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