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