get rid of pointless checks for dentry->sb == NULL
[linux-2.6-block.git] / include / linux / fs.h
CommitLineData
1da177e4
LT
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
1da177e4
LT
9#include <linux/limits.h>
10#include <linux/ioctl.h>
7cc01581 11#include <linux/blk_types.h>
b31d42a5 12#include <linux/types.h>
1da177e4
LT
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
0ac1ee0b
TG
26#define INR_OPEN_CUR 1024 /* Initial setting for nfile rlimits */
27#define INR_OPEN_MAX 4096 /* Hard limit for nfile rlimits */
1da177e4
LT
28
29#define BLOCK_SIZE_BITS 10
30#define BLOCK_SIZE (1<<BLOCK_SIZE_BITS)
31
01bf466e
DH
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 */
1ae7075b 35#define SEEK_MAX SEEK_END
01bf466e 36
367a51a3 37struct fstrim_range {
3a3a1af3
LM
38 __u64 start;
39 __u64 len;
40 __u64 minlen;
367a51a3
LC
41};
42
1da177e4
LT
43/* And dynamically-tunable limits and defaults: */
44struct files_stat_struct {
518de9b3
ED
45 unsigned long nr_files; /* read only */
46 unsigned long nr_free_files; /* read only */
47 unsigned long max_files; /* tunable */
1da177e4 48};
1da177e4
LT
49
50struct inodes_stat_t {
51 int nr_inodes;
52 int nr_unused;
9e7bf24b 53 int dummy[5]; /* padding for sysctl ABI compatibility */
1da177e4 54};
1da177e4 55
1da177e4
LT
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
e6305c43 63#define MAY_ACCESS 16
a110343f 64#define MAY_OPEN 32
9cfcac81 65#define MAY_CHDIR 64
1fc0f78c 66#define MAY_NOT_BLOCK 128 /* called from RCU mode, don't block */
1da177e4 67
55ec8217
PT
68/*
69 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
70 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
71 */
72
fc9161e5 73/* file is open for reading */
19adf9c5 74#define FMODE_READ ((__force fmode_t)0x1)
fc9161e5 75/* file is open for writing */
19adf9c5 76#define FMODE_WRITE ((__force fmode_t)0x2)
fc9161e5 77/* file is seekable */
19adf9c5 78#define FMODE_LSEEK ((__force fmode_t)0x4)
55ec8217 79/* file can be accessed using pread */
19adf9c5 80#define FMODE_PREAD ((__force fmode_t)0x8)
55ec8217 81/* file can be accessed using pwrite */
19adf9c5 82#define FMODE_PWRITE ((__force fmode_t)0x10)
fc9161e5 83/* File is opened for execution with sys_execve / sys_uselib */
19adf9c5 84#define FMODE_EXEC ((__force fmode_t)0x20)
fc9161e5 85/* File is opened with O_NDELAY (only set for block devices) */
19adf9c5 86#define FMODE_NDELAY ((__force fmode_t)0x40)
fc9161e5 87/* File is opened with O_EXCL (only set for block devices) */
19adf9c5 88#define FMODE_EXCL ((__force fmode_t)0x80)
fc9161e5
CH
89/* File is opened using open(.., 3, ..) and is writeable only for ioctls
90 (specialy hack for floppy.c) */
19adf9c5 91#define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
b500531e 92
4d4be482
CH
93/*
94 * Don't update ctime and mtime.
95 *
96 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
97 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
98 */
19adf9c5 99#define FMODE_NOCMTIME ((__force fmode_t)0x800)
4d4be482 100
0141450f 101/* Expect random access pattern */
19adf9c5 102#define FMODE_RANDOM ((__force fmode_t)0x1000)
0141450f 103
4a3956c7
KH
104/* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
105#define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
106
1abf0c71
AV
107/* File is opened with O_PATH; almost nothing can be done with it */
108#define FMODE_PATH ((__force fmode_t)0x4000)
109
ecf081d1 110/* File was opened by fanotify and shouldn't generate fanotify events */
13bcbc00 111#define FMODE_NONOTIFY ((__force fmode_t)0x1000000)
ecf081d1 112
48e70bc1
JA
113/*
114 * The below are the various read and write types that we support. Some of
115 * them include behavioral modifiers that send information down to the
116 * block layer and IO scheduler. Terminology:
117 *
118 * The block layer uses device plugging to defer IO a little bit, in
119 * the hope that we will see more IO very shortly. This increases
120 * coalescing of adjacent IO and thus reduces the number of IOs we
121 * have to send to the device. It also allows for better queuing,
122 * if the IO isn't mergeable. If the caller is going to be waiting
123 * for the IO, then he must ensure that the device is unplugged so
124 * that the IO is dispatched to the driver.
125 *
126 * All IO is handled async in Linux. This is fine for background
127 * writes, but for reads or writes that someone waits for completion
128 * on, we want to notify the block layer and IO scheduler so that they
129 * know about it. That allows them to make better scheduling
130 * decisions. So when the below references 'sync' and 'async', it
131 * is referencing this priority hint.
132 *
133 * With that in mind, the available types are:
134 *
135 * READ A normal read operation. Device will be plugged.
136 * READ_SYNC A synchronous read. Device is not plugged, caller can
137 * immediately wait on this read without caring about
138 * unplugging.
139 * READA Used for read-ahead operations. Lower priority, and the
7cc01581 140 * block layer could (in theory) choose to ignore this
48e70bc1
JA
141 * request if it runs into resource problems.
142 * WRITE A normal async write. Device will be plugged.
721a9602 143 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
48e70bc1 144 * the hint that someone will be waiting on this IO
721a9602
JA
145 * shortly. The write equivalent of READ_SYNC.
146 * WRITE_ODIRECT Special case write for O_DIRECT only.
4fed947c
TH
147 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
148 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
149 * non-volatile media on completion.
150 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
151 * by a cache flush and data is guaranteed to be on
152 * non-volatile media on completion.
48e70bc1
JA
153 *
154 */
7cc01581
TH
155#define RW_MASK REQ_WRITE
156#define RWA_MASK REQ_RAHEAD
7b6d91da
CH
157
158#define READ 0
7cc01581
TH
159#define WRITE RW_MASK
160#define READA RWA_MASK
7b6d91da 161
721a9602 162#define READ_SYNC (READ | REQ_SYNC)
7b6d91da 163#define READ_META (READ | REQ_META)
721a9602
JA
164#define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
165#define WRITE_ODIRECT (WRITE | REQ_SYNC)
7b6d91da 166#define WRITE_META (WRITE | REQ_META)
721a9602
JA
167#define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
168#define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
169#define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
1da177e4
LT
170
171#define SEL_IN 1
172#define SEL_OUT 2
173#define SEL_EX 4
174
175/* public flags for file_system_type */
176#define FS_REQUIRES_DEV 1
177#define FS_BINARY_MOUNTDATA 2
79c0b2df 178#define FS_HAS_SUBTYPE 4
1da177e4 179#define FS_REVAL_DOT 16384 /* Check the paths ".", ".." for staleness */
349457cc
MF
180#define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move()
181 * during rename() internally.
182 */
183
1da177e4
LT
184/*
185 * These are the fs-independent mount-flags: up to 32 flags are supported
186 */
187#define MS_RDONLY 1 /* Mount read-only */
188#define MS_NOSUID 2 /* Ignore suid and sgid bits */
189#define MS_NODEV 4 /* Disallow access to device special files */
190#define MS_NOEXEC 8 /* Disallow program execution */
191#define MS_SYNCHRONOUS 16 /* Writes are synced at once */
192#define MS_REMOUNT 32 /* Alter flags of a mounted FS */
193#define MS_MANDLOCK 64 /* Allow mandatory locks on an FS */
194#define MS_DIRSYNC 128 /* Directory modifications are synchronous */
195#define MS_NOATIME 1024 /* Do not update access times. */
196#define MS_NODIRATIME 2048 /* Do not update directory access times */
197#define MS_BIND 4096
198#define MS_MOVE 8192
199#define MS_REC 16384
9b04c997
TT
200#define MS_VERBOSE 32768 /* War is peace. Verbosity is silence.
201 MS_VERBOSE is deprecated. */
202#define MS_SILENT 32768
bf066c7d 203#define MS_POSIXACL (1<<16) /* VFS does not apply the umask */
9676f0c6 204#define MS_UNBINDABLE (1<<17) /* change to unbindable */
07b20889 205#define MS_PRIVATE (1<<18) /* change to private */
a58b0eb8 206#define MS_SLAVE (1<<19) /* change to slave */
03e06e68 207#define MS_SHARED (1<<20) /* change to shared */
47ae32d6 208#define MS_RELATIME (1<<21) /* Update atime relative to mtime/ctime. */
8bf9725c 209#define MS_KERNMOUNT (1<<22) /* this is a kern_mount call */
7a224228 210#define MS_I_VERSION (1<<23) /* Update inode I_version field */
d0adde57 211#define MS_STRICTATIME (1<<24) /* Always perform atime updates */
9e1f1de0 212#define MS_NOSEC (1<<28)
7a4dec53 213#define MS_BORN (1<<29)
1da177e4
LT
214#define MS_ACTIVE (1<<30)
215#define MS_NOUSER (1<<31)
216
217/*
218 * Superblock flags that can be altered by MS_REMOUNT
219 */
08b9fe6b 220#define MS_RMT_MASK (MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK|MS_I_VERSION)
1da177e4
LT
221
222/*
223 * Old magic mount flag and mask
224 */
225#define MS_MGC_VAL 0xC0ED0000
226#define MS_MGC_MSK 0xffff0000
227
228/* Inode flags - they have nothing to superblock flags now */
229
230#define S_SYNC 1 /* Writes are synced at once */
231#define S_NOATIME 2 /* Do not update access times */
232#define S_APPEND 4 /* Append-only file */
233#define S_IMMUTABLE 8 /* Immutable file */
234#define S_DEAD 16 /* removed, but still open directory */
235#define S_NOQUOTA 32 /* Inode is not counted to quota */
236#define S_DIRSYNC 64 /* Directory modifications are synchronous */
237#define S_NOCMTIME 128 /* Do not update file c/mtime */
238#define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
239#define S_PRIVATE 512 /* Inode is fs-internal */
196f5181 240#define S_IMA 1024 /* Inode has an associated IMA struct */
9875cf80 241#define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
69b45732 242#define S_NOSEC 4096 /* no suid or xattr security attributes */
1da177e4
LT
243
244/*
245 * Note that nosuid etc flags are inode-specific: setting some file-system
246 * flags just means all the inodes inherit those flags by default. It might be
247 * possible to override it selectively if you really wanted to with some
248 * ioctl() that is not currently implemented.
249 *
250 * Exception: MS_RDONLY is always applied to the entire file system.
251 *
252 * Unfortunately, it is possible to change a filesystems flags with it mounted
253 * with files in use. This means that all of the inodes will not have their
254 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
255 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
256 */
257#define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))
258
259#define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
260#define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
261 ((inode)->i_flags & S_SYNC))
262#define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
263 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
264#define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
37756ced 265#define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
7a224228 266#define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1da177e4
LT
267
268#define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
269#define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
270#define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1da177e4
LT
271#define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
272
273#define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
274#define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
275#define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
276#define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
196f5181 277#define IS_IMA(inode) ((inode)->i_flags & S_IMA)
9875cf80 278#define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
69b45732 279#define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1da177e4
LT
280
281/* the read-only stuff doesn't really belong here, but any other place is
282 probably as bad and I don't want to create yet another include file. */
283
284#define BLKROSET _IO(0x12,93) /* set device read-only (0 = read-write) */
285#define BLKROGET _IO(0x12,94) /* get read-only status (0 = read_write) */
286#define BLKRRPART _IO(0x12,95) /* re-read partition table */
287#define BLKGETSIZE _IO(0x12,96) /* return device size /512 (long *arg) */
288#define BLKFLSBUF _IO(0x12,97) /* flush buffer cache */
289#define BLKRASET _IO(0x12,98) /* set read ahead for block device */
290#define BLKRAGET _IO(0x12,99) /* get current read ahead setting */
291#define BLKFRASET _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
292#define BLKFRAGET _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
293#define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
294#define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
295#define BLKSSZGET _IO(0x12,104)/* get block device sector size */
296#if 0
297#define BLKPG _IO(0x12,105)/* See blkpg.h */
298
299/* Some people are morons. Do not use sizeof! */
300
301#define BLKELVGET _IOR(0x12,106,size_t)/* elevator get */
302#define BLKELVSET _IOW(0x12,107,size_t)/* elevator set */
303/* This was here just to show that the number is taken -
304 probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
305#endif
306/* A jump here: 108-111 have been used for various private purposes. */
307#define BLKBSZGET _IOR(0x12,112,size_t)
308#define BLKBSZSET _IOW(0x12,113,size_t)
309#define BLKGETSIZE64 _IOR(0x12,114,size_t) /* return device size in bytes (u64 *arg) */
2056a782
JA
310#define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup)
311#define BLKTRACESTART _IO(0x12,116)
312#define BLKTRACESTOP _IO(0x12,117)
313#define BLKTRACETEARDOWN _IO(0x12,118)
d30a2605 314#define BLKDISCARD _IO(0x12,119)
ac481c20
MP
315#define BLKIOMIN _IO(0x12,120)
316#define BLKIOOPT _IO(0x12,121)
317#define BLKALIGNOFF _IO(0x12,122)
318#define BLKPBSZGET _IO(0x12,123)
98262f27 319#define BLKDISCARDZEROES _IO(0x12,124)
8d57a98c 320#define BLKSECDISCARD _IO(0x12,125)
1da177e4
LT
321
322#define BMAP_IOCTL 1 /* obsolete - kept for compatibility */
323#define FIBMAP _IO(0x00,1) /* bmap access */
324#define FIGETBSZ _IO(0x00,2) /* get the block size used for bmap */
fcccf502
TS
325#define FIFREEZE _IOWR('X', 119, int) /* Freeze */
326#define FITHAW _IOWR('X', 120, int) /* Thaw */
367a51a3 327#define FITRIM _IOWR('X', 121, struct fstrim_range) /* Trim */
1da177e4 328
36695673
DH
329#define FS_IOC_GETFLAGS _IOR('f', 1, long)
330#define FS_IOC_SETFLAGS _IOW('f', 2, long)
331#define FS_IOC_GETVERSION _IOR('v', 1, long)
332#define FS_IOC_SETVERSION _IOW('v', 2, long)
c4b929b8 333#define FS_IOC_FIEMAP _IOWR('f', 11, struct fiemap)
36695673
DH
334#define FS_IOC32_GETFLAGS _IOR('f', 1, int)
335#define FS_IOC32_SETFLAGS _IOW('f', 2, int)
336#define FS_IOC32_GETVERSION _IOR('v', 1, int)
337#define FS_IOC32_SETVERSION _IOW('v', 2, int)
338
339/*
340 * Inode flags (FS_IOC_GETFLAGS / FS_IOC_SETFLAGS)
341 */
342#define FS_SECRM_FL 0x00000001 /* Secure deletion */
343#define FS_UNRM_FL 0x00000002 /* Undelete */
344#define FS_COMPR_FL 0x00000004 /* Compress file */
345#define FS_SYNC_FL 0x00000008 /* Synchronous updates */
346#define FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
347#define FS_APPEND_FL 0x00000020 /* writes to file may only append */
348#define FS_NODUMP_FL 0x00000040 /* do not dump file */
349#define FS_NOATIME_FL 0x00000080 /* do not update atime */
350/* Reserved for compression usage... */
351#define FS_DIRTY_FL 0x00000100
352#define FS_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
353#define FS_NOCOMP_FL 0x00000400 /* Don't compress */
354#define FS_ECOMPR_FL 0x00000800 /* Compression error */
355/* End compression flags --- maybe not all used */
356#define FS_BTREE_FL 0x00001000 /* btree format dir */
357#define FS_INDEX_FL 0x00001000 /* hash-indexed directory */
358#define FS_IMAGIC_FL 0x00002000 /* AFS directory */
359#define FS_JOURNAL_DATA_FL 0x00004000 /* Reserved for ext3 */
360#define FS_NOTAIL_FL 0x00008000 /* file tail should not be merged */
361#define FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
362#define FS_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
128e5eba
SW
363#define FS_EXTENT_FL 0x00080000 /* Extents */
364#define FS_DIRECTIO_FL 0x00100000 /* Use direct i/o */
32471f6e 365#define FS_NOCOW_FL 0x00800000 /* Do not cow file */
36695673
DH
366#define FS_RESERVED_FL 0x80000000 /* reserved for ext2 lib */
367
368#define FS_FL_USER_VISIBLE 0x0003DFFF /* User visible flags */
369#define FS_FL_USER_MODIFIABLE 0x000380FF /* User modifiable flags */
370
371
e46e4903
AM
372#define SYNC_FILE_RANGE_WAIT_BEFORE 1
373#define SYNC_FILE_RANGE_WRITE 2
374#define SYNC_FILE_RANGE_WAIT_AFTER 4
375
1da177e4
LT
376#ifdef __KERNEL__
377
378#include <linux/linkage.h>
379#include <linux/wait.h>
380#include <linux/types.h>
381#include <linux/kdev_t.h>
382#include <linux/dcache.h>
3f8206d4 383#include <linux/path.h>
1da177e4
LT
384#include <linux/stat.h>
385#include <linux/cache.h>
1da177e4
LT
386#include <linux/list.h>
387#include <linux/radix-tree.h>
388#include <linux/prio_tree.h>
389#include <linux/init.h>
914e2637 390#include <linux/pid.h>
1b1dcc1b 391#include <linux/mutex.h>
3bd858ab 392#include <linux/capability.h>
6188e10d 393#include <linux/semaphore.h>
c4b929b8 394#include <linux/fiemap.h>
ceb5bdc2 395#include <linux/rculist_bl.h>
07b8ce1e 396#include <linux/atomic.h>
1da177e4 397
1da177e4
LT
398#include <asm/byteorder.h>
399
a5694255 400struct export_operations;
a885c8c4 401struct hd_geometry;
1da177e4
LT
402struct iovec;
403struct nameidata;
92198f7e 404struct kiocb;
57cc7215 405struct kobject;
1da177e4
LT
406struct pipe_inode_info;
407struct poll_table_struct;
408struct kstatfs;
409struct vm_area_struct;
410struct vfsmount;
745ca247 411struct cred;
1da177e4 412
74bf17cf 413extern void __init inode_init(void);
1da177e4 414extern void __init inode_init_early(void);
1da177e4
LT
415extern void __init files_init(unsigned long);
416
dded4f4d 417extern struct files_stat_struct files_stat;
518de9b3 418extern unsigned long get_max_files(void);
dded4f4d
JE
419extern int sysctl_nr_open;
420extern struct inodes_stat_t inodes_stat;
421extern int leases_enable, lease_break_time;
dded4f4d 422
1da177e4
LT
423struct buffer_head;
424typedef int (get_block_t)(struct inode *inode, sector_t iblock,
425 struct buffer_head *bh_result, int create);
92198f7e 426typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
40e2e973
CH
427 ssize_t bytes, void *private, int ret,
428 bool is_async);
1da177e4
LT
429
430/*
431 * Attribute flags. These should be or-ed together to figure out what
432 * has been changed!
433 */
9767d749
MS
434#define ATTR_MODE (1 << 0)
435#define ATTR_UID (1 << 1)
436#define ATTR_GID (1 << 2)
437#define ATTR_SIZE (1 << 3)
438#define ATTR_ATIME (1 << 4)
439#define ATTR_MTIME (1 << 5)
440#define ATTR_CTIME (1 << 6)
441#define ATTR_ATIME_SET (1 << 7)
442#define ATTR_MTIME_SET (1 << 8)
443#define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
444#define ATTR_ATTR_FLAG (1 << 10)
445#define ATTR_KILL_SUID (1 << 11)
446#define ATTR_KILL_SGID (1 << 12)
447#define ATTR_FILE (1 << 13)
448#define ATTR_KILL_PRIV (1 << 14)
449#define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
450#define ATTR_TIMES_SET (1 << 16)
1da177e4
LT
451
452/*
453 * This is the Inode Attributes structure, used for notify_change(). It
454 * uses the above definitions as flags, to know which values have changed.
455 * Also, in this manner, a Filesystem can look at only the values it cares
456 * about. Basically, these are the attributes that the VFS layer can
457 * request to change from the FS layer.
458 *
459 * Derek Atkins <warlord@MIT.EDU> 94-10-20
460 */
461struct iattr {
462 unsigned int ia_valid;
463 umode_t ia_mode;
464 uid_t ia_uid;
465 gid_t ia_gid;
466 loff_t ia_size;
467 struct timespec ia_atime;
468 struct timespec ia_mtime;
469 struct timespec ia_ctime;
cc4e69de
MS
470
471 /*
25985edc 472 * Not an attribute, but an auxiliary info for filesystems wanting to
cc4e69de
MS
473 * implement an ftruncate() like method. NOTE: filesystem should
474 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
475 */
476 struct file *ia_file;
1da177e4
LT
477};
478
1da177e4
LT
479/*
480 * Includes for diskquotas.
481 */
482#include <linux/quota.h>
483
994fc28c
ZB
484/**
485 * enum positive_aop_returns - aop return codes with specific semantics
486 *
487 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
488 * completed, that the page is still locked, and
489 * should be considered active. The VM uses this hint
490 * to return the page to the active list -- it won't
491 * be a candidate for writeback again in the near
492 * future. Other callers must be careful to unlock
493 * the page if they get this return. Returned by
494 * writepage();
495 *
496 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
497 * unlocked it and the page might have been truncated.
498 * The caller should back up to acquiring a new page and
499 * trying again. The aop will be taking reasonable
500 * precautions not to livelock. If the caller held a page
501 * reference, it should drop it before retrying. Returned
55144768 502 * by readpage().
994fc28c
ZB
503 *
504 * address_space_operation functions return these large constants to indicate
505 * special semantics to the caller. These are much larger than the bytes in a
506 * page to allow for functions that return the number of bytes operated on in a
507 * given page.
508 */
509
510enum positive_aop_returns {
511 AOP_WRITEPAGE_ACTIVATE = 0x80000,
512 AOP_TRUNCATED_PAGE = 0x80001,
513};
514
afddba49 515#define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
89e10787 516#define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
54566b2c
NP
517#define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
518 * helper code (eg buffer layer)
519 * to clear GFP_FS from alloc */
afddba49 520
1da177e4
LT
521/*
522 * oh the beauties of C type declarations.
523 */
524struct page;
525struct address_space;
526struct writeback_control;
1da177e4 527
2f718ffc
NP
528struct iov_iter {
529 const struct iovec *iov;
530 unsigned long nr_segs;
531 size_t iov_offset;
532 size_t count;
533};
534
535size_t iov_iter_copy_from_user_atomic(struct page *page,
536 struct iov_iter *i, unsigned long offset, size_t bytes);
537size_t iov_iter_copy_from_user(struct page *page,
538 struct iov_iter *i, unsigned long offset, size_t bytes);
539void iov_iter_advance(struct iov_iter *i, size_t bytes);
afddba49 540int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes);
2f718ffc
NP
541size_t iov_iter_single_seg_count(struct iov_iter *i);
542
543static inline void iov_iter_init(struct iov_iter *i,
544 const struct iovec *iov, unsigned long nr_segs,
545 size_t count, size_t written)
546{
547 i->iov = iov;
548 i->nr_segs = nr_segs;
549 i->iov_offset = 0;
550 i->count = count + written;
551
552 iov_iter_advance(i, written);
553}
554
555static inline size_t iov_iter_count(struct iov_iter *i)
556{
557 return i->count;
558}
559
8ab22b9a
HH
560/*
561 * "descriptor" for what we're up to with a read.
562 * This allows us to use the same read code yet
563 * have multiple different users of the data that
564 * we read from a file.
565 *
566 * The simplest case just copies the data to user
567 * mode.
568 */
569typedef struct {
570 size_t written;
571 size_t count;
572 union {
573 char __user *buf;
574 void *data;
575 } arg;
576 int error;
577} read_descriptor_t;
578
579typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
580 unsigned long, unsigned long);
2f718ffc 581
1da177e4
LT
582struct address_space_operations {
583 int (*writepage)(struct page *page, struct writeback_control *wbc);
584 int (*readpage)(struct file *, struct page *);
1da177e4
LT
585
586 /* Write back some dirty pages from this mapping. */
587 int (*writepages)(struct address_space *, struct writeback_control *);
588
4741c9fd 589 /* Set a page dirty. Return true if this dirtied it */
1da177e4
LT
590 int (*set_page_dirty)(struct page *page);
591
592 int (*readpages)(struct file *filp, struct address_space *mapping,
593 struct list_head *pages, unsigned nr_pages);
594
afddba49
NP
595 int (*write_begin)(struct file *, struct address_space *mapping,
596 loff_t pos, unsigned len, unsigned flags,
597 struct page **pagep, void **fsdata);
598 int (*write_end)(struct file *, struct address_space *mapping,
599 loff_t pos, unsigned len, unsigned copied,
600 struct page *page, void *fsdata);
601
1da177e4
LT
602 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
603 sector_t (*bmap)(struct address_space *, sector_t);
2ff28e22 604 void (*invalidatepage) (struct page *, unsigned long);
27496a8c 605 int (*releasepage) (struct page *, gfp_t);
6072d13c 606 void (*freepage)(struct page *);
1da177e4
LT
607 ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
608 loff_t offset, unsigned long nr_segs);
70688e4d
NP
609 int (*get_xip_mem)(struct address_space *, pgoff_t, int,
610 void **, unsigned long *);
e965f963 611 /* migrate the contents of a page to the specified target */
2d1db3b1
CL
612 int (*migratepage) (struct address_space *,
613 struct page *, struct page *);
e3db7691 614 int (*launder_page) (struct page *);
8ab22b9a
HH
615 int (*is_partially_uptodate) (struct page *, read_descriptor_t *,
616 unsigned long);
25718736 617 int (*error_remove_page)(struct address_space *, struct page *);
1da177e4
LT
618};
619
7dcda1c9
JA
620extern const struct address_space_operations empty_aops;
621
afddba49
NP
622/*
623 * pagecache_write_begin/pagecache_write_end must be used by general code
624 * to write into the pagecache.
625 */
626int pagecache_write_begin(struct file *, struct address_space *mapping,
627 loff_t pos, unsigned len, unsigned flags,
628 struct page **pagep, void **fsdata);
629
630int pagecache_write_end(struct file *, struct address_space *mapping,
631 loff_t pos, unsigned len, unsigned copied,
632 struct page *page, void *fsdata);
633
1da177e4
LT
634struct backing_dev_info;
635struct address_space {
636 struct inode *host; /* owner: inode, block_device */
637 struct radix_tree_root page_tree; /* radix tree of all pages */
19fd6231 638 spinlock_t tree_lock; /* and lock protecting it */
1da177e4
LT
639 unsigned int i_mmap_writable;/* count VM_SHARED mappings */
640 struct prio_tree_root i_mmap; /* tree of private and shared mappings */
641 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
3d48ae45 642 struct mutex i_mmap_mutex; /* protect tree, count, list */
08142579 643 /* Protected by tree_lock together with the radix tree */
1da177e4
LT
644 unsigned long nrpages; /* number of total pages */
645 pgoff_t writeback_index;/* writeback starts here */
f5e54d6e 646 const struct address_space_operations *a_ops; /* methods */
1da177e4
LT
647 unsigned long flags; /* error bits/gfp mask */
648 struct backing_dev_info *backing_dev_info; /* device readahead, etc */
649 spinlock_t private_lock; /* for use by the address_space */
650 struct list_head private_list; /* ditto */
651 struct address_space *assoc_mapping; /* ditto */
652} __attribute__((aligned(sizeof(long))));
653 /*
654 * On most architectures that alignment is already the case; but
25985edc 655 * must be enforced here for CRIS, to let the least significant bit
1da177e4
LT
656 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
657 */
658
659struct block_device {
660 dev_t bd_dev; /* not a kdev_t - it's a search key */
f3ccfcda 661 int bd_openers;
1da177e4 662 struct inode * bd_inode; /* will die */
87d8fe1e 663 struct super_block * bd_super;
c039e313 664 struct mutex bd_mutex; /* open/close mutex */
1da177e4 665 struct list_head bd_inodes;
6b4517a7 666 void * bd_claiming;
1da177e4
LT
667 void * bd_holder;
668 int bd_holders;
77ea887e 669 bool bd_write_holder;
641dc636 670#ifdef CONFIG_SYSFS
49731baa 671 struct list_head bd_holder_disks;
641dc636 672#endif
1da177e4
LT
673 struct block_device * bd_contains;
674 unsigned bd_block_size;
675 struct hd_struct * bd_part;
676 /* number of times partitions within this device have been opened. */
677 unsigned bd_part_count;
678 int bd_invalidated;
679 struct gendisk * bd_disk;
680 struct list_head bd_list;
1da177e4
LT
681 /*
682 * Private data. You must have bd_claim'ed the block_device
683 * to use this. NOTE: bd_claim allows an owner to claim
684 * the same device multiple times, the owner must take special
685 * care to not mess up bd_private for that case.
686 */
687 unsigned long bd_private;
fcccf502
TS
688
689 /* The counter of freeze processes */
690 int bd_fsfreeze_count;
691 /* Mutex for freeze */
692 struct mutex bd_fsfreeze_mutex;
1da177e4
LT
693};
694
695/*
696 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
697 * radix trees
698 */
699#define PAGECACHE_TAG_DIRTY 0
700#define PAGECACHE_TAG_WRITEBACK 1
f446daae 701#define PAGECACHE_TAG_TOWRITE 2
1da177e4
LT
702
703int mapping_tagged(struct address_space *mapping, int tag);
704
705/*
706 * Might pages of this file be mapped into userspace?
707 */
708static inline int mapping_mapped(struct address_space *mapping)
709{
710 return !prio_tree_empty(&mapping->i_mmap) ||
711 !list_empty(&mapping->i_mmap_nonlinear);
712}
713
714/*
715 * Might pages of this file have been modified in userspace?
716 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
717 * marks vma as VM_SHARED if it is shared, and the file was opened for
718 * writing i.e. vma may be mprotected writable even if now readonly.
719 */
720static inline int mapping_writably_mapped(struct address_space *mapping)
721{
722 return mapping->i_mmap_writable != 0;
723}
724
725/*
726 * Use sequence counter to get consistent i_size on 32-bit processors.
727 */
728#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
729#include <linux/seqlock.h>
730#define __NEED_I_SIZE_ORDERED
731#define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
732#else
733#define i_size_ordered_init(inode) do { } while (0)
734#endif
735
f19d4a8f
AV
736struct posix_acl;
737#define ACL_NOT_CACHED ((void *)(-1))
738
1da177e4 739struct inode {
44a7d7a8
NP
740 /* RCU path lookup touches following: */
741 umode_t i_mode;
742 uid_t i_uid;
743 gid_t i_gid;
744 const struct inode_operations *i_op;
745 struct super_block *i_sb;
746
747 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
748 unsigned int i_flags;
79568f5b 749 unsigned long i_state;
13e12d14
LT
750#ifdef CONFIG_SECURITY
751 void *i_security;
752#endif
44a7d7a8
NP
753 struct mutex i_mutex;
754
13e12d14 755
44a7d7a8
NP
756 unsigned long dirtied_when; /* jiffies of first dirtying */
757
1da177e4 758 struct hlist_node i_hash;
7ccf19a8
NP
759 struct list_head i_wb_list; /* backing dev IO list */
760 struct list_head i_lru; /* inode LRU list */
1da177e4 761 struct list_head i_sb_list;
fa0d7e3d
NP
762 union {
763 struct list_head i_dentry;
764 struct rcu_head i_rcu;
765 };
1da177e4
LT
766 unsigned long i_ino;
767 atomic_t i_count;
1da177e4 768 unsigned int i_nlink;
1da177e4 769 dev_t i_rdev;
4e70af56 770 unsigned int i_blkbits;
7a224228 771 u64 i_version;
1da177e4 772 loff_t i_size;
83b7b44e
ED
773#ifdef __NEED_I_SIZE_ORDERED
774 seqcount_t i_size_seqcount;
775#endif
1da177e4
LT
776 struct timespec i_atime;
777 struct timespec i_mtime;
778 struct timespec i_ctime;
a0f62ac6 779 blkcnt_t i_blocks;
1da177e4 780 unsigned short i_bytes;
1da177e4 781 struct rw_semaphore i_alloc_sem;
99ac48f5 782 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
1da177e4
LT
783 struct file_lock *i_flock;
784 struct address_space *i_mapping;
785 struct address_space i_data;
786#ifdef CONFIG_QUOTA
787 struct dquot *i_dquot[MAXQUOTAS];
788#endif
1da177e4 789 struct list_head i_devices;
4c154168
TT
790 union {
791 struct pipe_inode_info *i_pipe;
eaf796e7 792 struct block_device *i_bdev;
577c4eb0 793 struct cdev *i_cdev;
4c154168 794 };
1da177e4
LT
795
796 __u32 i_generation;
797
3be25f49
EP
798#ifdef CONFIG_FSNOTIFY
799 __u32 i_fsnotify_mask; /* all events this inode cares about */
e61ce867 800 struct hlist_head i_fsnotify_marks;
0eeca283
RL
801#endif
802
a178d202 803#ifdef CONFIG_IMA
a68a27b6 804 atomic_t i_readcount; /* struct files open RO */
a178d202 805#endif
1da177e4 806 atomic_t i_writecount;
f19d4a8f
AV
807#ifdef CONFIG_FS_POSIX_ACL
808 struct posix_acl *i_acl;
809 struct posix_acl *i_default_acl;
50462062 810#endif
8e18e294 811 void *i_private; /* fs or device private pointer */
1da177e4
LT
812};
813
1d3382cb
AV
814static inline int inode_unhashed(struct inode *inode)
815{
816 return hlist_unhashed(&inode->i_hash);
817}
818
f2eace23
IM
819/*
820 * inode->i_mutex nesting subclasses for the lock validator:
821 *
822 * 0: the object of the current VFS operation
823 * 1: parent
824 * 2: child/target
825 * 3: quota file
826 *
827 * The locking order between these classes is
4df46240 828 * parent -> child -> normal -> xattr -> quota
f2eace23
IM
829 */
830enum inode_i_mutex_lock_class
831{
832 I_MUTEX_NORMAL,
833 I_MUTEX_PARENT,
834 I_MUTEX_CHILD,
4df46240 835 I_MUTEX_XATTR,
f2eace23
IM
836 I_MUTEX_QUOTA
837};
838
1da177e4
LT
839/*
840 * NOTE: in a 32bit arch with a preemptable kernel and
841 * an UP compile the i_size_read/write must be atomic
842 * with respect to the local cpu (unlike with preempt disabled),
843 * but they don't need to be atomic with respect to other cpus like in
844 * true SMP (so they need either to either locally disable irq around
845 * the read or for example on x86 they can be still implemented as a
846 * cmpxchg8b without the need of the lock prefix). For SMP compiles
847 * and 64bit archs it makes no difference if preempt is enabled or not.
848 */
48ed214d 849static inline loff_t i_size_read(const struct inode *inode)
1da177e4
LT
850{
851#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
852 loff_t i_size;
853 unsigned int seq;
854
855 do {
856 seq = read_seqcount_begin(&inode->i_size_seqcount);
857 i_size = inode->i_size;
858 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
859 return i_size;
860#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
861 loff_t i_size;
862
863 preempt_disable();
864 i_size = inode->i_size;
865 preempt_enable();
866 return i_size;
867#else
868 return inode->i_size;
869#endif
870}
871
7762f5a0
MS
872/*
873 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
874 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
875 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
876 */
1da177e4
LT
877static inline void i_size_write(struct inode *inode, loff_t i_size)
878{
879#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
880 write_seqcount_begin(&inode->i_size_seqcount);
881 inode->i_size = i_size;
882 write_seqcount_end(&inode->i_size_seqcount);
883#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
884 preempt_disable();
885 inode->i_size = i_size;
886 preempt_enable();
887#else
888 inode->i_size = i_size;
889#endif
890}
891
48ed214d 892static inline unsigned iminor(const struct inode *inode)
1da177e4
LT
893{
894 return MINOR(inode->i_rdev);
895}
896
48ed214d 897static inline unsigned imajor(const struct inode *inode)
1da177e4
LT
898{
899 return MAJOR(inode->i_rdev);
900}
901
902extern struct block_device *I_BDEV(struct inode *inode);
903
904struct fown_struct {
905 rwlock_t lock; /* protects pid, uid, euid fields */
609d7fa9
EB
906 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
907 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
1da177e4 908 uid_t uid, euid; /* uid/euid of process setting the owner */
1da177e4
LT
909 int signum; /* posix.1b rt signal to be delivered on IO */
910};
911
912/*
913 * Track a single file's readahead state
914 */
915struct file_ra_state {
937085aa
FW
916 pgoff_t start; /* where readahead started */
917 unsigned int size; /* # of readahead pages */
918 unsigned int async_size; /* do asynchronous readahead when
f9acc8c7 919 there are only # of pages ahead */
5ce1110b 920
937085aa 921 unsigned int ra_pages; /* Maximum readahead window */
1ebf26a9 922 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
f4e6b498 923 loff_t prev_pos; /* Cache last read() position */
1da177e4 924};
1da177e4 925
5ce1110b
FW
926/*
927 * Check if @index falls in the readahead windows.
928 */
929static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
930{
f9acc8c7
FW
931 return (index >= ra->start &&
932 index < ra->start + ra->size);
5ce1110b
FW
933}
934
ad775f5a
DH
935#define FILE_MNT_WRITE_TAKEN 1
936#define FILE_MNT_WRITE_RELEASED 2
937
1da177e4 938struct file {
2f512016
ED
939 /*
940 * fu_list becomes invalid after file_free is called and queued via
941 * fu_rcuhead for RCU freeing
942 */
943 union {
944 struct list_head fu_list;
945 struct rcu_head fu_rcuhead;
946 } f_u;
0f7fc9e4
JJS
947 struct path f_path;
948#define f_dentry f_path.dentry
949#define f_vfsmnt f_path.mnt
99ac48f5 950 const struct file_operations *f_op;
4a6a4499 951 spinlock_t f_lock; /* f_ep_links, f_flags, no IRQ */
6416ccb7
NP
952#ifdef CONFIG_SMP
953 int f_sb_list_cpu;
954#endif
516e0cc5 955 atomic_long_t f_count;
1da177e4 956 unsigned int f_flags;
aeb5d727 957 fmode_t f_mode;
1da177e4
LT
958 loff_t f_pos;
959 struct fown_struct f_owner;
d76b0d9b 960 const struct cred *f_cred;
1da177e4
LT
961 struct file_ra_state f_ra;
962
2b47c361 963 u64 f_version;
50462062 964#ifdef CONFIG_SECURITY
1da177e4 965 void *f_security;
50462062 966#endif
1da177e4
LT
967 /* needed for tty driver, and maybe others */
968 void *private_data;
969
970#ifdef CONFIG_EPOLL
971 /* Used by fs/eventpoll.c to link all the hooks to this file */
972 struct list_head f_ep_links;
1da177e4
LT
973#endif /* #ifdef CONFIG_EPOLL */
974 struct address_space *f_mapping;
ad775f5a
DH
975#ifdef CONFIG_DEBUG_WRITECOUNT
976 unsigned long f_mnt_write_state;
977#endif
1da177e4 978};
1da177e4 979
990d6c2d
AK
980struct file_handle {
981 __u32 handle_bytes;
982 int handle_type;
983 /* file identifier */
984 unsigned char f_handle[0];
985};
986
516e0cc5 987#define get_file(x) atomic_long_inc(&(x)->f_count)
d7065da0 988#define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
516e0cc5 989#define file_count(x) atomic_long_read(&(x)->f_count)
1da177e4 990
ad775f5a
DH
991#ifdef CONFIG_DEBUG_WRITECOUNT
992static inline void file_take_write(struct file *f)
993{
994 WARN_ON(f->f_mnt_write_state != 0);
995 f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN;
996}
997static inline void file_release_write(struct file *f)
998{
999 f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED;
1000}
1001static inline void file_reset_write(struct file *f)
1002{
1003 f->f_mnt_write_state = 0;
1004}
1005static inline void file_check_state(struct file *f)
1006{
1007 /*
1008 * At this point, either both or neither of these bits
1009 * should be set.
1010 */
1011 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN);
1012 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED);
1013}
1014static inline int file_check_writeable(struct file *f)
1015{
1016 if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN)
1017 return 0;
1018 printk(KERN_WARNING "writeable file with no "
1019 "mnt_want_write()\n");
1020 WARN_ON(1);
1021 return -EINVAL;
1022}
1023#else /* !CONFIG_DEBUG_WRITECOUNT */
1024static inline void file_take_write(struct file *filp) {}
1025static inline void file_release_write(struct file *filp) {}
1026static inline void file_reset_write(struct file *filp) {}
1027static inline void file_check_state(struct file *filp) {}
1028static inline int file_check_writeable(struct file *filp)
1029{
1030 return 0;
1031}
1032#endif /* CONFIG_DEBUG_WRITECOUNT */
1033
1da177e4
LT
1034#define MAX_NON_LFS ((1UL<<31) - 1)
1035
1036/* Page cache limit. The filesystems should put that into their s_maxbytes
1037 limits, otherwise bad things can happen in VM. */
1038#if BITS_PER_LONG==32
1039#define MAX_LFS_FILESIZE (((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
1040#elif BITS_PER_LONG==64
1041#define MAX_LFS_FILESIZE 0x7fffffffffffffffUL
1042#endif
1043
1044#define FL_POSIX 1
1045#define FL_FLOCK 2
1046#define FL_ACCESS 8 /* not trying to lock, just looking */
f475ae95 1047#define FL_EXISTS 16 /* when unlocking, test for existence */
1da177e4 1048#define FL_LEASE 32 /* lease held on this file */
75e1fcc0 1049#define FL_CLOSE 64 /* unlock on close */
1da177e4
LT
1050#define FL_SLEEP 128 /* A blocking lock */
1051
bde74e4b
MS
1052/*
1053 * Special return value from posix_lock_file() and vfs_lock_file() for
1054 * asynchronous locking.
1055 */
1056#define FILE_LOCK_DEFERRED 1
1057
1da177e4
LT
1058/*
1059 * The POSIX file lock owner is determined by
1060 * the "struct files_struct" in the thread group
1061 * (or NULL for no owner - BSD locks).
1062 *
1063 * Lockd stuffs a "host" pointer into this.
1064 */
1065typedef struct files_struct *fl_owner_t;
1066
1067struct file_lock_operations {
1da177e4
LT
1068 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1069 void (*fl_release_private)(struct file_lock *);
1070};
1071
1072struct lock_manager_operations {
1073 int (*fl_compare_owner)(struct file_lock *, struct file_lock *);
1074 void (*fl_notify)(struct file_lock *); /* unblock callback */
2beb6614 1075 int (*fl_grant)(struct file_lock *, struct file_lock *, int);
1da177e4
LT
1076 void (*fl_release_private)(struct file_lock *);
1077 void (*fl_break)(struct file_lock *);
1da177e4
LT
1078 int (*fl_change)(struct file_lock **, int);
1079};
1080
af558e33
BF
1081struct lock_manager {
1082 struct list_head list;
1083};
1084
1085void locks_start_grace(struct lock_manager *);
1086void locks_end_grace(struct lock_manager *);
1087int locks_in_grace(void);
1088
1da177e4
LT
1089/* that will die - we need it for nfs_lock_info */
1090#include <linux/nfs_fs_i.h>
1091
1092struct file_lock {
1093 struct file_lock *fl_next; /* singly linked list for this inode */
1094 struct list_head fl_link; /* doubly linked list of all locks */
1095 struct list_head fl_block; /* circular list of blocked processes */
1096 fl_owner_t fl_owner;
afc1246f
RK
1097 unsigned char fl_flags;
1098 unsigned char fl_type;
1da177e4 1099 unsigned int fl_pid;
ab1f1611 1100 struct pid *fl_nspid;
1da177e4
LT
1101 wait_queue_head_t fl_wait;
1102 struct file *fl_file;
1da177e4
LT
1103 loff_t fl_start;
1104 loff_t fl_end;
1105
1106 struct fasync_struct * fl_fasync; /* for lease break notifications */
1107 unsigned long fl_break_time; /* for nonblocking lease breaks */
1108
6aed6285 1109 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
7b021967 1110 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1da177e4
LT
1111 union {
1112 struct nfs_lock_info nfs_fl;
8d0a8a9d 1113 struct nfs4_lock_info nfs4_fl;
e8d6c554
DH
1114 struct {
1115 struct list_head link; /* link in AFS vnode's pending_locks list */
1116 int state; /* state of grant or error if -ve */
1117 } afs;
1da177e4
LT
1118 } fl_u;
1119};
1120
1121/* The following constant reflects the upper bound of the file/locking space */
1122#ifndef OFFSET_MAX
1123#define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1124#define OFFSET_MAX INT_LIMIT(loff_t)
1125#define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1126#endif
1127
1da177e4
LT
1128#include <linux/fcntl.h>
1129
bfcd17a6
TP
1130extern void send_sigio(struct fown_struct *fown, int fd, int band);
1131
bfcd17a6 1132#ifdef CONFIG_FILE_LOCKING
1da177e4 1133extern int fcntl_getlk(struct file *, struct flock __user *);
c293621b
PS
1134extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1135 struct flock __user *);
1da177e4
LT
1136
1137#if BITS_PER_LONG == 32
1138extern int fcntl_getlk64(struct file *, struct flock64 __user *);
c293621b
PS
1139extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1140 struct flock64 __user *);
1da177e4
LT
1141#endif
1142
1da177e4
LT
1143extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1144extern int fcntl_getlease(struct file *filp);
1145
1146/* fs/locks.c */
05fa3135 1147void locks_free_lock(struct file_lock *fl);
1da177e4 1148extern void locks_init_lock(struct file_lock *);
c5b1f0d9 1149extern struct file_lock * locks_alloc_lock(void);
1da177e4 1150extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1a747ee0 1151extern void __locks_copy_lock(struct file_lock *, const struct file_lock *);
1da177e4
LT
1152extern void locks_remove_posix(struct file *, fl_owner_t);
1153extern void locks_remove_flock(struct file *);
a9e61e25 1154extern void locks_release_private(struct file_lock *);
6d34ac19 1155extern void posix_test_lock(struct file *, struct file_lock *);
150b3934 1156extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1da177e4 1157extern int posix_lock_file_wait(struct file *, struct file_lock *);
64a318ee 1158extern int posix_unblock_lock(struct file *, struct file_lock *);
3ee17abd 1159extern int vfs_test_lock(struct file *, struct file_lock *);
150b3934 1160extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
9b9d2ab4 1161extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1da177e4
LT
1162extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
1163extern int __break_lease(struct inode *inode, unsigned int flags);
1164extern void lease_get_mtime(struct inode *, struct timespec *time);
0af1a450 1165extern int generic_setlease(struct file *, long, struct file_lock **);
a9933cea 1166extern int vfs_setlease(struct file *, long, struct file_lock **);
1da177e4
LT
1167extern int lease_modify(struct file_lock **, int);
1168extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
1169extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
b89f4321
AB
1170extern void lock_flocks(void);
1171extern void unlock_flocks(void);
bfcd17a6 1172#else /* !CONFIG_FILE_LOCKING */
c2aca5e5
SW
1173static inline int fcntl_getlk(struct file *file, struct flock __user *user)
1174{
1175 return -EINVAL;
1176}
1177
1178static inline int fcntl_setlk(unsigned int fd, struct file *file,
1179 unsigned int cmd, struct flock __user *user)
1180{
1181 return -EACCES;
1182}
1183
bfcd17a6 1184#if BITS_PER_LONG == 32
c2aca5e5
SW
1185static inline int fcntl_getlk64(struct file *file, struct flock64 __user *user)
1186{
1187 return -EINVAL;
1188}
1189
1190static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1191 unsigned int cmd, struct flock64 __user *user)
1192{
1193 return -EACCES;
1194}
bfcd17a6 1195#endif
c2aca5e5
SW
1196static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1197{
1198 return 0;
1199}
1200
1201static inline int fcntl_getlease(struct file *filp)
1202{
1203 return 0;
1204}
1205
1206static inline void locks_init_lock(struct file_lock *fl)
1207{
1208 return;
1209}
1210
1211static inline void __locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1212{
1213 return;
1214}
1215
1216static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1217{
1218 return;
1219}
1220
1221static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1222{
1223 return;
1224}
1225
1226static inline void locks_remove_flock(struct file *filp)
1227{
1228 return;
1229}
1230
1231static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1232{
1233 return;
1234}
1235
1236static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1237 struct file_lock *conflock)
1238{
1239 return -ENOLCK;
1240}
1241
1242static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1243{
1244 return -ENOLCK;
1245}
1246
1247static inline int posix_unblock_lock(struct file *filp,
1248 struct file_lock *waiter)
1249{
1250 return -ENOENT;
1251}
1252
1253static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1254{
1255 return 0;
1256}
1257
1258static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1259 struct file_lock *fl, struct file_lock *conf)
1260{
1261 return -ENOLCK;
1262}
1263
1264static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1265{
1266 return 0;
1267}
1268
1269static inline int flock_lock_file_wait(struct file *filp,
1270 struct file_lock *request)
1271{
1272 return -ENOLCK;
1273}
1274
1275static inline int __break_lease(struct inode *inode, unsigned int mode)
1276{
1277 return 0;
1278}
1279
1280static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1281{
1282 return;
1283}
1284
1285static inline int generic_setlease(struct file *filp, long arg,
1286 struct file_lock **flp)
1287{
1288 return -EINVAL;
1289}
1290
1291static inline int vfs_setlease(struct file *filp, long arg,
1292 struct file_lock **lease)
1293{
1294 return -EINVAL;
1295}
1296
1297static inline int lease_modify(struct file_lock **before, int arg)
1298{
1299 return -EINVAL;
1300}
1301
1302static inline int lock_may_read(struct inode *inode, loff_t start,
1303 unsigned long len)
1304{
1305 return 1;
1306}
1307
1308static inline int lock_may_write(struct inode *inode, loff_t start,
1309 unsigned long len)
1310{
1311 return 1;
1312}
1313
b89f4321
AB
1314static inline void lock_flocks(void)
1315{
1316}
1317
1318static inline void unlock_flocks(void)
1319{
1320}
1321
bfcd17a6
TP
1322#endif /* !CONFIG_FILE_LOCKING */
1323
1da177e4
LT
1324
1325struct fasync_struct {
989a2979
ED
1326 spinlock_t fa_lock;
1327 int magic;
1328 int fa_fd;
1329 struct fasync_struct *fa_next; /* singly linked list */
1330 struct file *fa_file;
1331 struct rcu_head fa_rcu;
1da177e4
LT
1332};
1333
1334#define FASYNC_MAGIC 0x4601
1335
1336/* SMP safe fasync helpers: */
1337extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
f7347ce4
LT
1338extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1339extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1340extern struct fasync_struct *fasync_alloc(void);
1341extern void fasync_free(struct fasync_struct *);
1342
1da177e4
LT
1343/* can be called from interrupts */
1344extern void kill_fasync(struct fasync_struct **, int, int);
1da177e4 1345
609d7fa9 1346extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1da177e4
LT
1347extern int f_setown(struct file *filp, unsigned long arg, int force);
1348extern void f_delown(struct file *filp);
609d7fa9 1349extern pid_t f_getown(struct file *filp);
1da177e4
LT
1350extern int send_sigurg(struct fown_struct *fown);
1351
1352/*
1353 * Umount options
1354 */
1355
1356#define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1357#define MNT_DETACH 0x00000002 /* Just detach from the tree */
1358#define MNT_EXPIRE 0x00000004 /* Mark for expiry */
db1f05bb
MS
1359#define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1360#define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1da177e4
LT
1361
1362extern struct list_head super_blocks;
1363extern spinlock_t sb_lock;
1364
1da177e4
LT
1365struct super_block {
1366 struct list_head s_list; /* Keep this first */
1367 dev_t s_dev; /* search index; _not_ kdev_t */
1da177e4 1368 unsigned char s_dirt;
270ba5f7
RK
1369 unsigned char s_blocksize_bits;
1370 unsigned long s_blocksize;
42cb56ae 1371 loff_t s_maxbytes; /* Max file size */
1da177e4 1372 struct file_system_type *s_type;
ee9b6d61 1373 const struct super_operations *s_op;
61e225dc 1374 const struct dquot_operations *dq_op;
0d54b217 1375 const struct quotactl_ops *s_qcop;
39655164 1376 const struct export_operations *s_export_op;
1da177e4
LT
1377 unsigned long s_flags;
1378 unsigned long s_magic;
1379 struct dentry *s_root;
1380 struct rw_semaphore s_umount;
7892f2f4 1381 struct mutex s_lock;
1da177e4 1382 int s_count;
1da177e4 1383 atomic_t s_active;
50462062 1384#ifdef CONFIG_SECURITY
1da177e4 1385 void *s_security;
50462062 1386#endif
bb435453 1387 const struct xattr_handler **s_xattr;
1da177e4
LT
1388
1389 struct list_head s_inodes; /* all inodes */
ceb5bdc2 1390 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
6416ccb7
NP
1391#ifdef CONFIG_SMP
1392 struct list_head __percpu *s_files;
1393#else
1da177e4 1394 struct list_head s_files;
6416ccb7 1395#endif
b5c84bf6 1396 /* s_dentry_lru, s_nr_dentry_unused protected by dcache.c lru locks */
da3bbdd4
KM
1397 struct list_head s_dentry_lru; /* unused dentry lru */
1398 int s_nr_dentry_unused; /* # of dentry on lru */
1da177e4
LT
1399
1400 struct block_device *s_bdev;
32a88aa1 1401 struct backing_dev_info *s_bdi;
acaebfd8 1402 struct mtd_info *s_mtd;
1da177e4
LT
1403 struct list_head s_instances;
1404 struct quota_info s_dquot; /* Diskquota specific options */
1405
1406 int s_frozen;
1407 wait_queue_head_t s_wait_unfrozen;
1408
1409 char s_id[32]; /* Informational name */
0bba0169 1410 u8 s_uuid[16]; /* UUID */
1da177e4
LT
1411
1412 void *s_fs_info; /* Filesystem private info */
30c40d2c 1413 fmode_t s_mode;
1da177e4 1414
270ba5f7
RK
1415 /* Granularity of c/m/atime in ns.
1416 Cannot be worse than a second */
1417 u32 s_time_gran;
1418
1da177e4
LT
1419 /*
1420 * The next field is for VFS *only*. No filesystems have any business
1421 * even looking at it. You had been warned.
1422 */
a11f3a05 1423 struct mutex s_vfs_rename_mutex; /* Kludge */
1da177e4 1424
79c0b2df
MS
1425 /*
1426 * Filesystem subtype. If non-empty the filesystem type field
1427 * in /proc/mounts will be "type.subtype"
1428 */
1429 char *s_subtype;
b3b304a2
MS
1430
1431 /*
1432 * Saved mount options for lazy filesystems using
1433 * generic_show_options()
1434 */
4d2deb40 1435 char __rcu *s_options;
c8aebb0c 1436 const struct dentry_operations *s_d_op; /* default d_op for dentries */
9fdfdcf1
DM
1437
1438 /*
1439 * Saved pool identifier for cleancache (-1 means none)
1440 */
1441 int cleancache_poolid;
1da177e4
LT
1442};
1443
1444extern struct timespec current_fs_time(struct super_block *sb);
1445
1446/*
1447 * Snapshotting support.
1448 */
1449enum {
1450 SB_UNFROZEN = 0,
1451 SB_FREEZE_WRITE = 1,
1452 SB_FREEZE_TRANS = 2,
1453};
1454
1455#define vfs_check_frozen(sb, level) \
1456 wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
1457
914e2637
AV
1458#define get_fs_excl() atomic_inc(&current->fs_excl)
1459#define put_fs_excl() atomic_dec(&current->fs_excl)
1460#define has_fs_excl() atomic_read(&current->fs_excl)
22e2c507 1461
e795b717
SH
1462/*
1463 * until VFS tracks user namespaces for inodes, just make all files
1464 * belong to init_user_ns
1465 */
1466extern struct user_namespace init_user_ns;
1467#define inode_userns(inode) (&init_user_ns)
2e149670 1468extern bool inode_owner_or_capable(const struct inode *inode);
3bd858ab 1469
914e2637
AV
1470/* not quite ready to be deprecated, but... */
1471extern void lock_super(struct super_block *);
1472extern void unlock_super(struct super_block *);
1da177e4
LT
1473
1474/*
1475 * VFS helper functions..
1476 */
1477extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *);
1478extern int vfs_mkdir(struct inode *, struct dentry *, int);
1479extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t);
db2e747b 1480extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1da177e4
LT
1481extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
1482extern int vfs_rmdir(struct inode *, struct dentry *);
1483extern int vfs_unlink(struct inode *, struct dentry *);
1484extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1485
1486/*
1487 * VFS dentry helper functions.
1488 */
1489extern void dentry_unhash(struct dentry *dentry);
1490
8c744fb8
CH
1491/*
1492 * VFS file helper functions.
1493 */
a1bd120d
DM
1494extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1495 mode_t mode);
c4b929b8
MF
1496/*
1497 * VFS FS_IOC_FIEMAP helper definitions.
1498 */
1499struct fiemap_extent_info {
1500 unsigned int fi_flags; /* Flags as passed from user */
1501 unsigned int fi_extents_mapped; /* Number of mapped extents */
1502 unsigned int fi_extents_max; /* Size of fiemap_extent array */
ecf5632d
NK
1503 struct fiemap_extent __user *fi_extents_start; /* Start of
1504 fiemap_extent array */
c4b929b8
MF
1505};
1506int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1507 u64 phys, u64 len, u32 flags);
1508int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1509
1da177e4
LT
1510/*
1511 * File types
1512 *
1513 * NOTE! These match bits 12..15 of stat.st_mode
1514 * (ie "(i_mode >> 12) & 15").
1515 */
1516#define DT_UNKNOWN 0
1517#define DT_FIFO 1
1518#define DT_CHR 2
1519#define DT_DIR 4
1520#define DT_BLK 6
1521#define DT_REG 8
1522#define DT_LNK 10
1523#define DT_SOCK 12
1524#define DT_WHT 14
1525
1da177e4
LT
1526/*
1527 * This is the "filldir" function type, used by readdir() to let
1528 * the kernel specify what kind of dirent layout it wants to have.
1529 * This allows the kernel to read directories into kernel space or
1530 * to have different dirent layouts depending on the binary type.
1531 */
afefdbb2 1532typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
08f85851 1533struct block_device_operations;
1da177e4 1534
1da177e4
LT
1535/* These macros are for out of kernel modules to test that
1536 * the kernel supports the unlocked_ioctl and compat_ioctl
1537 * fields in struct file_operations. */
1538#define HAVE_COMPAT_IOCTL 1
1539#define HAVE_UNLOCKED_IOCTL 1
1540
1541/*
1542 * NOTE:
b19dd42f
AB
1543 * all file operations except setlease can be called without
1544 * the big kernel lock held in all filesystems.
1da177e4
LT
1545 */
1546struct file_operations {
1547 struct module *owner;
1548 loff_t (*llseek) (struct file *, loff_t, int);
1549 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1da177e4 1550 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
027445c3
BP
1551 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1552 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1da177e4
LT
1553 int (*readdir) (struct file *, void *, filldir_t);
1554 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1da177e4
LT
1555 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1556 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1557 int (*mmap) (struct file *, struct vm_area_struct *);
1558 int (*open) (struct inode *, struct file *);
75e1fcc0 1559 int (*flush) (struct file *, fl_owner_t id);
1da177e4 1560 int (*release) (struct inode *, struct file *);
7ea80859 1561 int (*fsync) (struct file *, int datasync);
1da177e4
LT
1562 int (*aio_fsync) (struct kiocb *, int datasync);
1563 int (*fasync) (int, struct file *, int);
1564 int (*lock) (struct file *, int, struct file_lock *);
1da177e4
LT
1565 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1566 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1567 int (*check_flags)(int);
1da177e4 1568 int (*flock) (struct file *, int, struct file_lock *);
cbb7e577
JA
1569 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1570 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
f9ffed26 1571 int (*setlease)(struct file *, long, struct file_lock **);
2fe17c10
CH
1572 long (*fallocate)(struct file *file, int mode, loff_t offset,
1573 loff_t len);
1da177e4
LT
1574};
1575
1576struct inode_operations {
1da177e4 1577 struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
44a7d7a8 1578 void * (*follow_link) (struct dentry *, struct nameidata *);
10556cb2 1579 int (*permission) (struct inode *, int);
7e40145e 1580 int (*check_acl)(struct inode *, int);
44a7d7a8
NP
1581
1582 int (*readlink) (struct dentry *, char __user *,int);
1583 void (*put_link) (struct dentry *, struct nameidata *, void *);
1584
1585 int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
1da177e4
LT
1586 int (*link) (struct dentry *,struct inode *,struct dentry *);
1587 int (*unlink) (struct inode *,struct dentry *);
1588 int (*symlink) (struct inode *,struct dentry *,const char *);
1589 int (*mkdir) (struct inode *,struct dentry *,int);
1590 int (*rmdir) (struct inode *,struct dentry *);
1591 int (*mknod) (struct inode *,struct dentry *,int,dev_t);
1592 int (*rename) (struct inode *, struct dentry *,
1593 struct inode *, struct dentry *);
1da177e4 1594 void (*truncate) (struct inode *);
1da177e4
LT
1595 int (*setattr) (struct dentry *, struct iattr *);
1596 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1597 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1598 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1599 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1600 int (*removexattr) (struct dentry *, const char *);
f6b3ec23 1601 void (*truncate_range)(struct inode *, loff_t, loff_t);
c4b929b8
MF
1602 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1603 u64 len);
44a7d7a8 1604} ____cacheline_aligned;
1da177e4
LT
1605
1606struct seq_file;
1607
eed4e51f
BP
1608ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1609 unsigned long nr_segs, unsigned long fast_segs,
1610 struct iovec *fast_pointer,
1611 struct iovec **ret_pointer);
1612
1da177e4
LT
1613extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1614extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1615extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1616 unsigned long, loff_t *);
1617extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1618 unsigned long, loff_t *);
1619
1da177e4
LT
1620struct super_operations {
1621 struct inode *(*alloc_inode)(struct super_block *sb);
1622 void (*destroy_inode)(struct inode *);
1623
aa385729 1624 void (*dirty_inode) (struct inode *, int flags);
a9185b41 1625 int (*write_inode) (struct inode *, struct writeback_control *wbc);
45321ac5 1626 int (*drop_inode) (struct inode *);
be7ce416 1627 void (*evict_inode) (struct inode *);
1da177e4
LT
1628 void (*put_super) (struct super_block *);
1629 void (*write_super) (struct super_block *);
1630 int (*sync_fs)(struct super_block *sb, int wait);
c4be0c1d
TS
1631 int (*freeze_fs) (struct super_block *);
1632 int (*unfreeze_fs) (struct super_block *);
726c3342 1633 int (*statfs) (struct dentry *, struct kstatfs *);
1da177e4 1634 int (*remount_fs) (struct super_block *, int *, char *);
42faad99 1635 void (*umount_begin) (struct super_block *);
1da177e4
LT
1636
1637 int (*show_options)(struct seq_file *, struct vfsmount *);
c7f404b4
AV
1638 int (*show_devname)(struct seq_file *, struct vfsmount *);
1639 int (*show_path)(struct seq_file *, struct vfsmount *);
b4629fe2 1640 int (*show_stats)(struct seq_file *, struct vfsmount *);
0e51a720 1641#ifdef CONFIG_QUOTA
1da177e4
LT
1642 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1643 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
0e51a720 1644#endif
87d8fe1e 1645 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1da177e4
LT
1646};
1647
1c0eeaf5 1648/*
250df6ed 1649 * Inode state bits. Protected by inode->i_lock
1c0eeaf5
JE
1650 *
1651 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1652 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1653 *
1654 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1655 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1656 * various stages of removing an inode.
1657 *
eaff8079 1658 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1c0eeaf5 1659 *
e7ca2d41
JE
1660 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1661 * fdatasync(). i_atime is the usual cause.
000cb48e
JK
1662 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1663 * these changes separately from I_DIRTY_SYNC so that we
1664 * don't have to write inode on fdatasync() when only
1665 * mtime has changed in it.
1c0eeaf5 1666 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
eaff8079
CH
1667 * I_NEW Serves as both a mutex and completion notification.
1668 * New inodes set I_NEW. If two processes both create
1669 * the same inode, one of them will release its inode and
1670 * wait for I_NEW to be released before returning.
1671 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1672 * also cause waiting on I_NEW, without I_NEW actually
1673 * being set. find_inode() uses this to prevent returning
1674 * nearly-dead inodes.
1c0eeaf5
JE
1675 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1676 * is zero. I_FREEING must be set when I_WILL_FREE is
1677 * cleared.
1678 * I_FREEING Set when inode is about to be freed but still has dirty
1679 * pages or buffers attached or the inode itself is still
1680 * dirty.
b57922d9 1681 * I_CLEAR Added by end_writeback(). In this state the inode is clean
a4ffdde6 1682 * and can be destroyed. Inode keeps I_FREEING.
1c0eeaf5
JE
1683 *
1684 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1685 * prohibited for many purposes. iget() must wait for
1686 * the inode to be completely released, then create it
1687 * anew. Other functions will just ignore such inodes,
eaff8079 1688 * if appropriate. I_NEW is used for waiting.
1c0eeaf5 1689 *
eaff8079
CH
1690 * I_SYNC Synchonized write of dirty inode data. The bits is
1691 * set during data writeback, and cleared with a wakeup
1692 * on the bit address once it is done.
1c0eeaf5 1693 *
1c0eeaf5 1694 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1c0eeaf5 1695 */
9e38d86f
NP
1696#define I_DIRTY_SYNC (1 << 0)
1697#define I_DIRTY_DATASYNC (1 << 1)
1698#define I_DIRTY_PAGES (1 << 2)
eaff8079
CH
1699#define __I_NEW 3
1700#define I_NEW (1 << __I_NEW)
9e38d86f
NP
1701#define I_WILL_FREE (1 << 4)
1702#define I_FREEING (1 << 5)
1703#define I_CLEAR (1 << 6)
eaff8079 1704#define __I_SYNC 7
1c0eeaf5 1705#define I_SYNC (1 << __I_SYNC)
9e38d86f 1706#define I_REFERENCED (1 << 8)
1da177e4
LT
1707
1708#define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1709
1710extern void __mark_inode_dirty(struct inode *, int);
1711static inline void mark_inode_dirty(struct inode *inode)
1712{
1713 __mark_inode_dirty(inode, I_DIRTY);
1714}
1715
1716static inline void mark_inode_dirty_sync(struct inode *inode)
1717{
1718 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1719}
1720
71c42157
DH
1721/**
1722 * inc_nlink - directly increment an inode's link count
1723 * @inode: inode
1724 *
1725 * This is a low-level filesystem helper to replace any
1726 * direct filesystem manipulation of i_nlink. Currently,
1727 * it is only here for parity with dec_nlink().
1728 */
d8c76e6f 1729static inline void inc_nlink(struct inode *inode)
66d21730
AD
1730{
1731 inode->i_nlink++;
d8c76e6f
DH
1732}
1733
1734static inline void inode_inc_link_count(struct inode *inode)
1735{
1736 inc_nlink(inode);
66d21730
AD
1737 mark_inode_dirty(inode);
1738}
1739
71c42157
DH
1740/**
1741 * drop_nlink - directly drop an inode's link count
1742 * @inode: inode
1743 *
1744 * This is a low-level filesystem helper to replace any
1745 * direct filesystem manipulation of i_nlink. In cases
1746 * where we are attempting to track writes to the
1747 * filesystem, a decrement to zero means an imminent
1748 * write when the file is truncated and actually unlinked
1749 * on the filesystem.
1750 */
9a53c3a7 1751static inline void drop_nlink(struct inode *inode)
66d21730
AD
1752{
1753 inode->i_nlink--;
9a53c3a7
DH
1754}
1755
71c42157
DH
1756/**
1757 * clear_nlink - directly zero an inode's link count
1758 * @inode: inode
1759 *
1760 * This is a low-level filesystem helper to replace any
1761 * direct filesystem manipulation of i_nlink. See
1762 * drop_nlink() for why we care about i_nlink hitting zero.
1763 */
ce71ec36
DH
1764static inline void clear_nlink(struct inode *inode)
1765{
1766 inode->i_nlink = 0;
1767}
1768
9a53c3a7
DH
1769static inline void inode_dec_link_count(struct inode *inode)
1770{
1771 drop_nlink(inode);
66d21730
AD
1772 mark_inode_dirty(inode);
1773}
1774
25ec56b5
JNC
1775/**
1776 * inode_inc_iversion - increments i_version
1777 * @inode: inode that need to be updated
1778 *
1779 * Every time the inode is modified, the i_version field will be incremented.
1780 * The filesystem has to be mounted with i_version flag
1781 */
1782
1783static inline void inode_inc_iversion(struct inode *inode)
1784{
1785 spin_lock(&inode->i_lock);
1786 inode->i_version++;
1787 spin_unlock(&inode->i_lock);
1788}
1789
869243a0 1790extern void touch_atime(struct vfsmount *mnt, struct dentry *dentry);
1da177e4
LT
1791static inline void file_accessed(struct file *file)
1792{
1793 if (!(file->f_flags & O_NOATIME))
0f7fc9e4 1794 touch_atime(file->f_path.mnt, file->f_path.dentry);
1da177e4
LT
1795}
1796
1797int sync_inode(struct inode *inode, struct writeback_control *wbc);
c3765016 1798int sync_inode_metadata(struct inode *inode, int wait);
1da177e4 1799
1da177e4
LT
1800struct file_system_type {
1801 const char *name;
1802 int fs_flags;
c96e41e9
AV
1803 struct dentry *(*mount) (struct file_system_type *, int,
1804 const char *, void *);
1da177e4
LT
1805 void (*kill_sb) (struct super_block *);
1806 struct module *owner;
1807 struct file_system_type * next;
1808 struct list_head fs_supers;
d475fd42 1809
cf516249 1810 struct lock_class_key s_lock_key;
897c6ff9 1811 struct lock_class_key s_umount_key;
51ee049e 1812 struct lock_class_key s_vfs_rename_key;
d475fd42
PZ
1813
1814 struct lock_class_key i_lock_key;
1815 struct lock_class_key i_mutex_key;
14358e6d 1816 struct lock_class_key i_mutex_dir_key;
d475fd42 1817 struct lock_class_key i_alloc_sem_key;
1da177e4
LT
1818};
1819
ceefda69
AV
1820extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1821 void *data, int (*fill_super)(struct super_block *, void *, int));
152a0836
AV
1822extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1823 int flags, const char *dev_name, void *data,
1824 int (*fill_super)(struct super_block *, void *, int));
fc14f2fe
AV
1825extern struct dentry *mount_single(struct file_system_type *fs_type,
1826 int flags, void *data,
1827 int (*fill_super)(struct super_block *, void *, int));
3c26ff6e
AV
1828extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1829 int flags, void *data,
1830 int (*fill_super)(struct super_block *, void *, int));
1da177e4
LT
1831void generic_shutdown_super(struct super_block *sb);
1832void kill_block_super(struct super_block *sb);
1833void kill_anon_super(struct super_block *sb);
1834void kill_litter_super(struct super_block *sb);
1835void deactivate_super(struct super_block *sb);
74dbbdd7 1836void deactivate_locked_super(struct super_block *sb);
1da177e4
LT
1837int set_anon_super(struct super_block *s, void *data);
1838struct super_block *sget(struct file_system_type *type,
1839 int (*test)(struct super_block *,void *),
1840 int (*set)(struct super_block *,void *),
1841 void *data);
51139ada 1842extern struct dentry *mount_pseudo(struct file_system_type *, char *,
c74a1cbb
AV
1843 const struct super_operations *ops,
1844 const struct dentry_operations *dops,
1845 unsigned long);
1da177e4 1846
140236b4
AB
1847static inline void sb_mark_dirty(struct super_block *sb)
1848{
1849 sb->s_dirt = 1;
1850}
1851static inline void sb_mark_clean(struct super_block *sb)
1852{
1853 sb->s_dirt = 0;
1854}
1855static inline int sb_is_dirty(struct super_block *sb)
1856{
1857 return sb->s_dirt;
1858}
1859
1da177e4
LT
1860/* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1861#define fops_get(fops) \
1862 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1863#define fops_put(fops) \
1864 do { if (fops) module_put((fops)->owner); } while(0)
1865
1866extern int register_filesystem(struct file_system_type *);
1867extern int unregister_filesystem(struct file_system_type *);
8bf9725c
PE
1868extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1869#define kern_mount(type) kern_mount_data(type, NULL)
1da177e4
LT
1870extern int may_umount_tree(struct vfsmount *);
1871extern int may_umount(struct vfsmount *);
1872extern long do_mount(char *, char *, char *, unsigned long, void *);
589ff870 1873extern struct vfsmount *collect_mounts(struct path *);
8aec0809 1874extern void drop_collected_mounts(struct vfsmount *);
1f707137
AV
1875extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1876 struct vfsmount *);
ebabe9a9 1877extern int vfs_statfs(struct path *, struct kstatfs *);
c8b91acc
AV
1878extern int user_statfs(const char __user *, struct kstatfs *);
1879extern int fd_statfs(int, struct kstatfs *);
ebabe9a9 1880extern int statfs_by_dentry(struct dentry *, struct kstatfs *);
18e9e510
JB
1881extern int freeze_super(struct super_block *super);
1882extern int thaw_super(struct super_block *super);
1da177e4 1883
ce3b0f8d
AV
1884extern int current_umask(void);
1885
f87fd4c2 1886/* /sys/fs */
00d26666 1887extern struct kobject *fs_kobj;
f87fd4c2 1888
435f49a5 1889#define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
bfcd17a6
TP
1890extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1891
1da177e4
LT
1892#define FLOCK_VERIFY_READ 1
1893#define FLOCK_VERIFY_WRITE 2
1894
bfcd17a6 1895#ifdef CONFIG_FILE_LOCKING
1da177e4
LT
1896extern int locks_mandatory_locked(struct inode *);
1897extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1898
1899/*
1900 * Candidates for mandatory locking have the setgid bit set
1901 * but no group execute bit - an otherwise meaningless combination.
1902 */
a16877ca
PE
1903
1904static inline int __mandatory_lock(struct inode *ino)
1905{
1906 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1907}
1908
1909/*
1910 * ... and these candidates should be on MS_MANDLOCK mounted fs,
1911 * otherwise these will be advisory locks
1912 */
1913
1914static inline int mandatory_lock(struct inode *ino)
1915{
1916 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1917}
1da177e4
LT
1918
1919static inline int locks_verify_locked(struct inode *inode)
1920{
a16877ca 1921 if (mandatory_lock(inode))
1da177e4
LT
1922 return locks_mandatory_locked(inode);
1923 return 0;
1924}
1925
1da177e4
LT
1926static inline int locks_verify_truncate(struct inode *inode,
1927 struct file *filp,
1928 loff_t size)
1929{
a16877ca 1930 if (inode->i_flock && mandatory_lock(inode))
1da177e4
LT
1931 return locks_mandatory_area(
1932 FLOCK_VERIFY_WRITE, inode, filp,
1933 size < inode->i_size ? size : inode->i_size,
1934 (size < inode->i_size ? inode->i_size - size
1935 : size - inode->i_size)
1936 );
1937 return 0;
1938}
1939
1940static inline int break_lease(struct inode *inode, unsigned int mode)
1941{
1942 if (inode->i_flock)
1943 return __break_lease(inode, mode);
1944 return 0;
1945}
bfcd17a6 1946#else /* !CONFIG_FILE_LOCKING */
af5df566
SW
1947static inline int locks_mandatory_locked(struct inode *inode)
1948{
1949 return 0;
1950}
1951
1952static inline int locks_mandatory_area(int rw, struct inode *inode,
1953 struct file *filp, loff_t offset,
1954 size_t count)
1955{
1956 return 0;
1957}
1958
1959static inline int __mandatory_lock(struct inode *inode)
1960{
1961 return 0;
1962}
1963
1964static inline int mandatory_lock(struct inode *inode)
1965{
1966 return 0;
1967}
1968
1969static inline int locks_verify_locked(struct inode *inode)
1970{
1971 return 0;
1972}
1973
1974static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
1975 size_t size)
1976{
1977 return 0;
1978}
1979
1980static inline int break_lease(struct inode *inode, unsigned int mode)
1981{
1982 return 0;
1983}
1984
bfcd17a6 1985#endif /* CONFIG_FILE_LOCKING */
1da177e4
LT
1986
1987/* fs/open.c */
1988
4a30131e
N
1989extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
1990 struct file *filp);
3e63cbb1
AJ
1991extern int do_fallocate(struct file *file, int mode, loff_t offset,
1992 loff_t len);
8e8a1407 1993extern long do_sys_open(int dfd, const char __user *filename, int flags,
5590ff0d 1994 int mode);
1da177e4 1995extern struct file *filp_open(const char *, int, int);
73d049a4
AV
1996extern struct file *file_open_root(struct dentry *, struct vfsmount *,
1997 const char *, int);
745ca247
DH
1998extern struct file * dentry_open(struct dentry *, struct vfsmount *, int,
1999 const struct cred *);
1da177e4
LT
2000extern int filp_close(struct file *, fl_owner_t id);
2001extern char * getname(const char __user *);
2002
3e63cbb1
AJ
2003/* fs/ioctl.c */
2004
2005extern int ioctl_preallocate(struct file *filp, void __user *argp);
2006
1da177e4
LT
2007/* fs/dcache.c */
2008extern void __init vfs_caches_init_early(void);
2009extern void __init vfs_caches_init(unsigned long);
2010
b86c089b
CL
2011extern struct kmem_cache *names_cachep;
2012
3446a8aa
VN
2013#define __getname_gfp(gfp) kmem_cache_alloc(names_cachep, (gfp))
2014#define __getname() __getname_gfp(GFP_KERNEL)
2015#define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
1da177e4
LT
2016#ifndef CONFIG_AUDITSYSCALL
2017#define putname(name) __putname(name)
2018#else
2019extern void putname(const char *name);
2020#endif
2021
9361401e 2022#ifdef CONFIG_BLOCK
1da177e4 2023extern int register_blkdev(unsigned int, const char *);
f4480240 2024extern void unregister_blkdev(unsigned int, const char *);
1da177e4 2025extern struct block_device *bdget(dev_t);
dddac6a7 2026extern struct block_device *bdgrab(struct block_device *bdev);
1da177e4
LT
2027extern void bd_set_size(struct block_device *, loff_t size);
2028extern void bd_forget(struct inode *inode);
2029extern void bdput(struct block_device *);
585d3bc0
NP
2030extern void invalidate_bdev(struct block_device *);
2031extern int sync_blockdev(struct block_device *bdev);
2032extern struct super_block *freeze_bdev(struct block_device *);
c2d75438 2033extern void emergency_thaw_all(void);
585d3bc0
NP
2034extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2035extern int fsync_bdev(struct block_device *);
9361401e
DH
2036#else
2037static inline void bd_forget(struct inode *inode) {}
47e4491b
AV
2038static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2039static inline void invalidate_bdev(struct block_device *bdev) {}
2040
2041static inline struct super_block *freeze_bdev(struct block_device *sb)
2042{
2043 return NULL;
2044}
2045
2046static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2047{
2048 return 0;
2049}
9361401e 2050#endif
60b0680f 2051extern int sync_filesystem(struct super_block *);
9361401e 2052extern const struct file_operations def_blk_fops;
4b6f5d20
AV
2053extern const struct file_operations def_chr_fops;
2054extern const struct file_operations bad_sock_fops;
2055extern const struct file_operations def_fifo_fops;
9361401e 2056#ifdef CONFIG_BLOCK
1da177e4 2057extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
56b26add 2058extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
1da177e4 2059extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
e525fd89 2060extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
d4d77629
TH
2061extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2062 void *holder);
2063extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2064 void *holder);
e525fd89 2065extern int blkdev_put(struct block_device *bdev, fmode_t mode);
641dc636 2066#ifdef CONFIG_SYSFS
e09b457b 2067extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
49731baa
TH
2068extern void bd_unlink_disk_holder(struct block_device *bdev,
2069 struct gendisk *disk);
641dc636 2070#else
e09b457b
TH
2071static inline int bd_link_disk_holder(struct block_device *bdev,
2072 struct gendisk *disk)
2073{
2074 return 0;
2075}
49731baa
TH
2076static inline void bd_unlink_disk_holder(struct block_device *bdev,
2077 struct gendisk *disk)
2078{
2079}
641dc636 2080#endif
9361401e 2081#endif
1da177e4
LT
2082
2083/* fs/char_dev.c */
68eef3b4 2084#define CHRDEV_MAJOR_HASH_SIZE 255
1da177e4
LT
2085extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2086extern int register_chrdev_region(dev_t, unsigned, const char *);
1905b1bf
TH
2087extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2088 unsigned int count, const char *name,
2089 const struct file_operations *fops);
2090extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2091 unsigned int count, const char *name);
1da177e4 2092extern void unregister_chrdev_region(dev_t, unsigned);
68eef3b4 2093extern void chrdev_show(struct seq_file *,off_t);
1da177e4 2094
1905b1bf
TH
2095static inline int register_chrdev(unsigned int major, const char *name,
2096 const struct file_operations *fops)
2097{
2098 return __register_chrdev(major, 0, 256, name, fops);
2099}
2100
2101static inline void unregister_chrdev(unsigned int major, const char *name)
2102{
2103 __unregister_chrdev(major, 0, 256, name);
2104}
2105
1da177e4
LT
2106/* fs/block_dev.c */
2107#define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
1f014290 2108#define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
9361401e
DH
2109
2110#ifdef CONFIG_BLOCK
2111#define BLKDEV_MAJOR_HASH_SIZE 255
1da177e4
LT
2112extern const char *__bdevname(dev_t, char *buffer);
2113extern const char *bdevname(struct block_device *bdev, char *buffer);
2114extern struct block_device *lookup_bdev(const char *);
68eef3b4 2115extern void blkdev_show(struct seq_file *,off_t);
30c40d2c 2116
9361401e
DH
2117#else
2118#define BLKDEV_MAJOR_HASH_SIZE 0
2119#endif
1da177e4
LT
2120
2121extern void init_special_inode(struct inode *, umode_t, dev_t);
2122
2123/* Invalid inode operations -- fs/bad_inode.c */
2124extern void make_bad_inode(struct inode *);
2125extern int is_bad_inode(struct inode *);
2126
d2d9648e
DV
2127extern const struct file_operations read_pipefifo_fops;
2128extern const struct file_operations write_pipefifo_fops;
2129extern const struct file_operations rdwr_pipefifo_fops;
1da177e4
LT
2130
2131extern int fs_may_remount_ro(struct super_block *);
2132
9361401e 2133#ifdef CONFIG_BLOCK
1da177e4
LT
2134/*
2135 * return READ, READA, or WRITE
2136 */
2137#define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2138
2139/*
2140 * return data direction, READ or WRITE
2141 */
2142#define bio_data_dir(bio) ((bio)->bi_rw & 1)
2143
c3279d14
AP
2144extern void check_disk_size_change(struct gendisk *disk,
2145 struct block_device *bdev);
0c002c2f 2146extern int revalidate_disk(struct gendisk *);
1da177e4 2147extern int check_disk_change(struct block_device *);
93b270f7 2148extern int __invalidate_device(struct block_device *, bool);
1da177e4 2149extern int invalidate_partition(struct gendisk *, int);
9361401e 2150#endif
1da177e4
LT
2151unsigned long invalidate_mapping_pages(struct address_space *mapping,
2152 pgoff_t start, pgoff_t end);
54bc4855 2153
1da177e4
LT
2154static inline void invalidate_remote_inode(struct inode *inode)
2155{
2156 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2157 S_ISLNK(inode->i_mode))
fc0ecff6 2158 invalidate_mapping_pages(inode->i_mapping, 0, -1);
1da177e4
LT
2159}
2160extern int invalidate_inode_pages2(struct address_space *mapping);
2161extern int invalidate_inode_pages2_range(struct address_space *mapping,
2162 pgoff_t start, pgoff_t end);
2163extern int write_inode_now(struct inode *, int);
2164extern int filemap_fdatawrite(struct address_space *);
2165extern int filemap_flush(struct address_space *);
2166extern int filemap_fdatawait(struct address_space *);
d3bccb6f
JK
2167extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2168 loff_t lend);
1da177e4
LT
2169extern int filemap_write_and_wait(struct address_space *mapping);
2170extern int filemap_write_and_wait_range(struct address_space *mapping,
2171 loff_t lstart, loff_t lend);
ebcf28e1
AM
2172extern int __filemap_fdatawrite_range(struct address_space *mapping,
2173 loff_t start, loff_t end, int sync_mode);
f4c0a0fd
JK
2174extern int filemap_fdatawrite_range(struct address_space *mapping,
2175 loff_t start, loff_t end);
ebcf28e1 2176
8018ab05
CH
2177extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2178 int datasync);
2179extern int vfs_fsync(struct file *file, int datasync);
148f948b 2180extern int generic_write_sync(struct file *file, loff_t pos, loff_t count);
1da177e4 2181extern void sync_supers(void);
1da177e4
LT
2182extern void emergency_sync(void);
2183extern void emergency_remount(void);
9361401e 2184#ifdef CONFIG_BLOCK
1da177e4 2185extern sector_t bmap(struct inode *, sector_t);
9361401e 2186#endif
1da177e4 2187extern int notify_change(struct dentry *, struct iattr *);
f419a2e3 2188extern int inode_permission(struct inode *, int);
2830ba7f 2189extern int generic_permission(struct inode *, int);
1da177e4 2190
f696a365
MS
2191static inline bool execute_ok(struct inode *inode)
2192{
2193 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2194}
2195
07b8ce1e
AV
2196/*
2197 * get_write_access() gets write permission for a file.
2198 * put_write_access() releases this write permission.
2199 * This is used for regular files.
2200 * We cannot support write (and maybe mmap read-write shared) accesses and
2201 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2202 * can have the following values:
2203 * 0: no writers, no VM_DENYWRITE mappings
2204 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2205 * > 0: (i_writecount) users are writing to the file.
2206 *
2207 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2208 * except for the cases where we don't hold i_writecount yet. Then we need to
2209 * use {get,deny}_write_access() - these functions check the sign and refuse
2210 * to do the change if sign is wrong.
2211 */
2212static inline int get_write_access(struct inode *inode)
2213{
2214 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2215}
2216static inline int deny_write_access(struct file *file)
2217{
2218 struct inode *inode = file->f_path.dentry->d_inode;
2219 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2220}
1da177e4
LT
2221static inline void put_write_access(struct inode * inode)
2222{
2223 atomic_dec(&inode->i_writecount);
2224}
2225static inline void allow_write_access(struct file *file)
2226{
2227 if (file)
0f7fc9e4 2228 atomic_inc(&file->f_path.dentry->d_inode->i_writecount);
1da177e4 2229}
a5c96ebf
MZ
2230#ifdef CONFIG_IMA
2231static inline void i_readcount_dec(struct inode *inode)
2232{
2233 BUG_ON(!atomic_read(&inode->i_readcount));
2234 atomic_dec(&inode->i_readcount);
2235}
2236static inline void i_readcount_inc(struct inode *inode)
2237{
2238 atomic_inc(&inode->i_readcount);
2239}
2240#else
2241static inline void i_readcount_dec(struct inode *inode)
2242{
2243 return;
2244}
2245static inline void i_readcount_inc(struct inode *inode)
2246{
2247 return;
2248}
2249#endif
ed8cae8b 2250extern int do_pipe_flags(int *, int);
be61a86d
UD
2251extern struct file *create_read_pipe(struct file *f, int flags);
2252extern struct file *create_write_pipe(int flags);
d6cbd281 2253extern void free_write_pipe(struct file *);
1da177e4 2254
6777d773 2255extern int kernel_read(struct file *, loff_t, char *, unsigned long);
1da177e4
LT
2256extern struct file * open_exec(const char *);
2257
2258/* fs/dcache.c -- generic fs support functions */
2259extern int is_subdir(struct dentry *, struct dentry *);
2096f759 2260extern int path_is_under(struct path *, struct path *);
1da177e4
LT
2261extern ino_t find_inode_number(struct dentry *, struct qstr *);
2262
2263#include <linux/err.h>
2264
2265/* needed for stackable file system support */
2266extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
2267
2268extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
2269
54e34621 2270extern int inode_init_always(struct super_block *, struct inode *);
1da177e4 2271extern void inode_init_once(struct inode *);
2aa15890 2272extern void address_space_init_once(struct address_space *mapping);
7de9c6ee 2273extern void ihold(struct inode * inode);
1da177e4
LT
2274extern void iput(struct inode *);
2275extern struct inode * igrab(struct inode *);
2276extern ino_t iunique(struct super_block *, ino_t);
2277extern int inode_needs_sync(struct inode *inode);
45321ac5
AV
2278extern int generic_delete_inode(struct inode *inode);
2279extern int generic_drop_inode(struct inode *inode);
1da177e4 2280
88bd5121
AA
2281extern struct inode *ilookup5_nowait(struct super_block *sb,
2282 unsigned long hashval, int (*test)(struct inode *, void *),
2283 void *data);
1da177e4
LT
2284extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2285 int (*test)(struct inode *, void *), void *data);
2286extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2287
2288extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2289extern struct inode * iget_locked(struct super_block *, unsigned long);
261bca86
AV
2290extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2291extern int insert_inode_locked(struct inode *);
1da177e4 2292extern void unlock_new_inode(struct inode *);
85fe4025 2293extern unsigned int get_next_ino(void);
1da177e4 2294
1da177e4 2295extern void __iget(struct inode * inode);
b46980fe 2296extern void iget_failed(struct inode *);
b0683aa6 2297extern void end_writeback(struct inode *);
2e00c97e 2298extern void __destroy_inode(struct inode *);
1da177e4 2299extern struct inode *new_inode(struct super_block *);
ff0c7d15 2300extern void free_inode_nonrcu(struct inode *inode);
01de85e0 2301extern int should_remove_suid(struct dentry *);
2f1936b8 2302extern int file_remove_suid(struct file *);
1da177e4
LT
2303
2304extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2305extern void remove_inode_hash(struct inode *);
646ec461
CH
2306static inline void insert_inode_hash(struct inode *inode)
2307{
1da177e4
LT
2308 __insert_inode_hash(inode, inode->i_ino);
2309}
646ec461 2310extern void inode_sb_list_add(struct inode *inode);
1da177e4 2311
9361401e 2312#ifdef CONFIG_BLOCK
1da177e4
LT
2313extern void submit_bio(int, struct bio *);
2314extern int bdev_read_only(struct block_device *);
9361401e 2315#endif
1da177e4
LT
2316extern int set_blocksize(struct block_device *, int);
2317extern int sb_set_blocksize(struct super_block *, int);
2318extern int sb_min_blocksize(struct super_block *, int);
2319
2320extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2321extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2322extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
1da177e4 2323int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
027445c3 2324extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
e4dd9de3
JK
2325extern ssize_t __generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long,
2326 loff_t *);
027445c3 2327extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
1da177e4
LT
2328extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
2329 unsigned long *, loff_t, loff_t *, size_t, size_t);
2330extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
2331 unsigned long, loff_t, loff_t *, size_t, ssize_t);
2332extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2333extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
0ceb3314
DM
2334extern int generic_segment_checks(const struct iovec *iov,
2335 unsigned long *nr_segs, size_t *count, int access_flags);
88e6faef 2336
eef99380
CH
2337/* fs/block_dev.c */
2338extern ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov,
2339 unsigned long nr_segs, loff_t pos);
7ea80859 2340extern int blkdev_fsync(struct file *filp, int datasync);
eef99380 2341
88e6faef 2342/* fs/splice.c */
cbb7e577 2343extern ssize_t generic_file_splice_read(struct file *, loff_t *,
88e6faef 2344 struct pipe_inode_info *, size_t, unsigned int);
6818173b
MS
2345extern ssize_t default_file_splice_read(struct file *, loff_t *,
2346 struct pipe_inode_info *, size_t, unsigned int);
88e6faef 2347extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
cbb7e577 2348 struct file *, loff_t *, size_t, unsigned int);
88e6faef 2349extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
cbb7e577
JA
2350 struct file *out, loff_t *, size_t len, unsigned int flags);
2351extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
88e6faef
AM
2352 size_t len, unsigned int flags);
2353
1da177e4
LT
2354extern void
2355file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
ae6afc3f 2356extern loff_t noop_llseek(struct file *file, loff_t offset, int origin);
1da177e4
LT
2357extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
2358extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
9465efc9
AK
2359extern loff_t generic_file_llseek_unlocked(struct file *file, loff_t offset,
2360 int origin);
1da177e4
LT
2361extern int generic_file_open(struct inode * inode, struct file * filp);
2362extern int nonseekable_open(struct inode * inode, struct file * filp);
2363
ceffc078 2364#ifdef CONFIG_FS_XIP
eb6fe0c3
CO
2365extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2366 loff_t *ppos);
ceffc078 2367extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
eb6fe0c3
CO
2368extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2369 size_t len, loff_t *ppos);
ceffc078 2370extern int xip_truncate_page(struct address_space *mapping, loff_t from);
6d79125b
CO
2371#else
2372static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2373{
2374 return 0;
2375}
ceffc078
CO
2376#endif
2377
9361401e 2378#ifdef CONFIG_BLOCK
facd07b0
JB
2379typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2380 loff_t file_offset);
1da177e4
LT
2381
2382enum {
1e431f5c
CH
2383 /* need locking between buffered and direct access */
2384 DIO_LOCKING = 0x01,
2385
2386 /* filesystem does not support filling holes */
2387 DIO_SKIP_HOLES = 0x02,
1da177e4
LT
2388};
2389
eafdc7d1
CH
2390void dio_end_io(struct bio *bio, int error);
2391
2392ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2393 struct block_device *bdev, const struct iovec *iov, loff_t offset,
2394 unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
2395 dio_submit_t submit_io, int flags);
7bb46a67 2396
1da177e4
LT
2397static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
2398 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1d8fa7a2 2399 loff_t offset, unsigned long nr_segs, get_block_t get_block,
1da177e4
LT
2400 dio_iodone_t end_io)
2401{
2402 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
facd07b0 2403 nr_segs, get_block, end_io, NULL,
1e431f5c 2404 DIO_LOCKING | DIO_SKIP_HOLES);
1da177e4 2405}
9361401e 2406#endif
1da177e4 2407
4b6f5d20 2408extern const struct file_operations generic_ro_fops;
1da177e4
LT
2409
2410#define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2411
2412extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
2413extern int vfs_follow_link(struct nameidata *, const char *);
2414extern int page_readlink(struct dentry *, char __user *, int);
cc314eef
LT
2415extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2416extern void page_put_link(struct dentry *, struct nameidata *, void *);
0adb25d2 2417extern int __page_symlink(struct inode *inode, const char *symname, int len,
54566b2c 2418 int nofs);
1da177e4 2419extern int page_symlink(struct inode *inode, const char *symname, int len);
c5ef1c42 2420extern const struct inode_operations page_symlink_inode_operations;
1da177e4
LT
2421extern int generic_readlink(struct dentry *, char __user *, int);
2422extern void generic_fillattr(struct inode *, struct kstat *);
2423extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
b462707e 2424void __inode_add_bytes(struct inode *inode, loff_t bytes);
1da177e4
LT
2425void inode_add_bytes(struct inode *inode, loff_t bytes);
2426void inode_sub_bytes(struct inode *inode, loff_t bytes);
2427loff_t inode_get_bytes(struct inode *inode);
2428void inode_set_bytes(struct inode *inode, loff_t bytes);
2429
2430extern int vfs_readdir(struct file *, filldir_t, void *);
2431
c7887325
DH
2432extern int vfs_stat(const char __user *, struct kstat *);
2433extern int vfs_lstat(const char __user *, struct kstat *);
1da177e4 2434extern int vfs_fstat(unsigned int, struct kstat *);
c7887325 2435extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
1da177e4 2436
deb21db7
EZ
2437extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2438 unsigned long arg);
e9079cce 2439extern int __generic_block_fiemap(struct inode *inode,
3a3076f4
JB
2440 struct fiemap_extent_info *fieinfo,
2441 loff_t start, loff_t len,
2442 get_block_t *get_block);
68c9d702
JB
2443extern int generic_block_fiemap(struct inode *inode,
2444 struct fiemap_extent_info *fieinfo, u64 start,
2445 u64 len, get_block_t *get_block);
1da177e4 2446
c18479fe
MS
2447extern void get_filesystem(struct file_system_type *fs);
2448extern void put_filesystem(struct file_system_type *fs);
1da177e4
LT
2449extern struct file_system_type *get_fs_type(const char *name);
2450extern struct super_block *get_super(struct block_device *);
4504230a 2451extern struct super_block *get_active_super(struct block_device *bdev);
1da177e4
LT
2452extern struct super_block *user_get_super(dev_t);
2453extern void drop_super(struct super_block *sb);
01a05b33 2454extern void iterate_supers(void (*)(struct super_block *, void *), void *);
43e15cdb
AV
2455extern void iterate_supers_type(struct file_system_type *,
2456 void (*)(struct super_block *, void *), void *);
1da177e4
LT
2457
2458extern int dcache_dir_open(struct inode *, struct file *);
2459extern int dcache_dir_close(struct inode *, struct file *);
2460extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2461extern int dcache_readdir(struct file *, void *, filldir_t);
7bb46a67 2462extern int simple_setattr(struct dentry *, struct iattr *);
1da177e4 2463extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
726c3342 2464extern int simple_statfs(struct dentry *, struct kstatfs *);
1da177e4
LT
2465extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2466extern int simple_unlink(struct inode *, struct dentry *);
2467extern int simple_rmdir(struct inode *, struct dentry *);
2468extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1b061d92 2469extern int noop_fsync(struct file *, int);
1da177e4
LT
2470extern int simple_empty(struct dentry *);
2471extern int simple_readpage(struct file *file, struct page *page);
afddba49
NP
2472extern int simple_write_begin(struct file *file, struct address_space *mapping,
2473 loff_t pos, unsigned len, unsigned flags,
2474 struct page **pagep, void **fsdata);
2475extern int simple_write_end(struct file *file, struct address_space *mapping,
2476 loff_t pos, unsigned len, unsigned copied,
2477 struct page *page, void *fsdata);
1da177e4
LT
2478
2479extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
2480extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
4b6f5d20 2481extern const struct file_operations simple_dir_operations;
c5ef1c42 2482extern const struct inode_operations simple_dir_inode_operations;
4b6f5d20 2483struct tree_descr { char *name; const struct file_operations *ops; int mode; };
1da177e4 2484struct dentry *d_alloc_name(struct dentry *, const char *);
7d683a09 2485extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
1f5ce9e9 2486extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
1da177e4
LT
2487extern void simple_release_fs(struct vfsmount **mount, int *count);
2488
93b07113
AM
2489extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2490 loff_t *ppos, const void *from, size_t available);
6a727b43
JS
2491extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2492 const void __user *from, size_t count);
1da177e4 2493
1b061d92 2494extern int generic_file_fsync(struct file *, int);
d5aacad5 2495
30ca22c7
PL
2496extern int generic_check_addressable(unsigned, u64);
2497
e965f963 2498#ifdef CONFIG_MIGRATION
2d1db3b1
CL
2499extern int buffer_migrate_page(struct address_space *,
2500 struct page *, struct page *);
e965f963
CL
2501#else
2502#define buffer_migrate_page NULL
2503#endif
2504
25d9e2d1 2505extern int inode_change_ok(const struct inode *, struct iattr *);
2506extern int inode_newsize_ok(const struct inode *, loff_t offset);
6a1a90ad 2507extern void setattr_copy(struct inode *inode, const struct iattr *attr);
1da177e4 2508
870f4817 2509extern void file_update_time(struct file *file);
1da177e4 2510
b3b304a2
MS
2511extern int generic_show_options(struct seq_file *m, struct vfsmount *mnt);
2512extern void save_mount_options(struct super_block *sb, char *options);
2a32cebd 2513extern void replace_mount_options(struct super_block *sb, char *options);
b3b304a2 2514
1da177e4
LT
2515static inline ino_t parent_ino(struct dentry *dentry)
2516{
2517 ino_t res;
2518
b5c84bf6
NP
2519 /*
2520 * Don't strictly need d_lock here? If the parent ino could change
2521 * then surely we'd have a deeper race in the caller?
2522 */
1da177e4
LT
2523 spin_lock(&dentry->d_lock);
2524 res = dentry->d_parent->d_inode->i_ino;
2525 spin_unlock(&dentry->d_lock);
2526 return res;
2527}
2528
1da177e4
LT
2529/* Transaction based IO helpers */
2530
2531/*
2532 * An argresp is stored in an allocated page and holds the
2533 * size of the argument or response, along with its content
2534 */
2535struct simple_transaction_argresp {
2536 ssize_t size;
2537 char data[0];
2538};
2539
2540#define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2541
2542char *simple_transaction_get(struct file *file, const char __user *buf,
2543 size_t size);
2544ssize_t simple_transaction_read(struct file *file, char __user *buf,
2545 size_t size, loff_t *pos);
2546int simple_transaction_release(struct inode *inode, struct file *file);
2547
76791ab2 2548void simple_transaction_set(struct file *file, size_t n);
1da177e4 2549
acaefc25
AB
2550/*
2551 * simple attribute files
2552 *
2553 * These attributes behave similar to those in sysfs:
2554 *
2555 * Writing to an attribute immediately sets a value, an open file can be
2556 * written to multiple times.
2557 *
2558 * Reading from an attribute creates a buffer from the value that might get
2559 * read with multiple read calls. When the attribute has been read
2560 * completely, no further read calls are possible until the file is opened
2561 * again.
2562 *
2563 * All attributes contain a text representation of a numeric value
2564 * that are accessed with the get() and set() functions.
2565 */
2566#define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2567static int __fops ## _open(struct inode *inode, struct file *file) \
2568{ \
2569 __simple_attr_check_format(__fmt, 0ull); \
2570 return simple_attr_open(inode, file, __get, __set, __fmt); \
2571} \
828c0950 2572static const struct file_operations __fops = { \
acaefc25
AB
2573 .owner = THIS_MODULE, \
2574 .open = __fops ## _open, \
74bedc4d 2575 .release = simple_attr_release, \
acaefc25
AB
2576 .read = simple_attr_read, \
2577 .write = simple_attr_write, \
1ec5584e 2578 .llseek = generic_file_llseek, \
acaefc25
AB
2579};
2580
2581static inline void __attribute__((format(printf, 1, 2)))
2582__simple_attr_check_format(const char *fmt, ...)
2583{
2584 /* don't do anything, just let the compiler check the arguments; */
2585}
2586
2587int simple_attr_open(struct inode *inode, struct file *file,
8b88b099 2588 int (*get)(void *, u64 *), int (*set)(void *, u64),
acaefc25 2589 const char *fmt);
74bedc4d 2590int simple_attr_release(struct inode *inode, struct file *file);
acaefc25
AB
2591ssize_t simple_attr_read(struct file *file, char __user *buf,
2592 size_t len, loff_t *ppos);
2593ssize_t simple_attr_write(struct file *file, const char __user *buf,
2594 size_t len, loff_t *ppos);
2595
4be28540 2596struct ctl_table;
8d65af78 2597int proc_nr_files(struct ctl_table *table, int write,
62239ac2 2598 void __user *buffer, size_t *lenp, loff_t *ppos);
312d3ca8
CH
2599int proc_nr_dentry(struct ctl_table *table, int write,
2600 void __user *buffer, size_t *lenp, loff_t *ppos);
cffbc8aa
DC
2601int proc_nr_inodes(struct ctl_table *table, int write,
2602 void __user *buffer, size_t *lenp, loff_t *ppos);
38e23c95 2603int __init get_filesystem_list(char *buf);
62239ac2 2604
3cd90ea4 2605#define __FMODE_EXEC ((__force int) FMODE_EXEC)
1a44bc8c
NK
2606#define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
2607
6d125529 2608#define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
ecf081d1 2609#define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
1a44bc8c 2610 (flag & __FMODE_NONOTIFY)))
5300990c 2611
69b45732
AK
2612static inline int is_sxid(mode_t mode)
2613{
2614 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
2615}
2616
2617static inline void inode_has_no_xattr(struct inode *inode)
2618{
9e1f1de0 2619 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
69b45732
AK
2620 inode->i_flags |= S_NOSEC;
2621}
2622
1da177e4
LT
2623#endif /* __KERNEL__ */
2624#endif /* _LINUX_FS_H */