and shouldn't generate fanotify events */
114 #define FMODE_NONOTIFY ((__force fmode_t)0x1000000)
115
116 /*
117 * The below are the various read and write types that we support. Some of
118 * them include behavioral modifiers that send information down to the
119 * block layer and IO scheduler. Terminology:
120 *
121 * The block layer uses device plugging to defer IO a little bit, in
122 * the hope that we will see more IO very shortly. This increases
123 * coalescing of adjacent IO and thus reduces the number of IOs we
124 * have to send to the device. It also allows for better queuing,
125 * if the IO isn't mergeable. If the caller is going to be waiting
126 * for the IO, then he must ensure that the device is unplugged so
127 * that the IO is dispatched to the driver.
128 *
129 * All IO is handled async in Linux. This is fine for background
130 * writes, but for reads or writes that someone waits for completion
131 * on, we want to notify the block layer and IO scheduler so that they
132 * know about it. That allows them to make better scheduling
133 * decisions. So when the below references 'sync' and 'async', it
134 * is referencing this priority hint.
135 *
136 * With that in mind, the available types are:
137 *
138 * READ A normal read operation. Device will be plugged.
139 * READ_SYNC A synchronous read. Device is not plugged, caller can
140 * immediately wait on this read without caring about
141 * unplugging.
142 * READA Used for read-ahead operations. Lower priority, and the
143 * block layer could (in theory) choose to ignore this
144 * request if it runs into resource problems.
145 * WRITE A normal async write. Device will be plugged.
146 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
147 * the hint that someone will be waiting on this IO
148 * shortly. The write equivalent of READ_SYNC.
149 * WRITE_ODIRECT Special case write for O_DIRECT only.
150 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
151 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
152 * non-volatile media on completion.
153 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
154 * by a cache flush and data is guaranteed to be on
155 * non-volatile media on completion.
156 *
157 */
158 #define RW_MASK REQ_WRITE
159 #define RWA_MASK REQ_RAHEAD
160
161 #define READ 0
162 #define WRITE RW_MASK
163 #define READA RWA_MASK
164
165 #define READ_SYNC (READ | REQ_SYNC)
166 #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
167 #define WRITE_ODIRECT (WRITE | REQ_SYNC)
168 #define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
169 #define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
170 #define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
171
172 #define SEL_IN 1
173 #define SEL_OUT 2
174 #define SEL_EX 4
175
176 /* public flags for file_system_type */
177 #define FS_REQUIRES_DEV 1
178 #define FS_BINARY_MOUNTDATA 2
179 #define FS_HAS_SUBTYPE 4
180 #define FS_REVAL_DOT 16384 /* Check the paths ".", ".." for staleness */
181 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move()
182 * during rename() internally.
183 */
184
185 /*
186 * These are the fs-independent mount-flags: up to 32 flags are supported
187 */
188 #define MS_RDONLY 1 /* Mount read-only */
189 #define MS_NOSUID 2 /* Ignore suid and sgid bits */
190 #define MS_NODEV 4 /* Disallow access to device special files */
191 #define MS_NOEXEC 8 /* Disallow program execution */
192 #define MS_SYNCHRONOUS 16 /* Writes are synced at once */
193 #define MS_REMOUNT 32 /* Alter flags of a mounted FS */
194 #define MS_MANDLOCK 64 /* Allow mandatory locks on an FS */
195 #define MS_DIRSYNC 128 /* Directory modifications are synchronous */
196 #define MS_NOATIME 1024 /* Do not update access times. */
197 #define MS_NODIRATIME 2048 /* Do not update directory access times */
198 #define MS_BIND 4096
199 #define MS_MOVE 8192
200 #define MS_REC 16384
201 #define MS_VERBOSE 32768 /* War is peace. Verbosity is silence.
202 MS_VERBOSE is deprecated. */
203 #define MS_SILENT 32768
204 #define MS_POSIXACL (1<<16) /* VFS does not apply the umask */
205 #define MS_UNBINDABLE (1<<17) /* change to unbindable */
206 #define MS_PRIVATE (1<<18) /* change to private */
207 #define MS_SLAVE (1<<19) /* change to slave */
208 #define MS_SHARED (1<<20) /* change to shared */
209 #define MS_RELATIME (1<<21) /* Update atime relative to mtime/ctime. */
210 #define MS_KERNMOUNT (1<<22) /* this is a kern_mount call */
211 #define MS_I_VERSION (1<<23) /* Update inode I_version field */
212 #define MS_STRICTATIME (1<<24) /* Always perform atime updates */
213 #define MS_NOSEC (1<<28)
214 #define MS_BORN (1<<29)
215 #define MS_ACTIVE (1<<30)
216 #define MS_NOUSER (1<<31)
217
218 /*
219 * Superblock flags that can be altered by MS_REMOUNT
220 */
221 #define MS_RMT_MASK (MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK|MS_I_VERSION)
222
223 /*
224 * Old magic mount flag and mask
225 */
226 #define MS_MGC_VAL 0xC0ED0000
227 #define MS_MGC_MSK 0xffff0000
228
229 /* Inode flags - they have nothing to superblock flags now */
230
231 #define S_SYNC 1 /* Writes are synced at once */
232 #define S_NOATIME 2 /* Do not update access times */
233 #define S_APPEND 4 /* Append-only file */
234 #define S_IMMUTABLE 8 /* Immutable file */
235 #define S_DEAD 16 /* removed, but still open directory */
236 #define S_NOQUOTA 32 /* Inode is not counted to quota */
237 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
238 #define S_NOCMTIME 128 /* Do not update file c/mtime */
239 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
240 #define S_PRIVATE 512 /* Inode is fs-internal */
241 #define S_IMA 1024 /* Inode has an associated IMA struct */
242 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
243 #define S_NOSEC 4096 /* no suid or xattr security attributes */
244
245 /*
246 * Note that nosuid etc flags are inode-specific: setting some file-system
247 * flags just means all the inodes inherit those flags by default. It might be
248 * possible to override it selectively if you really wanted to with some
249 * ioctl() that is not currently implemented.
250 *
251 * Exception: MS_RDONLY is always applied to the entire file system.
252 *
253 * Unfortunately, it is possible to change a filesystems flags with it mounted
254 * with files in use. This means that all of the inodes will not have their
255 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
256 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
257 */
258 #define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))
259
260 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
261 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
262 ((inode)->i_flags & S_SYNC))
263 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
264 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
265 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
266 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
267 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
268
269 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
270 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
271 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
272 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
273
274 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
275 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
276 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
277 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
278 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
279 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
280 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
281
282 /* the read-only stuff doesn't really belong here, but any other place is
283 probably as bad and I don't want to create yet another include file. */
284
285 #define BLKROSET _IO(0x12,93) /* set device read-only (0 = read-write) */
286 #define BLKROGET _IO(0x12,94) /* get read-only status (0 = read_write) */
287 #define BLKRRPART _IO(0x12,95) /* re-read partition table */
288 #define BLKGETSIZE _IO(0x12,96) /* return device size /512 (long *arg) */
289 #define BLKFLSBUF _IO(0x12,97) /* flush buffer cache */
290 #define BLKRASET _IO(0x12,98) /* set read ahead for block device */
291 #define BLKRAGET _IO(0x12,99) /* get current read ahead setting */
292 #define BLKFRASET _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
293 #define BLKFRAGET _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
294 #define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
295 #define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
296 #define BLKSSZGET _IO(0x12,104)/* get block device sector size */
297 #if 0
298 #define BLKPG _IO(0x12,105)/* See blkpg.h */
299
300 /* Some people are morons. Do not use sizeof! */
301
302 #define BLKELVGET _IOR(0x12,106,size_t)/* elevator get */
303 #define BLKELVSET _IOW(0x12,107,size_t)/* elevator set */
304 /* This was here just to show that the number is taken -
305 probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
306 #endif
307 /* A jump here: 108-111 have been used for various private purposes. */
308 #define BLKBSZGET _IOR(0x12,112,size_t)
309 #define BLKBSZSET _IOW(0x12,113,size_t)
310 #define BLKGETSIZE64 _IOR(0x12,114,size_t) /* return device size in bytes (u64 *arg) */
311 #define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup)
312 #define BLKTRACESTART _IO(0x12,116)
313 #define BLKTRACESTOP _IO(0x12,117)
314 #define BLKTRACETEARDOWN _IO(0x12,118)
315 #define BLKDISCARD _IO(0x12,119)
316 #define BLKIOMIN _IO(0x12,120)
317 #define BLKIOOPT _IO(0x12,121)
318 #define BLKALIGNOFF _IO(0x12,122)
319 #define BLKPBSZGET _IO(0x12,123)
320 #define BLKDISCARDZEROES _IO(0x12,124)
321 #define BLKSECDISCARD _IO(0x12,125)
322
323 #define BMAP_IOCTL 1 /* obsolete - kept for compatibility */
324 #define FIBMAP _IO(0x00,1) /* bmap access */
325 #define FIGETBSZ _IO(0x00,2) /* get the block size used for bmap */
326 #define FIFREEZE _IOWR('X', 119, int) /* Freeze */
327 #define FITHAW _IOWR('X', 120, int) /* Thaw */
328 #define FITRIM _IOWR('X', 121, struct fstrim_range) /* Trim */
329
330 #define FS_IOC_GETFLAGS _IOR('f', 1, long)
331 #define FS_IOC_SETFLAGS _IOW('f', 2, long)
332 #define FS_IOC_GETVERSION _IOR('v', 1, long)
333 #define FS_IOC_SETVERSION _IOW('v', 2, long)
334 #define FS_IOC_FIEMAP _IOWR('f', 11, struct fiemap)
335 #define FS_IOC32_GETFLAGS _IOR('f', 1, int)
336 #define FS_IOC32_SETFLAGS _IOW('f', 2, int)
337 #define FS_IOC32_GETVERSION _IOR('v', 1, int)
338 #define FS_IOC32_SETVERSION _IOW('v', 2, int)
339
340 /*
341 * Inode flags (FS_IOC_GETFLAGS / FS_IOC_SETFLAGS)
342 */
343 #define FS_SECRM_FL 0x00000001 /* Secure deletion */
344 #define FS_UNRM_FL 0x00000002 /* Undelete */
345 #define FS_COMPR_FL 0x00000004 /* Compress file */
346 #define FS_SYNC_FL 0x00000008 /* Synchronous updates */
347 #define FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
348 #define FS_APPEND_FL 0x00000020 /* writes to file may only append */
349 #define FS_NODUMP_FL 0x00000040 /* do not dump file */
350 #define FS_NOATIME_FL 0x00000080 /* do not update atime */
351 /* Reserved for compression usage... */
352 #define FS_DIRTY_FL 0x00000100
353 #define FS_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
354 #define FS_NOCOMP_FL 0x00000400 /* Don't compress */
355 #define FS_ECOMPR_FL 0x00000800 /* Compression error */
356 /* End compression flags --- maybe not all used */
357 #define FS_BTREE_FL 0x00001000 /* btree format dir */
358 #define FS_INDEX_FL 0x00001000 /* hash-indexed directory */
359 #define FS_IMAGIC_FL 0x00002000 /* AFS directory */
360 #define FS_JOURNAL_DATA_FL 0x00004000 /* Reserved for ext3 */
361 #define FS_NOTAIL_FL 0x00008000 /* file tail should not be merged */
362 #define FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
363 #define FS_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
364 #define FS_EXTENT_FL 0x00080000 /* Extents */
365 #define FS_DIRECTIO_FL 0x00100000 /* Use direct i/o */
366 #define FS_NOCOW_FL 0x00800000 /* Do not cow file */
367 #define FS_RESERVED_FL 0x80000000 /* reserved for ext2 lib */
368
369 #define FS_FL_USER_VISIBLE 0x0003DFFF /* User visible flags */
370 #define FS_FL_USER_MODIFIABLE 0x000380FF /* User modifiable flags */
371
372
373 #define SYNC_FILE_RANGE_WAIT_BEFORE 1
374 #define SYNC_FILE_RANGE_WRITE 2
375 #define SYNC_FILE_RANGE_WAIT_AFTER 4
376
377 #ifdef __KERNEL__
378
379 #include <linux/linkage.h>
380 #include <linux/wait.h>
381 #include <linux/kdev_t.h>
382 #include <linux/dcache.h>
383 #include <linux/path.h>
384 #include <linux/stat.h>
385 #include <linux/cache.h>
386 #include <linux/list.h>
387 #include <linux/radix-tree.h>
388 #include <linux/prio_tree.h>
389 #include <linux/init.h>
390 #include <linux/pid.h>
391 #include <linux/mutex.h>
392 #include <linux/capability.h>
393 #include <linux/semaphore.h>
394 #include <linux/fiemap.h>
395 #include <linux/rculist_bl.h>
396 #include <linux/atomic.h>
397 #include <linux/shrinker.h>
398
399 #include <asm/byteorder.h>
400
401 struct export_operations;
402 struct hd_geometry;
403 struct iovec;
404 struct nameidata;
405 struct kiocb;
406 struct kobject;
407 struct pipe_inode_info;
408 struct poll_table_struct;
409 struct kstatfs;
410 struct vm_area_struct;
411 struct vfsmount;
412 struct cred;
413
414 extern void __init inode_init(void);
415 extern void __init inode_init_early(void);
416 extern void __init files_init(unsigned long);
417
418 extern struct files_stat_struct files_stat;
419 extern unsigned long get_max_files(void);
420 extern int sysctl_nr_open;
421 extern struct inodes_stat_t inodes_stat;
422 extern int leases_enable, lease_break_time;
423
424 struct buffer_head;
425 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
426 struct buffer_head *bh_result, int create);
427 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
428 ssize_t bytes, void *private, int ret,
429 bool is_async);
430
431 /*
432 * Attribute flags. These should be or-ed together to figure out what
433 * has been changed!
434 */
435 #define ATTR_MODE (1 << 0)
436 #define ATTR_UID (1 << 1)
437 #define ATTR_GID (1 << 2)
438 #define ATTR_SIZE (1 << 3)
439 #define ATTR_ATIME (1 << 4)
440 #define ATTR_MTIME (1 << 5)
441 #define ATTR_CTIME (1 << 6)
442 #define ATTR_ATIME_SET (1 << 7)
443 #define ATTR_MTIME_SET (1 << 8)
444 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
445 #define ATTR_ATTR_FLAG (1 << 10)
446 #define ATTR_KILL_SUID (1 << 11)
447 #define ATTR_KILL_SGID (1 << 12)
448 #define ATTR_FILE (1 << 13)
449 #define ATTR_KILL_PRIV (1 << 14)
450 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
451 #define ATTR_TIMES_SET (1 << 16)
452
453 /*
454 * This is the Inode Attributes structure, used for notify_change(). It
455 * uses the above definitions as flags, to know which values have changed.
456 * Also, in this manner, a Filesystem can look at only the values it cares
457 * about. Basically, these are the attributes that the VFS layer can
458 * request to change from the FS layer.
459 *
460 * Derek Atkins <warlord@MIT.EDU> 94-10-20
461 */
462 struct iattr {
463 unsigned int ia_valid;
464 umode_t ia_mode;
465 uid_t ia_uid;
466 gid_t ia_gid;
467 loff_t ia_size;
468 struct timespec ia_atime;
469 struct timespec ia_mtime;
470 struct timespec ia_ctime;
471
472 /*
473 * Not an attribute, but an auxiliary info for filesystems wanting to
474 * implement an ftruncate() like method. NOTE: filesystem should
475 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
476 */
477 struct file *ia_file;
478 };
479
480 /*
481 * Includes for diskquotas.
482 */
483 #include <linux/quota.h>
484
485 /**
486 * enum positive_aop_returns - aop return codes with specific semantics
487 *
488 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
489 * completed, that the page is still locked, and
490 * should be considered active. The VM uses this hint
491 * to return the page to the active list -- it won't
492 * be a candidate for writeback again in the near
493 * future. Other callers must be careful to unlock
494 * the page if they get this return. Returned by
495 * writepage();
496 *
497 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
498 * unlocked it and the page might have been truncated.
499 * The caller should back up to acquiring a new page and
500 * trying again. The aop will be taking reasonable
501 * precautions not to livelock. If the caller held a page
502 * reference, it should drop it before retrying. Returned
503 * by readpage().
504 *
505 * address_space_operation functions return these large constants to indicate
506 * special semantics to the caller. These are much larger than the bytes in a
507 * page to allow for functions that return the number of bytes operated on in a
508 * given page.
509 */
510
511 enum positive_aop_returns {
512 AOP_WRITEPAGE_ACTIVATE = 0x80000,
513 AOP_TRUNCATED_PAGE = 0x80001,
514 };
515
516 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
517 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
518 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
519 * helper code (eg buffer layer)
520 * to clear GFP_FS from alloc */
521
522 /*
523 * oh the beauties of C type declarations.
524 */
525 struct page;
526 struct address_space;
527 struct writeback_control;
528 enum migrate_mode;
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 /*
614 * migrate the contents of a page to the specified target. If sync
615 * is false, it must not block.
616 */
617 int (*migratepage) (struct address_space *,
618 struct page *, struct page *, enum migrate_mode);
619 int (*launder_page) (struct page *);
620 int (*is_partially_uptodate) (struct page *, read_descriptor_t *,
621 unsigned long);
622 int (*error_remove_page)(struct address_space *, struct page *);
623 };
624
625 extern const struct address_space_operations empty_aops;
626
627 /*
628 * pagecache_write_begin/pagecache_write_end must be used by general code
629 * to write into the pagecache.
630 */
631 int pagecache_write_begin(struct file *, struct address_space *mapping,
632 loff_t pos, unsigned len, unsigned flags,
633 struct page **pagep, void **fsdata);
634
635 int pagecache_write_end(struct file *, struct address_space *mapping,
636 loff_t pos, unsigned len, unsigned copied,
637 struct page *page, void *fsdata);
638
639 struct backing_dev_info;
640 struct address_space {
641 struct inode *host; /* owner: inode, block_device */
642 struct radix_tree_root page_tree; /* radix tree of all pages */
643 spinlock_t tree_lock; /* and lock protecting it */
644 unsigned int i_mmap_writable;/* count VM_SHARED mappings */
645 struct prio_tree_root i_mmap; /* tree of private and shared mappings */
646 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
647 struct mutex i_mmap_mutex; /* protect tree, count, list */
648 /* Protected by tree_lock together with the radix tree */
649 unsigned long nrpages; /* number of total pages */
650 pgoff_t writeback_index;/* writeback starts here */
651 const struct address_space_operations *a_ops; /* methods */
652 unsigned long flags; /* error bits/gfp mask */
653 struct backing_dev_info *backing_dev_info; /* device readahead, etc */
654 spinlock_t private_lock; /* for use by the address_space */
655 struct list_head private_list; /* ditto */
656 struct address_space *assoc_mapping; /* ditto */
657 } __attribute__((aligned(sizeof(long))));
658 /*
659 * On most architectures that alignment is already the case; but
660 * must be enforced here for CRIS, to let the least significant bit
661 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
662 */
663 struct request_queue;
664
665 struct block_device {
666 dev_t bd_dev; /* not a kdev_t - it's a search key */
667 int bd_openers;
668 struct inode * bd_inode; /* will die */
669 struct super_block * bd_super;
670 struct mutex bd_mutex; /* open/close mutex */
671 struct list_head bd_inodes;
672 void * bd_claiming;
673 void * bd_holder;
674 int bd_holders;
675 bool bd_write_holder;
676 #ifdef CONFIG_SYSFS
677 struct list_head bd_holder_disks;
678 #endif
679 struct block_device * bd_contains;
680 unsigned bd_block_size;
681 struct hd_struct * bd_part;
682 /* number of times partitions within this device have been opened. */
683 unsigned bd_part_count;
684 int bd_invalidated;
685 struct gendisk * bd_disk;
686 struct request_queue * bd_queue;
687 struct list_head bd_list;
688 /*
689 * Private data. You must have bd_claim'ed the block_device
690 * to use this. NOTE: bd_claim allows an owner to claim
691 * the same device multiple times, the owner must take special
692 * care to not mess up bd_private for that case.
693 */
694 unsigned long bd_private;
695
696 /* The counter of freeze processes */
697 int bd_fsfreeze_count;
698 /* Mutex for freeze */
699 struct mutex bd_fsfreeze_mutex;
700 };
701
702 /*
703 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
704 * radix trees
705 */
706 #define PAGECACHE_TAG_DIRTY 0
707 #define PAGECACHE_TAG_WRITEBACK 1
708 #define PAGECACHE_TAG_TOWRITE 2
709
710 int mapping_tagged(struct address_space *mapping, int tag);
711
712 /*
713 * Might pages of this file be mapped into userspace?
714 */
715 static inline int mapping_mapped(struct address_space *mapping)
716 {
717 return !prio_tree_empty(&mapping->i_mmap) ||
718 !list_empty(&mapping->i_mmap_nonlinear);
719 }
720
721 /*
722 * Might pages of this file have been modified in userspace?
723 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
724 * marks vma as VM_SHARED if it is shared, and the file was opened for
725 * writing i.e. vma may be mprotected writable even if now readonly.
726 */
727 static inline int mapping_writably_mapped(struct address_space *mapping)
728 {
729 return mapping->i_mmap_writable != 0;
730 }
731
732 /*
733 * Use sequence counter to get consistent i_size on 32-bit processors.
734 */
735 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
736 #include <linux/seqlock.h>
737 #define __NEED_I_SIZE_ORDERED
738 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
739 #else
740 #define i_size_ordered_init(inode) do { } while (0)
741 #endif
742
743 struct posix_acl;
744 #define ACL_NOT_CACHED ((void *)(-1))
745
746 #define IOP_FASTPERM 0x0001
747 #define IOP_LOOKUP 0x0002
748 #define IOP_NOFOLLOW 0x0004
749
750 /*
751 * Keep mostly read-only and often accessed (especially for
752 * the RCU path lookup and 'stat' data) fields at the beginning
753 * of the 'struct inode'
754 */
755 struct inode {
756 umode_t i_mode;
757 unsigned short i_opflags;
758 uid_t i_uid;
759 gid_t i_gid;
760 unsigned int i_flags;
761
762 #ifdef CONFIG_FS_POSIX_ACL
763 struct posix_acl *i_acl;
764 struct posix_acl *i_default_acl;
765 #endif
766
767 const struct inode_operations *i_op;
768 struct super_block *i_sb;
769 struct address_space *i_mapping;
770
771 #ifdef CONFIG_SECURITY
772 void *i_security;
773 #endif
774
775 /* Stat data, not accessed from path walking */
776 unsigned long i_ino;
777 /*
778 * Filesystems may only read i_nlink directly. They shall use the
779 * following functions for modification:
780 *
781 * (set|clear|inc|drop)_nlink
782 * inode_(inc|dec)_link_count
783 */
784 union {
785 const unsigned int i_nlink;
786 unsigned int __i_nlink;
787 };
788 dev_t i_rdev;
789 struct timespec i_atime;
790 struct timespec i_mtime;
791 struct timespec i_ctime;
792 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
793 unsigned short i_bytes;
794 blkcnt_t i_blocks;
795 loff_t i_size;
796
797 #ifdef __NEED_I_SIZE_ORDERED
798 seqcount_t i_size_seqcount;
799 #endif
800
801 /* Misc */
802 unsigned long i_state;
803 struct mutex i_mutex;
804
805 unsigned long dirtied_when; /* jiffies of first dirtying */
806
807 struct hlist_node i_hash;
808 struct list_head i_wb_list; /* backing dev IO list */
809 struct list_head i_lru; /* inode LRU list */
810 struct list_head i_sb_list;
811 union {
812 struct list_head i_dentry;
813 struct rcu_head i_rcu;
814 };
815 atomic_t i_count;
816 unsigned int i_blkbits;
817 u64 i_version;
818 atomic_t i_dio_count;
819 atomic_t i_writecount;
820 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
821 struct file_lock *i_flock;
822 struct address_space i_data;
823 #ifdef CONFIG_QUOTA
824 struct dquot *i_dquot[MAXQUOTAS];
825 #endif
826 struct list_head i_devices;
827 union {
828 struct pipe_inode_info *i_pipe;
829 struct block_device *i_bdev;
830 struct cdev *i_cdev;
831 };
832
833 __u32 i_generation;
834
835 #ifdef CONFIG_FSNOTIFY
836 __u32 i_fsnotify_mask; /* all events this inode cares about */
837 struct hlist_head i_fsnotify_marks;
838 #endif
839
840 #ifdef CONFIG_IMA
841 atomic_t i_readcount; /* struct files open RO */
842 #endif
843 void *i_private; /* fs or device private pointer */
844 };
845
846 static inline int inode_unhashed(struct inode *inode)
847 {
848 return hlist_unhashed(&inode->i_hash);
849 }
850
851 /*
852 * inode->i_mutex nesting subclasses for the lock validator:
853 *
854 * 0: the object of the current VFS operation
855 * 1: parent
856 * 2: child/target
857 * 3: quota file
858 *
859 * The locking order between these classes is
860 * parent -> child -> normal -> xattr -> quota
861 */
862 enum inode_i_mutex_lock_class
863 {
864 I_MUTEX_NORMAL,
865 I_MUTEX_PARENT,
866 I_MUTEX_CHILD,
867 I_MUTEX_XATTR,
868 I_MUTEX_QUOTA
869 };
870
871 /*
872 * NOTE: in a 32bit arch with a preemptable kernel and
873 * an UP compile the i_size_read/write must be atomic
874 * with respect to the local cpu (unlike with preempt disabled),
875 * but they don't need to be atomic with respect to other cpus like in
876 * true SMP (so they need either to either locally disable irq around
877 * the read or for example on x86 they can be still implemented as a
878 * cmpxchg8b without the need of the lock prefix). For SMP compiles
879 * and 64bit archs it makes no difference if preempt is enabled or not.
880 */
881 static inline loff_t i_size_read(const struct inode *inode)
882 {
883 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
884 loff_t i_size;
885 unsigned int seq;
886
887 do {
888 seq = read_seqcount_begin(&inode->i_size_seqcount);
889 i_size = inode->i_size;
890 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
891 return i_size;
892 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
893 loff_t i_size;
894
895 preempt_disable();
896 i_size = inode->i_size;
897 preempt_enable();
898 return i_size;
899 #else
900 return inode->i_size;
901 #endif
902 }
903
904 /*
905 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
906 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
907 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
908 */
909 static inline void i_size_write(struct inode *inode, loff_t i_size)
910 {
911 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
912 write_seqcount_begin(&inode->i_size_seqcount);
913 inode->i_size = i_size;
914 write_seqcount_end(&inode->i_size_seqcount);
915 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
916 preempt_disable();
917 inode->i_size = i_size;
918 preempt_enable();
919 #else
920 inode->i_size = i_size;
921 #endif
922 }
923
924 static inline unsigned iminor(const struct inode *inode)
925 {
926 return MINOR(inode->i_rdev);
927 }
928
929 static inline unsigned imajor(const struct inode *inode)
930 {
931 return MAJOR(inode->i_rdev);
932 }
933
934 extern struct block_device *I_BDEV(struct inode *inode);
935
936 struct fown_struct {
937 rwlock_t lock; /* protects pid, uid, euid fields */
938 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
939 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
940 uid_t uid, euid; /* uid/euid of process setting the owner */
941 int signum; /* posix.1b rt signal to be delivered on IO */
942 };
943
944 /*
945 * Track a single file's readahead state
946 */
947 struct file_ra_state {
948 pgoff_t start; /* where readahead started */
949 unsigned int size; /* # of readahead pages */
950 unsigned int async_size; /* do asynchronous readahead when
951 there are only # of pages ahead */
952
953 unsigned int ra_pages; /* Maximum readahead window */
954 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
955 loff_t prev_pos; /* Cache last read() position */
956 };
957
958 /*
959 * Check if @index falls in the readahead windows.
960 */
961 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
962 {
963 return (index >= ra->start &&
964 index < ra->start + ra->size);
965 }
966
967 #define FILE_MNT_WRITE_TAKEN 1
968 #define FILE_MNT_WRITE_RELEASED 2
969
970 struct file {
971 /*
972 * fu_list becomes invalid after file_free is called and queued via
973 * fu_rcuhead for RCU freeing
974 */
975 union {
976 struct list_head fu_list;
977 struct rcu_head fu_rcuhead;
978 } f_u;
979 struct path f_path;
980 #define f_dentry f_path.dentry
981 #define f_vfsmnt f_path.mnt
982 const struct file_operations *f_op;
983
984 /*
985 * Protects f_ep_links, f_flags, f_pos vs i_size in lseek SEEK_CUR.
986 * Must not be taken from IRQ context.
987 */
988 spinlock_t f_lock;
989 #ifdef CONFIG_SMP
990 int f_sb_list_cpu;
991 #endif
992 atomic_long_t f_count;
993 unsigned int f_flags;
994 fmode_t f_mode;
995 loff_t f_pos;
996 struct fown_struct f_owner;
997 const struct cred *f_cred;
998 struct file_ra_state f_ra;
999
1000 u64 f_version;
1001 #ifdef CONFIG_SECURITY
1002 void *f_security;
1003 #endif
1004 /* needed for tty driver, and maybe others */
1005 void *private_data;
1006
1007 #ifdef CONFIG_EPOLL
1008 /* Used by fs/eventpoll.c to link all the hooks to this file */
1009 struct list_head f_ep_links;
1010 struct list_head f_tfile_llink;
1011 #endif /* #ifdef CONFIG_EPOLL */
1012 struct address_space *f_mapping;
1013 #ifdef CONFIG_DEBUG_WRITECOUNT
1014 unsigned long f_mnt_write_state;
1015 #endif
1016 };
1017
1018 struct file_handle {
1019 __u32 handle_bytes;
1020 int handle_type;
1021 /* file identifier */
1022 unsigned char f_handle[0];
1023 };
1024
1025 #define get_file(x) atomic_long_inc(&(x)->f_count)
1026 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
1027 #define file_count(x) atomic_long_read(&(x)->f_count)
1028
1029 #ifdef CONFIG_DEBUG_WRITECOUNT
1030 static inline void file_take_write(struct file *f)
1031 {
1032 WARN_ON(f->f_mnt_write_state != 0);
1033 f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN;
1034 }
1035 static inline void file_release_write(struct file *f)
1036 {
1037 f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED;
1038 }
1039 static inline void file_reset_write(struct file *f)
1040 {
1041 f->f_mnt_write_state = 0;
1042 }
1043 static inline void file_check_state(struct file *f)
1044 {
1045 /*
1046 * At this point, either both or neither of these bits
1047 * should be set.
1048 */
1049 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN);
1050 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED);
1051 }
1052 static inline int file_check_writeable(struct file *f)
1053 {
1054 if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN)
1055 return 0;
1056 printk(KERN_WARNING "writeable file with no "
1057 "mnt_want_write()\n");
1058 WARN_ON(1);
1059 return -EINVAL;
1060 }
1061 #else /* !CONFIG_DEBUG_WRITECOUNT */
1062 static inline void file_take_write(struct file *filp) {}
1063 static inline void file_release_write(struct file *filp) {}
1064 static inline void file_reset_write(struct file *filp) {}
1065 static inline void file_check_state(struct file *filp) {}
1066 static inline int file_check_writeable(struct file *filp)
1067 {
1068 return 0;
1069 }
1070 #endif /* CONFIG_DEBUG_WRITECOUNT */
1071
1072 #define MAX_NON_LFS ((1UL<<31) - 1)
1073
1074 /* Page cache limit. The filesystems should put that into their s_maxbytes
1075 limits, otherwise bad things can happen in VM. */
1076 #if BITS_PER_LONG==32
1077 #define MAX_LFS_FILESIZE (((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
1078 #elif BITS_PER_LONG==64
1079 #define MAX_LFS_FILESIZE 0x7fffffffffffffffUL
1080 #endif
1081
1082 #define FL_POSIX 1
1083 #define FL_FLOCK 2
1084 #define FL_ACCESS 8 /* not trying to lock, just looking */
1085 #define FL_EXISTS 16 /* when unlocking, test for existence */
1086 #define FL_LEASE 32 /* lease held on this file */
1087 #define FL_CLOSE 64 /* unlock on close */
1088 #define FL_SLEEP 128 /* A blocking lock */
1089 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
1090 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
1091
1092 /*
1093 * Special return value from posix_lock_file() and vfs_lock_file() for
1094 * asynchronous locking.
1095 */
1096 #define FILE_LOCK_DEFERRED 1
1097
1098 /*
1099 * The POSIX file lock owner is determined by
1100 * the "struct files_struct" in the thread group
1101 * (or NULL for no owner - BSD locks).
1102 *
1103 * Lockd stuffs a "host" pointer into this.
1104 */
1105 typedef struct files_struct *fl_owner_t;
1106
1107 struct file_lock_operations {
1108 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1109 void (*fl_release_private)(struct file_lock *);
1110 };
1111
1112 struct lock_manager_operations {
1113 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
1114 void (*lm_notify)(struct file_lock *); /* unblock callback */
1115 int (*lm_grant)(struct file_lock *, struct file_lock *, int);
1116 void (*lm_release_private)(struct file_lock *);
1117 void (*lm_break)(struct file_lock *);
1118 int (*lm_change)(struct file_lock **, int);
1119 };
1120
1121 struct lock_manager {
1122 struct list_head list;
1123 };
1124
1125 void locks_start_grace(struct lock_manager *);
1126 void locks_end_grace(struct lock_manager *);
1127 int locks_in_grace(void);
1128
1129 /* that will die - we need it for nfs_lock_info */
1130 #include <linux/nfs_fs_i.h>
1131
1132 struct file_lock {
1133 struct file_lock *fl_next; /* singly linked list for this inode */
1134 struct list_head fl_link; /* doubly linked list of all locks */
1135 struct list_head fl_block; /* circular list of blocked processes */
1136 fl_owner_t fl_owner;
1137 unsigned int fl_flags;
1138 unsigned char fl_type;
1139 unsigned int fl_pid;
1140 struct pid *fl_nspid;
1141 wait_queue_head_t fl_wait;
1142 struct file *fl_file;
1143 loff_t fl_start;
1144 loff_t fl_end;
1145
1146 struct fasync_struct * fl_fasync; /* for lease break notifications */
1147 /* for lease breaks: */
1148 unsigned long fl_break_time;
1149 unsigned long fl_downgrade_time;
1150
1151 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1152 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1153 union {
1154 struct nfs_lock_info nfs_fl;
1155 struct nfs4_lock_info nfs4_fl;
1156 struct {
1157 struct list_head link; /* link in AFS vnode's pending_locks list */
1158 int state; /* state of grant or error if -ve */
1159 } afs;
1160 } fl_u;
1161 };
1162
1163 /* The following constant reflects the upper bound of the file/locking space */
1164 #ifndef OFFSET_MAX
1165 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1166 #define OFFSET_MAX INT_LIMIT(loff_t)
1167 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1168 #endif
1169
1170 #include <linux/fcntl.h>
1171
1172 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1173
1174 #ifdef CONFIG_FILE_LOCKING
1175 extern int fcntl_getlk(struct file *, struct flock __user *);
1176 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1177 struct flock __user *);
1178
1179 #if BITS_PER_LONG == 32
1180 extern int fcntl_getlk64(struct file *, struct flock64 __user *);
1181 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1182 struct flock64 __user *);
1183 #endif
1184
1185 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1186 extern int fcntl_getlease(struct file *filp);
1187
1188 /* fs/locks.c */
1189 void locks_free_lock(struct file_lock *fl);
1190 extern void locks_init_lock(struct file_lock *);
1191 extern struct file_lock * locks_alloc_lock(void);
1192 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1193 extern void __locks_copy_lock(struct file_lock *, const struct file_lock *);
1194 extern void locks_remove_posix(struct file *, fl_owner_t);
1195 extern void locks_remove_flock(struct file *);
1196 extern void locks_release_private(struct file_lock *);
1197 extern void posix_test_lock(struct file *, struct file_lock *);
1198 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1199 extern int posix_lock_file_wait(struct file *, struct file_lock *);
1200 extern int posix_unblock_lock(struct file *, struct file_lock *);
1201 extern int vfs_test_lock(struct file *, struct file_lock *);
1202 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1203 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1204 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
1205 extern int __break_lease(struct inode *inode, unsigned int flags);
1206 extern void lease_get_mtime(struct inode *, struct timespec *time);
1207 extern int generic_setlease(struct file *, long, struct file_lock **);
1208 extern int vfs_setlease(struct file *, long, struct file_lock **);
1209 extern int lease_modify(struct file_lock **, int);
1210 extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
1211 extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
1212 extern void lock_flocks(void);
1213 extern void unlock_flocks(void);
1214 #else /* !CONFIG_FILE_LOCKING */
1215 static inline int fcntl_getlk(struct file *file, struct flock __user *user)
1216 {
1217 return -EINVAL;
1218 }
1219
1220 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1221 unsigned int cmd, struct flock __user *user)
1222 {
1223 return -EACCES;
1224 }
1225
1226 #if BITS_PER_LONG == 32
1227 static inline int fcntl_getlk64(struct file *file, struct flock64 __user *user)
1228 {
1229 return -EINVAL;
1230 }
1231
1232 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1233 unsigned int cmd, struct flock64 __user *user)
1234 {
1235 return -EACCES;
1236 }
1237 #endif
1238 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1239 {
1240 return 0;
1241 }
1242
1243 static inline int fcntl_getlease(struct file *filp)
1244 {
1245 return 0;
1246 }
1247
1248 static inline void locks_init_lock(struct file_lock *fl)
1249 {
1250 return;
1251 }
1252
1253 static inline void __locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1254 {
1255 return;
1256 }
1257
1258 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1259 {
1260 return;
1261 }
1262
1263 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1264 {
1265 return;
1266 }
1267
1268 static inline void locks_remove_flock(struct file *filp)
1269 {
1270 return;
1271 }
1272
1273 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1274 {
1275 return;
1276 }
1277
1278 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1279 struct file_lock *conflock)
1280 {
1281 return -ENOLCK;
1282 }
1283
1284 static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1285 {
1286 return -ENOLCK;
1287 }
1288
1289 static inline int posix_unblock_lock(struct file *filp,
1290 struct file_lock *waiter)
1291 {
1292 return -ENOENT;
1293 }
1294
1295 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1296 {
1297 return 0;
1298 }
1299
1300 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1301 struct file_lock *fl, struct file_lock *conf)
1302 {
1303 return -ENOLCK;
1304 }
1305
1306 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1307 {
1308 return 0;
1309 }
1310
1311 static inline int flock_lock_file_wait(struct file *filp,
1312 struct file_lock *request)
1313 {
1314 return -ENOLCK;
1315 }
1316
1317 static inline int __break_lease(struct inode *inode, unsigned int mode)
1318 {
1319 return 0;
1320 }
1321
1322 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1323 {
1324 return;
1325 }
1326
1327 static inline int generic_setlease(struct file *filp, long arg,
1328 struct file_lock **flp)
1329 {
1330 return -EINVAL;
1331 }
1332
1333 static inline int vfs_setlease(struct file *filp, long arg,
1334 struct file_lock **lease)
1335 {
1336 return -EINVAL;
1337 }
1338
1339 static inline int lease_modify(struct file_lock **before, int arg)
1340 {
1341 return -EINVAL;
1342 }
1343
1344 static inline int lock_may_read(struct inode *inode, loff_t start,
1345 unsigned long len)
1346 {
1347 return 1;
1348 }
1349
1350 static inline int lock_may_write(struct inode *inode, loff_t start,
1351 unsigned long len)
1352 {
1353 return 1;
1354 }
1355
1356 static inline void lock_flocks(void)
1357 {
1358 }
1359
1360 static inline void unlock_flocks(void)
1361 {
1362 }
1363
1364 #endif /* !CONFIG_FILE_LOCKING */
1365
1366
1367 struct fasync_struct {
1368 spinlock_t fa_lock;
1369 int magic;
1370 int fa_fd;
1371 struct fasync_struct *fa_next; /* singly linked list */
1372 struct file *fa_file;
1373 struct rcu_head fa_rcu;
1374 };
1375
1376 #define FASYNC_MAGIC 0x4601
1377
1378 /* SMP safe fasync helpers: */
1379 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1380 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1381 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1382 extern struct fasync_struct *fasync_alloc(void);
1383 extern void fasync_free(struct fasync_struct *);
1384
1385 /* can be called from interrupts */
1386 extern void kill_fasync(struct fasync_struct **, int, int);
1387
1388 extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1389 extern int f_setown(struct file *filp, unsigned long arg, int force);
1390 extern void f_delown(struct file *filp);
1391 extern pid_t f_getown(struct file *filp);
1392 extern int send_sigurg(struct fown_struct *fown);
1393
1394 /*
1395 * Umount options
1396 */
1397
1398 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1399 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1400 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1401 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1402 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1403
1404 extern struct list_head super_blocks;
1405 extern spinlock_t sb_lock;
1406
1407 struct super_block {
1408 struct list_head s_list; /* Keep this first */
1409 dev_t s_dev; /* search index; _not_ kdev_t */
1410 unsigned char s_dirt;
1411 unsigned char s_blocksize_bits;
1412 unsigned long s_blocksize;
1413 loff_t s_maxbytes; /* Max file size */
1414 struct file_system_type *s_type;
1415 const struct super_operations *s_op;
1416 const struct dquot_operations *dq_op;
1417 const struct quotactl_ops *s_qcop;
1418 const struct export_operations *s_export_op;
1419 unsigned long s_flags;
1420 unsigned long s_magic;
1421 struct dentry *s_root;
1422 struct rw_semaphore s_umount;
1423 struct mutex s_lock;
1424 int s_count;
1425 atomic_t s_active;
1426 #ifdef CONFIG_SECURITY
1427 void *s_security;
1428 #endif
1429 const struct xattr_handler **s_xattr;
1430
1431 struct list_head s_inodes; /* all inodes */
1432 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1433 #ifdef CONFIG_SMP
1434 struct list_head __percpu *s_files;
1435 #else
1436 struct list_head s_files;
1437 #endif
1438 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1439 /* s_dentry_lru, s_nr_dentry_unused protected by dcache.c lru locks */
1440 struct list_head s_dentry_lru; /* unused dentry lru */
1441 int s_nr_dentry_unused; /* # of dentry on lru */
1442
1443 /* s_inode_lru_lock protects s_inode_lru and s_nr_inodes_unused */
1444 spinlock_t s_inode_lru_lock ____cacheline_aligned_in_smp;
1445 struct list_head s_inode_lru; /* unused inode lru */
1446 int s_nr_inodes_unused; /* # of inodes on lru */
1447
1448 struct block_device *s_bdev;
1449 struct backing_dev_info *s_bdi;
1450 struct mtd_info *s_mtd;
1451 struct hlist_node s_instances;
1452 struct quota_info s_dquot; /* Diskquota specific options */
1453
1454 int s_frozen;
1455 wait_queue_head_t s_wait_unfrozen;
1456
1457 char s_id[32]; /* Informational name */
1458 u8 s_uuid[16]; /* UUID */
1459
1460 void *s_fs_info; /* Filesystem private info */
1461 fmode_t s_mode;
1462
1463 /* Granularity of c/m/atime in ns.
1464 Cannot be worse than a second */
1465 u32 s_time_gran;
1466
1467 /*
1468 * The next field is for VFS *only*. No filesystems have any business
1469 * even looking at it. You had been warned.
1470 */
1471 struct mutex s_vfs_rename_mutex; /* Kludge */
1472
1473 /*
1474 * Filesystem subtype. If non-empty the filesystem type field
1475 * in /proc/mounts will be "type.subtype"
1476 */
1477 char *s_subtype;
1478
1479 /*
1480 * Saved mount options for lazy filesystems using
1481 * generic_show_options()
1482 */
1483 char __rcu *s_options;
1484 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1485
1486 /*
1487 * Saved pool identifier for cleancache (-1 means none)
1488 */
1489 int cleancache_poolid;
1490
1491 struct shrinker s_shrink; /* per-sb shrinker handle */
1492
1493 /* Number of inodes with nlink == 0 but still referenced */
1494 atomic_long_t s_remove_count;
1495
1496 /* Being remounted read-only */
1497 int s_readonly_remount;
1498 };
1499
1500 /* superblock cache pruning functions */
1501 extern void prune_icache_sb(struct super_block *sb, int nr_to_scan);
1502 extern void prune_dcache_sb(struct super_block *sb, int nr_to_scan);
1503
1504 extern struct timespec current_fs_time(struct super_block *sb);
1505
1506 /*
1507 * Snapshotting support.
1508 */
1509 enum {
1510 SB_UNFROZEN = 0,
1511 SB_FREEZE_WRITE = 1,
1512 SB_FREEZE_TRANS = 2,
1513 };
1514
1515 #define vfs_check_frozen(sb, level) \
1516 wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
1517
1518 /*
1519 * until VFS tracks user namespaces for inodes, just make all files
1520 * belong to init_user_ns
1521 */
1522 extern struct user_namespace init_user_ns;
1523 #define inode_userns(inode) (&init_user_ns)
1524 extern bool inode_owner_or_capable(const struct inode *inode);
1525
1526 /* not quite ready to be deprecated, but... */
1527 extern void lock_super(struct super_block *);
1528 extern void unlock_super(struct super_block *);
1529
1530 /*
1531 * VFS helper functions..
1532 */
1533 extern int vfs_create(struct inode *, struct dentry *, umode_t, struct nameidata *);
1534 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1535 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1536 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1537 extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
1538 extern int vfs_rmdir(struct inode *, struct dentry *);
1539 extern int vfs_unlink(struct inode *, struct dentry *);
1540 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1541
1542 /*
1543 * VFS dentry helper functions.
1544 */
1545 extern void dentry_unhash(struct dentry *dentry);
1546
1547 /*
1548 * VFS file helper functions.
1549 */
1550 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1551 umode_t mode);
1552 /*
1553 * VFS FS_IOC_FIEMAP helper definitions.
1554 */
1555 struct fiemap_extent_info {
1556 unsigned int fi_flags; /* Flags as passed from user */
1557 unsigned int fi_extents_mapped; /* Number of mapped extents */
1558 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1559 struct fiemap_extent __user *fi_extents_start; /* Start of
1560 fiemap_extent array */
1561 };
1562 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1563 u64 phys, u64 len, u32 flags);
1564 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1565
1566 /*
1567 * File types
1568 *
1569 * NOTE! These match bits 12..15 of stat.st_mode
1570 * (ie "(i_mode >> 12) & 15").
1571 */
1572 #define DT_UNKNOWN 0
1573 #define DT_FIFO 1
1574 #define DT_CHR 2
1575 #define DT_DIR 4
1576 #define DT_BLK 6
1577 #define DT_REG 8
1578 #define DT_LNK 10
1579 #define DT_SOCK 12
1580 #define DT_WHT 14
1581
1582 /*
1583 * This is the "filldir" function type, used by readdir() to let
1584 * the kernel specify what kind of dirent layout it wants to have.
1585 * This allows the kernel to read directories into kernel space or
1586 * to have different dirent layouts depending on the binary type.
1587 */
1588 typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
1589 struct block_device_operations;
1590
1591 /* These macros are for out of kernel modules to test that
1592 * the kernel supports the unlocked_ioctl and compat_ioctl
1593 * fields in struct file_operations. */
1594 #define HAVE_COMPAT_IOCTL 1
1595 #define HAVE_UNLOCKED_IOCTL 1
1596
1597 struct file_operations {
1598 struct module *owner;
1599 loff_t (*llseek) (struct file *, loff_t, int);
1600 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1601 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1602 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1603 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1604 int (*readdir) (struct file *, void *, filldir_t);
1605 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1606 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1607 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1608 int (*mmap) (struct file *, struct vm_area_struct *);
1609 int (*open) (struct inode *, struct file *);
1610 int (*flush) (struct file *, fl_owner_t id);
1611 int (*release) (struct inode *, struct file *);
1612 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1613 int (*aio_fsync) (struct kiocb *, int datasync);
1614 int (*fasync) (int, struct file *, int);
1615 int (*lock) (struct file *, int, struct file_lock *);
1616 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1617 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1618 int (*check_flags)(int);
1619 int (*flock) (struct file *, int, struct file_lock *);
1620 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1621 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1622 int (*setlease)(struct file *, long, struct file_lock **);
1623 long (*fallocate)(struct file *file, int mode, loff_t offset,
1624 loff_t len);
1625 };
1626
1627 struct inode_operations {
1628 struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
1629 void * (*follow_link) (struct dentry *, struct nameidata *);
1630 int (*permission) (struct inode *, int);
1631 struct posix_acl * (*get_acl)(struct inode *, int);
1632
1633 int (*readlink) (struct dentry *, char __user *,int);
1634 void (*put_link) (struct dentry *, struct nameidata *, void *);
1635
1636 int (*create) (struct inode *,struct dentry *,umode_t,struct nameidata *);
1637 int (*link) (struct dentry *,struct inode *,struct dentry *);
1638 int (*unlink) (struct inode *,struct dentry *);
1639 int (*symlink) (struct inode *,struct dentry *,const char *);
1640 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1641 int (*rmdir) (struct inode *,struct dentry *);
1642 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1643 int (*rename) (struct inode *, struct dentry *,
1644 struct inode *, struct dentry *);
1645 void (*truncate) (struct inode *);
1646 int (*setattr) (struct dentry *, struct iattr *);
1647 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1648 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1649 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1650 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1651 int (*removexattr) (struct dentry *, const char *);
1652 void (*truncate_range)(struct inode *, loff_t, loff_t);
1653 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1654 u64 len);
1655 } ____cacheline_aligned;
1656
1657 struct seq_file;
1658
1659 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1660 unsigned long nr_segs, unsigned long fast_segs,
1661 struct iovec *fast_pointer,
1662 struct iovec **ret_pointer,
1663 int check_access);
1664
1665 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1666 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1667 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1668 unsigned long, loff_t *);
1669 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1670 unsigned long, loff_t *);
1671
1672 struct super_operations {
1673 struct inode *(*alloc_inode)(struct super_block *sb);
1674 void (*destroy_inode)(struct inode *);
1675
1676 void (*dirty_inode) (struct inode *, int flags);
1677 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1678 int (*drop_inode) (struct inode *);
1679 void (*evict_inode) (struct inode *);
1680 void (*put_super) (struct super_block *);
1681 void (*write_super) (struct super_block *);
1682 int (*sync_fs)(struct super_block *sb, int wait);
1683 int (*freeze_fs) (struct super_block *);
1684 int (*unfreeze_fs) (struct super_block *);
1685 int (*statfs) (struct dentry *, struct kstatfs *);
1686 int (*remount_fs) (struct super_block *, int *, char *);
1687 void (*umount_begin) (struct super_block *);
1688
1689 int (*show_options)(struct seq_file *, struct dentry *);
1690 int (*show_devname)(struct seq_file *, struct dentry *);
1691 int (*show_path)(struct seq_file *, struct dentry *);
1692 int (*show_stats)(struct seq_file *, struct dentry *);
1693 #ifdef CONFIG_QUOTA
1694 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1695 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1696 #endif
1697 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1698 int (*nr_cached_objects)(struct super_block *);
1699 void (*free_cached_objects)(struct super_block *, int);
1700 };
1701
1702 /*
1703 * Inode state bits. Protected by inode->i_lock
1704 *
1705 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1706 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1707 *
1708 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1709 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1710 * various stages of removing an inode.
1711 *
1712 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1713 *
1714 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1715 * fdatasync(). i_atime is the usual cause.
1716 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1717 * these changes separately from I_DIRTY_SYNC so that we
1718 * don't have to write inode on fdatasync() when only
1719 * mtime has changed in it.
1720 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1721 * I_NEW Serves as both a mutex and completion notification.
1722 * New inodes set I_NEW. If two processes both create
1723 * the same inode, one of them will release its inode and
1724 * wait for I_NEW to be released before returning.
1725 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1726 * also cause waiting on I_NEW, without I_NEW actually
1727 * being set. find_inode() uses this to prevent returning
1728 * nearly-dead inodes.
1729 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1730 * is zero. I_FREEING must be set when I_WILL_FREE is
1731 * cleared.
1732 * I_FREEING Set when inode is about to be freed but still has dirty
1733 * pages or buffers attached or the inode itself is still
1734 * dirty.
1735 * I_CLEAR Added by end_writeback(). In this state the inode is clean
1736 * and can be destroyed. Inode keeps I_FREEING.
1737 *
1738 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1739 * prohibited for many purposes. iget() must wait for
1740 * the inode to be completely released, then create it
1741 * anew. Other functions will just ignore such inodes,
1742 * if appropriate. I_NEW is used for waiting.
1743 *
1744 * I_SYNC Synchonized write of dirty inode data. The bits is
1745 * set during data writeback, and cleared with a wakeup
1746 * on the bit address once it is done.
1747 *
1748 * I_REFERENCED Marks the inode as recently references on the LRU list.
1749 *
1750 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1751 *
1752 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1753 */
1754 #define I_DIRTY_SYNC (1 << 0)
1755 #define I_DIRTY_DATASYNC (1 << 1)
1756 #define I_DIRTY_PAGES (1 << 2)
1757 #define __I_NEW 3
1758 #define I_NEW (1 << __I_NEW)
1759 #define I_WILL_FREE (1 << 4)
1760 #define I_FREEING (1 << 5)
1761 #define I_CLEAR (1 << 6)
1762 #define __I_SYNC 7
1763 #define I_SYNC (1 << __I_SYNC)
1764 #define I_REFERENCED (1 << 8)
1765 #define __I_DIO_WAKEUP 9
1766 #define I_DIO_WAKEUP (1 << I_DIO_WAKEUP)
1767
1768 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1769
1770 extern void __mark_inode_dirty(struct inode *, int);
1771 static inline void mark_inode_dirty(struct inode *inode)
1772 {
1773 __mark_inode_dirty(inode, I_DIRTY);
1774 }
1775
1776 static inline void mark_inode_dirty_sync(struct inode *inode)
1777 {
1778 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1779 }
1780
1781 extern void inc_nlink(struct inode *inode);
1782 extern void drop_nlink(struct inode *inode);
1783 extern void clear_nlink(struct inode *inode);
1784 extern void set_nlink(struct inode *inode, unsigned int nlink);
1785
1786 static inline void inode_inc_link_count(struct inode *inode)
1787 {
1788 inc_nlink(inode);
1789 mark_inode_dirty(inode);
1790 }
1791
1792 static inline void inode_dec_link_count(struct inode *inode)
1793 {
1794 drop_nlink(inode);
1795 mark_inode_dirty(inode);
1796 }
1797
1798 /**
1799 * inode_inc_iversion - increments i_version
1800 * @inode: inode that need to be updated
1801 *
1802 * Every time the inode is modified, the i_version field will be incremented.
1803 * The filesystem has to be mounted with i_version flag
1804 */
1805
1806 static inline void inode_inc_iversion(struct inode *inode)
1807 {
1808 spin_lock(&inode->i_lock);
1809 inode->i_version++;
1810 spin_unlock(&inode->i_lock);
1811 }
1812
1813 extern void touch_atime(struct vfsmount *mnt, struct dentry *dentry);
1814 static inline void file_accessed(struct file *file)
1815 {
1816 if (!(file->f_flags & O_NOATIME))
1817 touch_atime(file->f_path.mnt, file->f_path.dentry);
1818 }
1819
1820 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1821 int sync_inode_metadata(struct inode *inode, int wait);
1822
1823 struct file_system_type {
1824 const char *name;
1825 int fs_flags;
1826 struct dentry *(*mount) (struct file_system_type *, int,
1827 const char *, void *);
1828 void (*kill_sb) (struct super_block *);
1829 struct module *owner;
1830 struct file_system_type * next;
1831 struct hlist_head fs_supers;
1832
1833 struct lock_class_key s_lock_key;
1834 struct lock_class_key s_umount_key;
1835 struct lock_class_key s_vfs_rename_key;
1836
1837 struct lock_class_key i_lock_key;
1838 struct lock_class_key i_mutex_key;
1839 struct lock_class_key i_mutex_dir_key;
1840 };
1841
1842 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1843 void *data, int (*fill_super)(struct super_block *, void *, int));
1844 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1845 int flags, const char *dev_name, void *data,
1846 int (*fill_super)(struct super_block *, void *, int));
1847 extern struct dentry *mount_single(struct file_system_type *fs_type,
1848 int flags, void *data,
1849 int (*fill_super)(struct super_block *, void *, int));
1850 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1851 int flags, void *data,
1852 int (*fill_super)(struct super_block *, void *, int));
1853 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
1854 void generic_shutdown_super(struct super_block *sb);
1855 void kill_block_super(struct super_block *sb);
1856 void kill_anon_super(struct super_block *sb);
1857 void kill_litter_super(struct super_block *sb);
1858 void deactivate_super(struct super_block *sb);
1859 void deactivate_locked_super(struct super_block *sb);
1860 int set_anon_super(struct super_block *s, void *data);
1861 int get_anon_bdev(dev_t *);
1862 void free_anon_bdev(dev_t);
1863 struct super_block *sget(struct file_system_type *type,
1864 int (*test)(struct super_block *,void *),
1865 int (*set)(struct super_block *,void *),
1866 void *data);
1867 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
1868 const struct super_operations *ops,
1869 const struct dentry_operations *dops,
1870 unsigned long);
1871
1872 static inline void sb_mark_dirty(struct super_block *sb)
1873 {
1874 sb->s_dirt = 1;
1875 }
1876 static inline void sb_mark_clean(struct super_block *sb)
1877 {
1878 sb->s_dirt = 0;
1879 }
1880 static inline int sb_is_dirty(struct super_block *sb)
1881 {
1882 return sb->s_dirt;
1883 }
1884
1885 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1886 #define fops_get(fops) \
1887 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1888 #define fops_put(fops) \
1889 do { if (fops) module_put((fops)->owner); } while(0)
1890
1891 extern int register_filesystem(struct file_system_type *);
1892 extern int unregister_filesystem(struct file_system_type *);
1893 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1894 #define kern_mount(type) kern_mount_data(type, NULL)
1895 extern void kern_unmount(struct vfsmount *mnt);
1896 extern int may_umount_tree(struct vfsmount *);
1897 extern int may_umount(struct vfsmount *);
1898 extern long do_mount(char *, char *, char *, unsigned long, void *);
1899 extern struct vfsmount *collect_mounts(struct path *);
1900 extern void drop_collected_mounts(struct vfsmount *);
1901 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1902 struct vfsmount *);
1903 extern int vfs_statfs(struct path *, struct kstatfs *);
1904 extern int user_statfs(const char __user *, struct kstatfs *);
1905 extern int fd_statfs(int, struct kstatfs *);
1906 extern int vfs_ustat(dev_t, struct kstatfs *);
1907 extern int freeze_super(struct super_block *super);
1908 extern int thaw_super(struct super_block *super);
1909 extern bool our_mnt(struct vfsmount *mnt);
1910
1911 extern int current_umask(void);
1912
1913 /* /sys/fs */
1914 extern struct kobject *fs_kobj;
1915
1916 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
1917 extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1918
1919 #define FLOCK_VERIFY_READ 1
1920 #define FLOCK_VERIFY_WRITE 2
1921
1922 #ifdef CONFIG_FILE_LOCKING
1923 extern int locks_mandatory_locked(struct inode *);
1924 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1925
1926 /*
1927 * Candidates for mandatory locking have the setgid bit set
1928 * but no group execute bit - an otherwise meaningless combination.
1929 */
1930
1931 static inline int __mandatory_lock(struct inode *ino)
1932 {
1933 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1934 }
1935
1936 /*
1937 * ... and these candidates should be on MS_MANDLOCK mounted fs,
1938 * otherwise these will be advisory locks
1939 */
1940
1941 static inline int mandatory_lock(struct inode *ino)
1942 {
1943 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1944 }
1945
1946 static inline int locks_verify_locked(struct inode *inode)
1947 {
1948 if (mandatory_lock(inode))
1949 return locks_mandatory_locked(inode);
1950 return 0;
1951 }
1952
1953 static inline int locks_verify_truncate(struct inode *inode,
1954 struct file *filp,
1955 loff_t size)
1956 {
1957 if (inode->i_flock && mandatory_lock(inode))
1958 return locks_mandatory_area(
1959 FLOCK_VERIFY_WRITE, inode, filp,
1960 size < inode->i_size ? size : inode->i_size,
1961 (size < inode->i_size ? inode->i_size - size
1962 : size - inode->i_size)
1963 );
1964 return 0;
1965 }
1966
1967 static inline int break_lease(struct inode *inode, unsigned int mode)
1968 {
1969 if (inode->i_flock)
1970 return __break_lease(inode, mode);
1971 return 0;
1972 }
1973 #else /* !CONFIG_FILE_LOCKING */
1974 static inline int locks_mandatory_locked(struct inode *inode)
1975 {
1976 return 0;
1977 }
1978
1979 static inline int locks_mandatory_area(int rw, struct inode *inode,
1980 struct file *filp, loff_t offset,
1981 size_t count)
1982 {
1983 return 0;
1984 }
1985
1986 static inline int __mandatory_lock(struct inode *inode)
1987 {
1988 return 0;
1989 }
1990
1991 static inline int mandatory_lock(struct inode *inode)
1992 {
1993 return 0;
1994 }
1995
1996 static inline int locks_verify_locked(struct inode *inode)
1997 {
1998 return 0;
1999 }
2000
2001 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2002 size_t size)
2003 {
2004 return 0;
2005 }
2006
2007 static inline int break_lease(struct inode *inode, unsigned int mode)
2008 {
2009 return 0;
2010 }
2011
2012 #endif /* CONFIG_FILE_LOCKING */
2013
2014 /* fs/open.c */
2015
2016 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2017 struct file *filp);
2018 extern int do_fallocate(struct file *file, int mode, loff_t offset,
2019 loff_t len);
2020 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2021 umode_t mode);
2022 extern struct file *filp_open(const char *, int, umode_t);
2023 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2024 const char *, int);
2025 extern struct file * dentry_open(struct dentry *, struct vfsmount *, int,
2026 const struct cred *);
2027 extern int filp_close(struct file *, fl_owner_t id);
2028 extern char * getname(const char __user *);
2029
2030 /* fs/ioctl.c */
2031
2032 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2033
2034 /* fs/dcache.c */
2035 extern void __init vfs_caches_init_early(void);
2036 extern void __init vfs_caches_init(unsigned long);
2037
2038 extern struct kmem_cache *names_cachep;
2039
2040 #define __getname_gfp(gfp) kmem_cache_alloc(names_cachep, (gfp))
2041 #define __getname() __getname_gfp(GFP_KERNEL)
2042 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2043 #ifndef CONFIG_AUDITSYSCALL
2044 #define putname(name) __putname(name)
2045 #else
2046 extern void putname(const char *name);
2047 #endif
2048
2049 #ifdef CONFIG_BLOCK
2050 extern int register_blkdev(unsigned int, const char *);
2051 extern void unregister_blkdev(unsigned int, const char *);
2052 extern struct block_device *bdget(dev_t);
2053 extern struct block_device *bdgrab(struct block_device *bdev);
2054 extern void bd_set_size(struct block_device *, loff_t size);
2055 extern void bd_forget(struct inode *inode);
2056 extern void bdput(struct block_device *);
2057 extern void invalidate_bdev(struct block_device *);
2058 extern int sync_blockdev(struct block_device *bdev);
2059 extern void kill_bdev(struct block_device *);
2060 extern struct super_block *freeze_bdev(struct block_device *);
2061 extern void emergency_thaw_all(void);
2062 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2063 extern int fsync_bdev(struct block_device *);
2064 #else
2065 static inline void bd_forget(struct inode *inode) {}
2066 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2067 static inline void kill_bdev(struct block_device *bdev) {}
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 #ifdef CONFIG_BLOCK
2161 /*
2162 * return READ, READA, or WRITE
2163 */
2164 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2165
2166 /*
2167 * return data direction, READ or WRITE
2168 */
2169 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2170
2171 extern void check_disk_size_change(struct gendisk *disk,
2172 struct block_device *bdev);
2173 extern int revalidate_disk(struct gendisk *);
2174 extern int check_disk_change(struct block_device *);
2175 extern int __invalidate_device(struct block_device *, bool);
2176 extern int invalidate_partition(struct gendisk *, int);
2177 #endif
2178 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2179 pgoff_t start, pgoff_t end);
2180
2181 static inline void invalidate_remote_inode(struct inode *inode)
2182 {
2183 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2184 S_ISLNK(inode->i_mode))
2185 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2186 }
2187 extern int invalidate_inode_pages2(struct address_space *mapping);
2188 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2189 pgoff_t start, pgoff_t end);
2190 extern int write_inode_now(struct inode *, int);
2191 extern int filemap_fdatawrite(struct address_space *);
2192 extern int filemap_flush(struct address_space *);
2193 extern int filemap_fdatawait(struct address_space *);
2194 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2195 loff_t lend);
2196 extern int filemap_write_and_wait(struct address_space *mapping);
2197 extern int filemap_write_and_wait_range(struct address_space *mapping,
2198 loff_t lstart, loff_t lend);
2199 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2200 loff_t start, loff_t end, int sync_mode);
2201 extern int filemap_fdatawrite_range(struct address_space *mapping,
2202 loff_t start, loff_t end);
2203
2204 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2205 int datasync);
2206 extern int vfs_fsync(struct file *file, int datasync);
2207 extern int generic_write_sync(struct file *file, loff_t pos, loff_t count);
2208 extern void sync_supers(void);
2209 extern void emergency_sync(void);
2210 extern void emergency_remount(void);
2211 #ifdef CONFIG_BLOCK
2212 extern sector_t bmap(struct inode *, sector_t);
2213 #endif
2214 extern int notify_change(struct dentry *, struct iattr *);
2215 extern int inode_permission(struct inode *, int);
2216 extern int generic_permission(struct inode *, int);
2217
2218 static inline bool execute_ok(struct inode *inode)
2219 {
2220 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2221 }
2222
2223 /*
2224 * get_write_access() gets write permission for a file.
2225 * put_write_access() releases this write permission.
2226 * This is used for regular files.
2227 * We cannot support write (and maybe mmap read-write shared) accesses and
2228 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2229 * can have the following values:
2230 * 0: no writers, no VM_DENYWRITE mappings
2231 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2232 * > 0: (i_writecount) users are writing to the file.
2233 *
2234 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2235 * except for the cases where we don't hold i_writecount yet. Then we need to
2236 * use {get,deny}_write_access() - these functions check the sign and refuse
2237 * to do the change if sign is wrong.
2238 */
2239 static inline int get_write_access(struct inode *inode)
2240 {
2241 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2242 }
2243 static inline int deny_write_access(struct file *file)
2244 {
2245 struct inode *inode = file->f_path.dentry->d_inode;
2246 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2247 }
2248 static inline void put_write_access(struct inode * inode)
2249 {
2250 atomic_dec(&inode->i_writecount);
2251 }
2252 static inline void allow_write_access(struct file *file)
2253 {
2254 if (file)
2255 atomic_inc(&file->f_path.dentry->d_inode->i_writecount);
2256 }
2257 #ifdef CONFIG_IMA
2258 static inline void i_readcount_dec(struct inode *inode)
2259 {
2260 BUG_ON(!atomic_read(&inode->i_readcount));
2261 atomic_dec(&inode->i_readcount);
2262 }
2263 static inline void i_readcount_inc(struct inode *inode)
2264 {
2265 atomic_inc(&inode->i_readcount);
2266 }
2267 #else
2268 static inline void i_readcount_dec(struct inode *inode)
2269 {
2270 return;
2271 }
2272 static inline void i_readcount_inc(struct inode *inode)
2273 {
2274 return;
2275 }
2276 #endif
2277 extern int do_pipe_flags(int *, int);
2278 extern struct file *create_read_pipe(struct file *f, int flags);
2279 extern struct file *create_write_pipe(int flags);
2280 extern void free_write_pipe(struct file *);
2281
2282 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2283 extern struct file * open_exec(const char *);
2284
2285 /* fs/dcache.c -- generic fs support functions */
2286 extern int is_subdir(struct dentry *, struct dentry *);
2287 extern int path_is_under(struct path *, struct path *);
2288 extern ino_t find_inode_number(struct dentry *, struct qstr *);
2289
2290 #include <linux/err.h>
2291
2292 /* needed for stackable file system support */
2293 extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
2294
2295 extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
2296
2297 extern int inode_init_always(struct super_block *, struct inode *);
2298 extern void inode_init_once(struct inode *);
2299 extern void address_space_init_once(struct address_space *mapping);
2300 extern void ihold(struct inode * inode);
2301 extern void iput(struct inode *);
2302 extern struct inode * igrab(struct inode *);
2303 extern ino_t iunique(struct super_block *, ino_t);
2304 extern int inode_needs_sync(struct inode *inode);
2305 extern int generic_delete_inode(struct inode *inode);
2306 extern int generic_drop_inode(struct inode *inode);
2307
2308 extern struct inode *ilookup5_nowait(struct super_block *sb,
2309 unsigned long hashval, int (*test)(struct inode *, void *),
2310 void *data);
2311 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2312 int (*test)(struct inode *, void *), void *data);
2313 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2314
2315 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2316 extern struct inode * iget_locked(struct super_block *, unsigned long);
2317 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2318 extern int insert_inode_locked(struct inode *);
2319 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2320 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2321 #else
2322 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2323 #endif
2324 extern void unlock_new_inode(struct inode *);
2325 extern unsigned int get_next_ino(void);
2326
2327 extern void __iget(struct inode * inode);
2328 extern void iget_failed(struct inode *);
2329 extern void end_writeback(struct inode *);
2330 extern void __destroy_inode(struct inode *);
2331 extern struct inode *new_inode_pseudo(struct super_block *sb);
2332 extern struct inode *new_inode(struct super_block *sb);
2333 extern void free_inode_nonrcu(struct inode *inode);
2334 extern int should_remove_suid(struct dentry *);
2335 extern int file_remove_suid(struct file *);
2336
2337 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2338 static inline void insert_inode_hash(struct inode *inode)
2339 {
2340 __insert_inode_hash(inode, inode->i_ino);
2341 }
2342
2343 extern void __remove_inode_hash(struct inode *);
2344 static inline void remove_inode_hash(struct inode *inode)
2345 {
2346 if (!inode_unhashed(inode))
2347 __remove_inode_hash(inode);
2348 }
2349
2350 extern void inode_sb_list_add(struct inode *inode);
2351
2352 #ifdef CONFIG_BLOCK
2353 extern void submit_bio(int, struct bio *);
2354 extern int bdev_read_only(struct block_device *);
2355 #endif
2356 extern int set_blocksize(struct block_device *, int);
2357 extern int sb_set_blocksize(struct super_block *, int);
2358 extern int sb_min_blocksize(struct super_block *, int);
2359
2360 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2361 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2362 extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
2363 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
2364 extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2365 extern ssize_t __generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long,
2366 loff_t *);
2367 extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2368 extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
2369 unsigned long *, loff_t, loff_t *, size_t, size_t);
2370 extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
2371 unsigned long, loff_t, loff_t *, size_t, ssize_t);
2372 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2373 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2374 extern int generic_segment_checks(const struct iovec *iov,
2375 unsigned long *nr_segs, size_t *count, int access_flags);
2376
2377 /* fs/block_dev.c */
2378 extern ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov,
2379 unsigned long nr_segs, loff_t pos);
2380 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2381 int datasync);
2382 extern void block_sync_page(struct page *page);
2383
2384 /* fs/splice.c */
2385 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2386 struct pipe_inode_info *, size_t, unsigned int);
2387 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2388 struct pipe_inode_info *, size_t, unsigned int);
2389 extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
2390 struct file *, loff_t *, size_t, unsigned int);
2391 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2392 struct file *out, loff_t *, size_t len, unsigned int flags);
2393 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2394 size_t len, unsigned int flags);
2395
2396 extern void
2397 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2398 extern loff_t noop_llseek(struct file *file, loff_t offset, int origin);
2399 extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
2400 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
2401 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2402 int origin, loff_t maxsize);
2403 extern int generic_file_open(struct inode * inode, struct file * filp);
2404 extern int nonseekable_open(struct inode * inode, struct file * filp);
2405
2406 #ifdef CONFIG_FS_XIP
2407 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2408 loff_t *ppos);
2409 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
2410 extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2411 size_t len, loff_t *ppos);
2412 extern int xip_truncate_page(struct address_space *mapping, loff_t from);
2413 #else
2414 static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2415 {
2416 return 0;
2417 }
2418 #endif
2419
2420 #ifdef CONFIG_BLOCK
2421 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2422 loff_t file_offset);
2423
2424 enum {
2425 /* need locking between buffered and direct access */
2426 DIO_LOCKING = 0x01,
2427
2428 /* filesystem does not support filling holes */
2429 DIO_SKIP_HOLES = 0x02,
2430 };
2431
2432 void dio_end_io(struct bio *bio, int error);
2433 void inode_dio_wait(struct inode *inode);
2434 void inode_dio_done(struct inode *inode);
2435
2436 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2437 struct block_device *bdev, const struct iovec *iov, loff_t offset,
2438 unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
2439 dio_submit_t submit_io, int flags);
2440
2441 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
2442 struct inode *inode, const struct iovec *iov, loff_t offset,
2443 unsigned long nr_segs, get_block_t get_block)
2444 {
2445 return __blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov,
2446 offset, nr_segs, get_block, NULL, NULL,
2447 DIO_LOCKING | DIO_SKIP_HOLES);
2448 }
2449 #else
2450 static inline void inode_dio_wait(struct inode *inode)
2451 {
2452 }
2453 #endif
2454
2455 extern const struct file_operations generic_ro_fops;
2456
2457 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2458
2459 extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
2460 extern int vfs_follow_link(struct nameidata *, const char *);
2461 extern int page_readlink(struct dentry *, char __user *, int);
2462 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2463 extern void page_put_link(struct dentry *, struct nameidata *, void *);
2464 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2465 int nofs);
2466 extern int page_symlink(struct inode *inode, const char *symname, int len);
2467 extern const struct inode_operations page_symlink_inode_operations;
2468 extern int generic_readlink(struct dentry *, char __user *, int);
2469 extern void generic_fillattr(struct inode *, struct kstat *);
2470 extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2471 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2472 void inode_add_bytes(struct inode *inode, loff_t bytes);
2473 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2474 loff_t inode_get_bytes(struct inode *inode);
2475 void inode_set_bytes(struct inode *inode, loff_t bytes);
2476
2477 extern int vfs_readdir(struct file *, filldir_t, void *);
2478
2479 extern int vfs_stat(const char __user *, struct kstat *);
2480 extern int vfs_lstat(const char __user *, struct kstat *);
2481 extern int vfs_fstat(unsigned int, struct kstat *);
2482 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2483
2484 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2485 unsigned long arg);
2486 extern int __generic_block_fiemap(struct inode *inode,
2487 struct fiemap_extent_info *fieinfo,
2488 loff_t start, loff_t len,
2489 get_block_t *get_block);
2490 extern int generic_block_fiemap(struct inode *inode,
2491 struct fiemap_extent_info *fieinfo, u64 start,
2492 u64 len, get_block_t *get_block);
2493
2494 extern void get_filesystem(struct file_system_type *fs);
2495 extern void put_filesystem(struct file_system_type *fs);
2496 extern struct file_system_type *get_fs_type(const char *name);
2497 extern struct super_block *get_super(struct block_device *);
2498 extern struct super_block *get_active_super(struct block_device *bdev);
2499 extern void drop_super(struct super_block *sb);
2500 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2501 extern void iterate_supers_type(struct file_system_type *,
2502 void (*)(struct super_block *, void *), void *);
2503
2504 extern int dcache_dir_open(struct inode *, struct file *);
2505 extern int dcache_dir_close(struct inode *, struct file *);
2506 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2507 extern int dcache_readdir(struct file *, void *, filldir_t);
2508 extern int simple_setattr(struct dentry *, struct iattr *);
2509 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2510 extern int simple_statfs(struct dentry *, struct kstatfs *);
2511 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2512 extern int simple_unlink(struct inode *, struct dentry *);
2513 extern int simple_rmdir(struct inode *, struct dentry *);
2514 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2515 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2516 extern int simple_empty(struct dentry *);
2517 extern int simple_readpage(struct file *file, struct page *page);
2518 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2519 loff_t pos, unsigned len, unsigned flags,
2520 struct page **pagep, void **fsdata);
2521 extern int simple_write_end(struct file *file, struct address_space *mapping,
2522 loff_t pos, unsigned len, unsigned copied,
2523 struct page *page, void *fsdata);
2524
2525 extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
2526 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2527 extern const struct file_operations simple_dir_operations;
2528 extern const struct inode_operations simple_dir_inode_operations;
2529 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2530 struct dentry *d_alloc_name(struct dentry *, const char *);
2531 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2532 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2533 extern void simple_release_fs(struct vfsmount **mount, int *count);
2534
2535 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2536 loff_t *ppos, const void *from, size_t available);
2537 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2538 const void __user *from, size_t count);
2539
2540 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2541
2542 extern int generic_check_addressable(unsigned, u64);
2543
2544 #ifdef CONFIG_MIGRATION
2545 extern int buffer_migrate_page(struct address_space *,
2546 struct page *, struct page *,
2547 enum migrate_mode);
2548 #else
2549 #define buffer_migrate_page NULL
2550 #endif
2551
2552 extern int inode_change_ok(const struct inode *, struct iattr *);
2553 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2554 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2555
2556 extern void file_update_time(struct file *file);
2557
2558 extern int generic_show_options(struct seq_file *m, struct dentry *root);
2559 extern void save_mount_options(struct super_block *sb, char *options);
2560 extern void replace_mount_options(struct super_block *sb, char *options);
2561
2562 static inline ino_t parent_ino(struct dentry *dentry)
2563 {
2564 ino_t res;
2565
2566 /*
2567 * Don't strictly need d_lock here? If the parent ino could change
2568 * then surely we'd have a deeper race in the caller?
2569 */
2570 spin_lock(&dentry->d_lock);
2571 res = dentry->d_parent->d_inode->i_ino;
2572 spin_unlock(&dentry->d_lock);
2573 return res;
2574 }
2575
2576 /* Transaction based IO helpers */
2577
2578 /*
2579 * An argresp is stored in an allocated page and holds the
2580 * size of the argument or response, along with its content
2581 */
2582 struct simple_transaction_argresp {
2583 ssize_t size;
2584 char data[0];
2585 };
2586
2587 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2588
2589 char *simple_transaction_get(struct file *file, const char __user *buf,
2590 size_t size);
2591 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2592 size_t size, loff_t *pos);
2593 int simple_transaction_release(struct inode *inode, struct file *file);
2594
2595 void simple_transaction_set(struct file *file, size_t n);
2596
2597 /*
2598 * simple attribute files
2599 *
2600 * These attributes behave similar to those in sysfs:
2601 *
2602 * Writing to an attribute immediately sets a value, an open file can be
2603 * written to multiple times.
2604 *
2605 * Reading from an attribute creates a buffer from the value that might get
2606 * read with multiple read calls. When the attribute has been read
2607 * completely, no further read calls are possible until the file is opened
2608 * again.
2609 *
2610 * All attributes contain a text representation of a numeric value
2611 * that are accessed with the get() and set() functions.
2612 */
2613 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2614 static int __fops ## _open(struct inode *inode, struct file *file) \
2615 { \
2616 __simple_attr_check_format(__fmt, 0ull); \
2617 return simple_attr_open(inode, file, __get, __set, __fmt); \
2618 } \
2619 static const struct file_operations __fops = { \
2620 .owner = THIS_MODULE, \
2621 .open = __fops ## _open, \
2622 .release = simple_attr_release, \
2623 .read = simple_attr_read, \
2624 .write = simple_attr_write, \
2625 .llseek = generic_file_llseek, \
2626 };
2627
2628 static inline __printf(1, 2)
2629 void __simple_attr_check_format(const char *fmt, ...)
2630 {
2631 /* don't do anything, just let the compiler check the arguments; */
2632 }
2633
2634 int simple_attr_open(struct inode *inode, struct file *file,
2635 int (*get)(void *, u64 *), int (*set)(void *, u64),
2636 const char *fmt);
2637 int simple_attr_release(struct inode *inode, struct file *file);
2638 ssize_t simple_attr_read(struct file *file, char __user *buf,
2639 size_t len, loff_t *ppos);
2640 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2641 size_t len, loff_t *ppos);
2642
2643 struct ctl_table;
2644 int proc_nr_files(struct ctl_table *table, int write,
2645 void __user *buffer, size_t *lenp, loff_t *ppos);
2646 int proc_nr_dentry(struct ctl_table *table, int write,
2647 void __user *buffer, size_t *lenp, loff_t *ppos);
2648 int proc_nr_inodes(struct ctl_table *table, int write,
2649 void __user *buffer, size_t *lenp, loff_t *ppos);
2650 int __init get_filesystem_list(char *buf);
2651
2652 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
2653 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
2654
2655 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2656 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2657 (flag & __FMODE_NONOTIFY)))
2658
2659 static inline int is_sxid(umode_t mode)
2660 {
2661 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
2662 }
2663
2664 static inline void inode_has_no_xattr(struct inode *inode)
2665 {
2666 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
2667 inode->i_flags |= S_NOSEC;
2668 }
2669
2670 #endif /* __KERNEL__ */
2671 #endif /* _LINUX_FS_H */
This page took 0.140621 seconds and 6 git commands to generate.