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