blob: 460d2cc21ec60031c346b687c197114dfb8d8221 [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001#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
Linus Torvalds1da177e2005-04-16 15:20:36 -07009#include <linux/limits.h>
10#include <linux/ioctl.h>
Tejun Heo7cc01582010-08-03 13:14:58 +020011#include <linux/blk_types.h>
Ingo Molnarb31d42a2010-10-28 16:39:24 +020012#include <linux/types.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070013
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
Tim Gardner0ac1ee02011-05-24 17:13:05 -070026#define INR_OPEN_CUR 1024 /* Initial setting for nfile rlimits */
27#define INR_OPEN_MAX 4096 /* Hard limit for nfile rlimits */
Linus Torvalds1da177e2005-04-16 15:20:36 -070028
29#define BLOCK_SIZE_BITS 10
30#define BLOCK_SIZE (1<<BLOCK_SIZE_BITS)
31
David Howells01bf4662006-07-10 04:44:53 -070032#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 */
Chris Snook1ae70752007-05-08 00:24:15 -070035#define SEEK_MAX SEEK_END
David Howells01bf4662006-07-10 04:44:53 -070036
Lukas Czerner367a51a2010-10-27 21:30:11 -040037struct fstrim_range {
Loïc Minier3a3a1af2010-11-24 12:56:53 -080038 __u64 start;
39 __u64 len;
40 __u64 minlen;
Lukas Czerner367a51a2010-10-27 21:30:11 -040041};
42
Linus Torvalds1da177e2005-04-16 15:20:36 -070043/* And dynamically-tunable limits and defaults: */
44struct files_stat_struct {
Eric Dumazet518de9b2010-10-26 14:22:44 -070045 unsigned long nr_files; /* read only */
46 unsigned long nr_free_files; /* read only */
47 unsigned long max_files; /* tunable */
Linus Torvalds1da177e2005-04-16 15:20:36 -070048};
Linus Torvalds1da177e2005-04-16 15:20:36 -070049
50struct inodes_stat_t {
51 int nr_inodes;
52 int nr_unused;
Stefan Richter9e7bf242007-07-15 23:40:25 -070053 int dummy[5]; /* padding for sysctl ABI compatibility */
Linus Torvalds1da177e2005-04-16 15:20:36 -070054};
Linus Torvalds1da177e2005-04-16 15:20:36 -070055
Linus Torvalds1da177e2005-04-16 15:20:36 -070056
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
Al Viroe6305c42008-07-15 21:03:57 -040063#define MAY_ACCESS 16
Al Viroa1103432008-07-17 09:19:08 -040064#define MAY_OPEN 32
Eric Paris9cfcac82010-07-23 11:43:51 -040065#define MAY_CHDIR 64
Al Viro1fc0f782011-06-20 18:59:02 -040066#define MAY_NOT_BLOCK 128 /* called from RCU mode, don't block */
Linus Torvalds1da177e2005-04-16 15:20:36 -070067
Paul Turner55ec8212009-02-18 14:48:15 -080068/*
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
Christoph Hellwigfc9161e2008-11-05 14:58:46 +010073/* file is open for reading */
Andrew Morton19adf9c2010-03-05 13:42:03 -080074#define FMODE_READ ((__force fmode_t)0x1)
Christoph Hellwigfc9161e2008-11-05 14:58:46 +010075/* file is open for writing */
Andrew Morton19adf9c2010-03-05 13:42:03 -080076#define FMODE_WRITE ((__force fmode_t)0x2)
Christoph Hellwigfc9161e2008-11-05 14:58:46 +010077/* file is seekable */
Andrew Morton19adf9c2010-03-05 13:42:03 -080078#define FMODE_LSEEK ((__force fmode_t)0x4)
Paul Turner55ec8212009-02-18 14:48:15 -080079/* file can be accessed using pread */
Andrew Morton19adf9c2010-03-05 13:42:03 -080080#define FMODE_PREAD ((__force fmode_t)0x8)
Paul Turner55ec8212009-02-18 14:48:15 -080081/* file can be accessed using pwrite */
Andrew Morton19adf9c2010-03-05 13:42:03 -080082#define FMODE_PWRITE ((__force fmode_t)0x10)
Christoph Hellwigfc9161e2008-11-05 14:58:46 +010083/* File is opened for execution with sys_execve / sys_uselib */
Andrew Morton19adf9c2010-03-05 13:42:03 -080084#define FMODE_EXEC ((__force fmode_t)0x20)
Christoph Hellwigfc9161e2008-11-05 14:58:46 +010085/* File is opened with O_NDELAY (only set for block devices) */
Andrew Morton19adf9c2010-03-05 13:42:03 -080086#define FMODE_NDELAY ((__force fmode_t)0x40)
Christoph Hellwigfc9161e2008-11-05 14:58:46 +010087/* File is opened with O_EXCL (only set for block devices) */
Andrew Morton19adf9c2010-03-05 13:42:03 -080088#define FMODE_EXCL ((__force fmode_t)0x80)
Christoph Hellwigfc9161e2008-11-05 14:58:46 +010089/* File is opened using open(.., 3, ..) and is writeable only for ioctls
90 (specialy hack for floppy.c) */
Andrew Morton19adf9c2010-03-05 13:42:03 -080091#define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
Oleg Drokinb5005312006-03-25 03:07:01 -080092
Christoph Hellwig4d4be482008-12-09 04:47:33 -050093/*
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 */
Andrew Morton19adf9c2010-03-05 13:42:03 -080099#define FMODE_NOCMTIME ((__force fmode_t)0x800)
Christoph Hellwig4d4be482008-12-09 04:47:33 -0500100
Wu Fengguang01414502010-03-05 13:42:03 -0800101/* Expect random access pattern */
Andrew Morton19adf9c2010-03-05 13:42:03 -0800102#define FMODE_RANDOM ((__force fmode_t)0x1000)
Wu Fengguang01414502010-03-05 13:42:03 -0800103
KAMEZAWA Hiroyuki4a3956c2010-10-01 14:20:22 -0700104/* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
105#define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
106
Al Viro1abf0c72011-03-13 03:51:11 -0400107/* File is opened with O_PATH; almost nothing can be done with it */
108#define FMODE_PATH ((__force fmode_t)0x4000)
109
Eric Parisecf081d2009-12-17 21:24:25 -0500110/* File was opened by fanotify and shouldn't generate fanotify events */
Andrew Morton13bcbc02010-08-10 18:01:09 -0700111#define FMODE_NONOTIFY ((__force fmode_t)0x1000000)
Eric Parisecf081d2009-12-17 21:24:25 -0500112
Jens Axboe48e70bc2009-04-14 08:19:27 +0200113/*
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
Tejun Heo7cc01582010-08-03 13:14:58 +0200140 * block layer could (in theory) choose to ignore this
Jens Axboe48e70bc2009-04-14 08:19:27 +0200141 * request if it runs into resource problems.
142 * WRITE A normal async write. Device will be plugged.
Jens Axboe721a9602011-03-09 11:56:30 +0100143 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
Jens Axboe48e70bc2009-04-14 08:19:27 +0200144 * the hint that someone will be waiting on this IO
Jens Axboe721a9602011-03-09 11:56:30 +0100145 * shortly. The write equivalent of READ_SYNC.
146 * WRITE_ODIRECT Special case write for O_DIRECT only.
Tejun Heo4fed9472010-09-03 11:56:17 +0200147 * 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.
Jens Axboe48e70bc2009-04-14 08:19:27 +0200153 *
154 */
Tejun Heo7cc01582010-08-03 13:14:58 +0200155#define RW_MASK REQ_WRITE
156#define RWA_MASK REQ_RAHEAD
Christoph Hellwig7b6d91d2010-08-07 18:20:39 +0200157
158#define READ 0
Tejun Heo7cc01582010-08-03 13:14:58 +0200159#define WRITE RW_MASK
160#define READA RWA_MASK
Christoph Hellwig7b6d91d2010-08-07 18:20:39 +0200161
Jens Axboe721a9602011-03-09 11:56:30 +0100162#define READ_SYNC (READ | REQ_SYNC)
Christoph Hellwig7b6d91d2010-08-07 18:20:39 +0200163#define READ_META (READ | REQ_META)
Jens Axboe721a9602011-03-09 11:56:30 +0100164#define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
165#define WRITE_ODIRECT (WRITE | REQ_SYNC)
Christoph Hellwig7b6d91d2010-08-07 18:20:39 +0200166#define WRITE_META (WRITE | REQ_META)
Jens Axboe721a9602011-03-09 11:56:30 +0100167#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)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700170
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
Miklos Szeredi79c0b2d2007-05-08 00:25:43 -0700178#define FS_HAS_SUBTYPE 4
Linus Torvalds1da177e2005-04-16 15:20:36 -0700179#define FS_REVAL_DOT 16384 /* Check the paths ".", ".." for staleness */
Mark Fasheh349457c2006-09-08 14:22:21 -0700180#define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move()
181 * during rename() internally.
182 */
183
Linus Torvalds1da177e2005-04-16 15:20:36 -0700184/*
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
Theodore Ts'o9b04c992006-03-24 03:15:10 -0800200#define MS_VERBOSE 32768 /* War is peace. Verbosity is silence.
201 MS_VERBOSE is deprecated. */
202#define MS_SILENT 32768
Miklos Szeredibf066c72006-01-08 01:03:19 -0800203#define MS_POSIXACL (1<<16) /* VFS does not apply the umask */
Ram Pai9676f0c2005-11-07 17:21:20 -0500204#define MS_UNBINDABLE (1<<17) /* change to unbindable */
Ram Pai07b20882005-11-07 17:19:07 -0500205#define MS_PRIVATE (1<<18) /* change to private */
Ram Paia58b0eb2005-11-07 17:20:48 -0500206#define MS_SLAVE (1<<19) /* change to slave */
Ram Pai03e06e62005-11-07 17:19:33 -0500207#define MS_SHARED (1<<20) /* change to shared */
Valerie Henson47ae32d2006-12-13 00:34:34 -0800208#define MS_RELATIME (1<<21) /* Update atime relative to mtime/ctime. */
Pavel Emelyanov8bf97252007-10-18 23:40:02 -0700209#define MS_KERNMOUNT (1<<22) /* this is a kern_mount call */
Jean Noel Cordenner7a224222008-01-28 23:58:27 -0500210#define MS_I_VERSION (1<<23) /* Update inode I_version field */
Matthew Garrettd0adde52009-03-26 17:49:56 +0000211#define MS_STRICTATIME (1<<24) /* Always perform atime updates */
Al Viro9e1f1de2011-06-03 18:24:58 -0400212#define MS_NOSEC (1<<28)
Al Viro7a4dec52010-08-09 12:05:43 -0400213#define MS_BORN (1<<29)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700214#define MS_ACTIVE (1<<30)
215#define MS_NOUSER (1<<31)
216
217/*
218 * Superblock flags that can be altered by MS_REMOUNT
219 */
Mimi Zohar08b9fe62008-10-13 00:09:50 -0400220#define MS_RMT_MASK (MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK|MS_I_VERSION)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700221
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 */
Eric Paris196f5182010-10-25 14:42:19 -0400240#define S_IMA 1024 /* Inode has an associated IMA struct */
David Howells9875cf82011-01-14 18:45:21 +0000241#define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
Andi Kleen69b45732011-05-28 08:25:51 -0700242#define S_NOSEC 4096 /* no suid or xattr security attributes */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700243
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)
Eric Dumazet37756ce2007-02-10 01:44:49 -0800265#define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
Jean Noel Cordenner7a224222008-01-28 23:58:27 -0500266#define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700267
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)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700271#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)
Eric Paris196f5182010-10-25 14:42:19 -0400277#define IS_IMA(inode) ((inode)->i_flags & S_IMA)
David Howells9875cf82011-01-14 18:45:21 +0000278#define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
Andi Kleen69b45732011-05-28 08:25:51 -0700279#define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700280
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) */
Jens Axboe2056a782006-03-23 20:00:26 +0100310#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)
David Woodhoused30a2602008-08-11 15:58:42 +0100314#define BLKDISCARD _IO(0x12,119)
Martin K. Petersenac481c22009-10-03 20:52:01 +0200315#define BLKIOMIN _IO(0x12,120)
316#define BLKIOOPT _IO(0x12,121)
317#define BLKALIGNOFF _IO(0x12,122)
318#define BLKPBSZGET _IO(0x12,123)
Martin K. Petersen98262f22009-12-03 09:24:48 +0100319#define BLKDISCARDZEROES _IO(0x12,124)
Adrian Hunter8d57a982010-08-11 14:17:49 -0700320#define BLKSECDISCARD _IO(0x12,125)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700321
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 */
Takashi Satofcccf502009-01-09 16:40:59 -0800325#define FIFREEZE _IOWR('X', 119, int) /* Freeze */
326#define FITHAW _IOWR('X', 120, int) /* Thaw */
Lukas Czerner367a51a2010-10-27 21:30:11 -0400327#define FITRIM _IOWR('X', 121, struct fstrim_range) /* Trim */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700328
David Howells36695672006-08-29 19:06:16 +0100329#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)
Mark Fashehc4b929b2008-10-08 19:44:18 -0400333#define FS_IOC_FIEMAP _IOWR('f', 11, struct fiemap)
David Howells36695672006-08-29 19:06:16 +0100334#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*/
Steven Whitehouse128e5eb2006-10-02 11:24:43 -0400363#define FS_EXTENT_FL 0x00080000 /* Extents */
364#define FS_DIRECTIO_FL 0x00100000 /* Use direct i/o */
liubo32471f62011-03-21 08:54:27 +0000365#define FS_NOCOW_FL 0x00800000 /* Do not cow file */
David Howells36695672006-08-29 19:06:16 +0100366#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
Andrew Mortone46e4902006-05-22 22:35:24 -0700372#define SYNC_FILE_RANGE_WAIT_BEFORE 1
373#define SYNC_FILE_RANGE_WRITE 2
374#define SYNC_FILE_RANGE_WAIT_AFTER 4
375
Linus Torvalds1da177e2005-04-16 15:20:36 -0700376#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>
Al Viro3f8206d2008-07-26 03:46:43 -0400383#include <linux/path.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700384#include <linux/stat.h>
385#include <linux/cache.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700386#include <linux/list.h>
387#include <linux/radix-tree.h>
388#include <linux/prio_tree.h>
389#include <linux/init.h>
Al Viro914e2632006-10-18 13:55:46 -0400390#include <linux/pid.h>
Jes Sorensen1b1dcc12006-01-09 15:59:24 -0800391#include <linux/mutex.h>
Satyam Sharma3bd858a2007-07-17 15:00:08 +0530392#include <linux/capability.h>
Matthew Wilcox6188e102008-04-18 22:21:05 -0400393#include <linux/semaphore.h>
Mark Fashehc4b929b2008-10-08 19:44:18 -0400394#include <linux/fiemap.h>
Nick Pigginceb5bdc2011-01-07 17:50:05 +1100395#include <linux/rculist_bl.h>
Al Viro07b8ce12011-06-20 10:52:57 -0400396#include <linux/atomic.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700397
Linus Torvalds1da177e2005-04-16 15:20:36 -0700398#include <asm/byteorder.h>
399
Christoph Hellwiga5694252007-07-17 04:04:28 -0700400struct export_operations;
Christoph Hellwiga885c8c2006-01-08 01:02:50 -0800401struct hd_geometry;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700402struct iovec;
403struct nameidata;
Christoph Hellwig92198f72005-06-23 22:00:59 -0700404struct kiocb;
Alexey Dobriyan57cc7212011-01-10 08:18:25 +0200405struct kobject;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700406struct pipe_inode_info;
407struct poll_table_struct;
408struct kstatfs;
409struct vm_area_struct;
410struct vfsmount;
David Howells745ca242008-11-14 10:39:22 +1100411struct cred;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700412
Denis Cheng74bf17c2007-10-16 23:26:30 -0700413extern void __init inode_init(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700414extern void __init inode_init_early(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700415extern void __init files_init(unsigned long);
416
Jan Engelhardtdded4f42008-12-01 14:34:50 -0800417extern struct files_stat_struct files_stat;
Eric Dumazet518de9b2010-10-26 14:22:44 -0700418extern unsigned long get_max_files(void);
Jan Engelhardtdded4f42008-12-01 14:34:50 -0800419extern int sysctl_nr_open;
420extern struct inodes_stat_t inodes_stat;
421extern int leases_enable, lease_break_time;
Jan Engelhardtdded4f42008-12-01 14:34:50 -0800422
Linus Torvalds1da177e2005-04-16 15:20:36 -0700423struct buffer_head;
424typedef int (get_block_t)(struct inode *inode, sector_t iblock,
425 struct buffer_head *bh_result, int create);
Christoph Hellwig92198f72005-06-23 22:00:59 -0700426typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
Christoph Hellwig40e2e972010-07-18 21:17:09 +0000427 ssize_t bytes, void *private, int ret,
428 bool is_async);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700429
430/*
431 * Attribute flags. These should be or-ed together to figure out what
432 * has been changed!
433 */
Miklos Szeredi9767d742008-07-01 15:01:26 +0200434#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)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700451
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;
Miklos Szeredicc4e69d2005-11-07 00:59:49 -0800470
471 /*
Lucas De Marchi25985ed2011-03-30 22:57:33 -0300472 * Not an attribute, but an auxiliary info for filesystems wanting to
Miklos Szeredicc4e69d2005-11-07 00:59:49 -0800473 * 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;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700477};
478
479/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700480 * Includes for diskquotas.
481 */
482#include <linux/quota.h>
483
Zach Brown994fc28c2005-12-15 14:28:17 -0800484/**
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
Nick Piggin55144762007-10-16 01:25:26 -0700502 * by readpage().
Zach Brown994fc28c2005-12-15 14:28:17 -0800503 *
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
Nick Pigginafddba42007-10-16 01:25:01 -0700515#define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
Nick Piggin89e10782007-10-16 01:25:07 -0700516#define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
Nick Piggin54566b22009-01-04 12:00:53 -0800517#define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
518 * helper code (eg buffer layer)
519 * to clear GFP_FS from alloc */
Nick Pigginafddba42007-10-16 01:25:01 -0700520
Linus Torvalds1da177e2005-04-16 15:20:36 -0700521/*
522 * oh the beauties of C type declarations.
523 */
524struct page;
525struct address_space;
526struct writeback_control;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700527
Nick Piggin2f718ff2007-10-16 01:24:59 -0700528struct 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);
Nick Pigginafddba42007-10-16 01:25:01 -0700540int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes);
Nick Piggin2f718ff2007-10-16 01:24:59 -0700541size_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
Hisashi Hifumi8ab22b92008-07-28 15:46:36 -0700560/*
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);
Nick Piggin2f718ff2007-10-16 01:24:59 -0700581
Linus Torvalds1da177e2005-04-16 15:20:36 -0700582struct address_space_operations {
583 int (*writepage)(struct page *page, struct writeback_control *wbc);
584 int (*readpage)(struct file *, struct page *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700585
586 /* Write back some dirty pages from this mapping. */
587 int (*writepages)(struct address_space *, struct writeback_control *);
588
Andrew Morton4741c9f2006-03-24 03:18:11 -0800589 /* Set a page dirty. Return true if this dirtied it */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700590 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
Nick Pigginafddba42007-10-16 01:25:01 -0700595 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
Linus Torvalds1da177e2005-04-16 15:20:36 -0700602 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
603 sector_t (*bmap)(struct address_space *, sector_t);
NeilBrown2ff28e22006-03-26 01:37:18 -0800604 void (*invalidatepage) (struct page *, unsigned long);
Al Viro27496a82005-10-21 03:20:48 -0400605 int (*releasepage) (struct page *, gfp_t);
Linus Torvalds6072d132010-12-01 13:35:19 -0500606 void (*freepage)(struct page *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700607 ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
608 loff_t offset, unsigned long nr_segs);
Nick Piggin70688e42008-04-28 02:13:02 -0700609 int (*get_xip_mem)(struct address_space *, pgoff_t, int,
610 void **, unsigned long *);
Christoph Lametere965f962006-02-01 03:05:41 -0800611 /* migrate the contents of a page to the specified target */
Christoph Lameter2d1db3b2006-06-23 02:03:33 -0700612 int (*migratepage) (struct address_space *,
613 struct page *, struct page *);
Trond Myklebuste3db7692007-01-10 23:15:39 -0800614 int (*launder_page) (struct page *);
Hisashi Hifumi8ab22b92008-07-28 15:46:36 -0700615 int (*is_partially_uptodate) (struct page *, read_descriptor_t *,
616 unsigned long);
Andi Kleen25718732009-09-16 11:50:13 +0200617 int (*error_remove_page)(struct address_space *, struct page *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700618};
619
Jens Axboe7dcda1c2011-04-05 23:51:48 +0200620extern const struct address_space_operations empty_aops;
621
Nick Pigginafddba42007-10-16 01:25:01 -0700622/*
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
Linus Torvalds1da177e2005-04-16 15:20:36 -0700634struct 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 */
Nick Piggin19fd6232008-07-25 19:45:32 -0700638 spinlock_t tree_lock; /* and lock protecting it */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700639 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 */
Peter Zijlstra3d48ae42011-05-24 17:12:06 -0700642 struct mutex i_mmap_mutex; /* protect tree, count, list */
Jan Kara08142572011-06-27 16:18:10 -0700643 /* Protected by tree_lock together with the radix tree */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700644 unsigned long nrpages; /* number of total pages */
645 pgoff_t writeback_index;/* writeback starts here */
Christoph Hellwigf5e54d62006-06-28 04:26:44 -0700646 const struct address_space_operations *a_ops; /* methods */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700647 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
Lucas De Marchi25985ed2011-03-30 22:57:33 -0300655 * must be enforced here for CRIS, to let the least significant bit
Linus Torvalds1da177e2005-04-16 15:20:36 -0700656 * 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 */
Richard Kennedyf3ccfcd2011-03-22 16:33:56 -0700661 int bd_openers;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700662 struct inode * bd_inode; /* will die */
Theodore Ts'o87d8fe12009-01-03 09:47:09 -0500663 struct super_block * bd_super;
Arjan van de Venc039e312006-03-23 03:00:28 -0800664 struct mutex bd_mutex; /* open/close mutex */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700665 struct list_head bd_inodes;
Tejun Heo6b4517a2010-04-07 18:53:59 +0900666 void * bd_claiming;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700667 void * bd_holder;
668 int bd_holders;
Tejun Heo77ea8872010-12-08 20:57:37 +0100669 bool bd_write_holder;
Jun'ichi Nomura641dc632006-03-27 01:17:57 -0800670#ifdef CONFIG_SYSFS
Tejun Heo49731ba2011-01-14 18:43:57 +0100671 struct list_head bd_holder_disks;
Jun'ichi Nomura641dc632006-03-27 01:17:57 -0800672#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700673 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;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700681 /*
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;
Takashi Satofcccf502009-01-09 16:40:59 -0800688
689 /* The counter of freeze processes */
690 int bd_fsfreeze_count;
691 /* Mutex for freeze */
692 struct mutex bd_fsfreeze_mutex;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700693};
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
Jan Karaf446daa2010-08-09 17:19:12 -0700701#define PAGECACHE_TAG_TOWRITE 2
Linus Torvalds1da177e2005-04-16 15:20:36 -0700702
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
Al Virof19d4a82009-06-08 19:50:45 -0400736struct posix_acl;
737#define ACL_NOT_CACHED ((void *)(-1))
738
Linus Torvalds1da177e2005-04-16 15:20:36 -0700739struct inode {
Nick Piggin44a7d7a2011-01-07 17:49:56 +1100740 /* 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;
Linus Torvalds79568f52011-06-20 20:13:49 -0700749 unsigned long i_state;
Linus Torvalds13e12d12011-06-08 15:18:19 -0700750#ifdef CONFIG_SECURITY
751 void *i_security;
752#endif
Nick Piggin44a7d7a2011-01-07 17:49:56 +1100753 struct mutex i_mutex;
754
Linus Torvalds13e12d12011-06-08 15:18:19 -0700755
Nick Piggin44a7d7a2011-01-07 17:49:56 +1100756 unsigned long dirtied_when; /* jiffies of first dirtying */
757
Linus Torvalds1da177e2005-04-16 15:20:36 -0700758 struct hlist_node i_hash;
Nick Piggin7ccf19a2010-10-21 11:49:30 +1100759 struct list_head i_wb_list; /* backing dev IO list */
760 struct list_head i_lru; /* inode LRU list */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700761 struct list_head i_sb_list;
Nick Pigginfa0d7e32011-01-07 17:49:49 +1100762 union {
763 struct list_head i_dentry;
764 struct rcu_head i_rcu;
765 };
Linus Torvalds1da177e2005-04-16 15:20:36 -0700766 unsigned long i_ino;
767 atomic_t i_count;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700768 unsigned int i_nlink;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700769 dev_t i_rdev;
Richard Kennedy4e70af52010-02-15 11:16:11 +0000770 unsigned int i_blkbits;
Jean Noel Cordenner7a224222008-01-28 23:58:27 -0500771 u64 i_version;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700772 loff_t i_size;
Eric Dumazet83b7b442006-12-06 20:38:53 -0800773#ifdef __NEED_I_SIZE_ORDERED
774 seqcount_t i_size_seqcount;
775#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700776 struct timespec i_atime;
777 struct timespec i_mtime;
778 struct timespec i_ctime;
Takashi Satoa0f62ac2006-03-26 01:37:52 -0800779 blkcnt_t i_blocks;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700780 unsigned short i_bytes;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700781 struct rw_semaphore i_alloc_sem;
Arjan van de Ven99ac48f2006-03-28 01:56:41 -0800782 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700783 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
Linus Torvalds1da177e2005-04-16 15:20:36 -0700789 struct list_head i_devices;
Theodore Ts'o4c154162006-09-27 01:50:47 -0700790 union {
791 struct pipe_inode_info *i_pipe;
Theodore Ts'oeaf796e2006-09-27 01:50:48 -0700792 struct block_device *i_bdev;
Theodore Ts'o577c4eb2006-09-27 01:50:49 -0700793 struct cdev *i_cdev;
Theodore Ts'o4c154162006-09-27 01:50:47 -0700794 };
Linus Torvalds1da177e2005-04-16 15:20:36 -0700795
796 __u32 i_generation;
797
Eric Paris3be25f42009-05-21 17:01:26 -0400798#ifdef CONFIG_FSNOTIFY
799 __u32 i_fsnotify_mask; /* all events this inode cares about */
Eric Parise61ce862009-12-17 21:24:24 -0500800 struct hlist_head i_fsnotify_marks;
Robert Love0eeca282005-07-12 17:06:03 -0400801#endif
802
Eric Parisa178d202010-10-25 14:41:59 -0400803#ifdef CONFIG_IMA
Mimi Zohara68a27b2010-11-02 10:10:56 -0400804 atomic_t i_readcount; /* struct files open RO */
Eric Parisa178d202010-10-25 14:41:59 -0400805#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700806 atomic_t i_writecount;
Al Virof19d4a82009-06-08 19:50:45 -0400807#ifdef CONFIG_FS_POSIX_ACL
808 struct posix_acl *i_acl;
809 struct posix_acl *i_default_acl;
810#endif
Theodore Ts'o8e18e292006-09-27 01:50:46 -0700811 void *i_private; /* fs or device private pointer */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700812};
813
Al Viro1d3382c2010-10-23 15:19:20 -0400814static inline int inode_unhashed(struct inode *inode)
815{
816 return hlist_unhashed(&inode->i_hash);
817}
818
Linus Torvalds1da177e2005-04-16 15:20:36 -0700819/*
Ingo Molnarf2eace22006-07-03 00:25:05 -0700820 * 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
Ingo Molnar4df46242006-08-27 01:23:56 -0700828 * parent -> child -> normal -> xattr -> quota
Ingo Molnarf2eace22006-07-03 00:25:05 -0700829 */
830enum inode_i_mutex_lock_class
831{
832 I_MUTEX_NORMAL,
833 I_MUTEX_PARENT,
834 I_MUTEX_CHILD,
Ingo Molnar4df46242006-08-27 01:23:56 -0700835 I_MUTEX_XATTR,
Ingo Molnarf2eace22006-07-03 00:25:05 -0700836 I_MUTEX_QUOTA
837};
838
839/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700840 * 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 */
Jan Engelhardt48ed2142006-12-06 20:35:37 -0800849static inline loff_t i_size_read(const struct inode *inode)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700850{
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
Miklos Szeredi7762f5a2006-10-17 00:10:07 -0700872/*
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 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700877static 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
Jan Engelhardt48ed2142006-12-06 20:35:37 -0800892static inline unsigned iminor(const struct inode *inode)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700893{
894 return MINOR(inode->i_rdev);
895}
896
Jan Engelhardt48ed2142006-12-06 20:35:37 -0800897static inline unsigned imajor(const struct inode *inode)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700898{
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 */
Eric W. Biederman609d7fa2006-10-02 02:17:15 -0700906 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 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700908 uid_t uid, euid; /* uid/euid of process setting the owner */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700909 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 {
Fengguang Wu937085a2007-10-16 01:24:31 -0700916 pgoff_t start; /* where readahead started */
917 unsigned int size; /* # of readahead pages */
918 unsigned int async_size; /* do asynchronous readahead when
Fengguang Wuf9acc8c2007-07-19 01:48:08 -0700919 there are only # of pages ahead */
Fengguang Wu5ce11102007-07-19 01:47:59 -0700920
Fengguang Wu937085a2007-10-16 01:24:31 -0700921 unsigned int ra_pages; /* Maximum readahead window */
Wu Fengguang1ebf26a2009-06-16 15:31:19 -0700922 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
Fengguang Wuf4e6b492007-10-16 01:24:33 -0700923 loff_t prev_pos; /* Cache last read() position */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700924};
Linus Torvalds1da177e2005-04-16 15:20:36 -0700925
Fengguang Wu5ce11102007-07-19 01:47:59 -0700926/*
Fengguang Wu5ce11102007-07-19 01:47:59 -0700927 * Check if @index falls in the readahead windows.
928 */
929static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
930{
Fengguang Wuf9acc8c2007-07-19 01:48:08 -0700931 return (index >= ra->start &&
932 index < ra->start + ra->size);
Fengguang Wu5ce11102007-07-19 01:47:59 -0700933}
934
Dave Hansenad775f52008-02-15 14:38:01 -0800935#define FILE_MNT_WRITE_TAKEN 1
936#define FILE_MNT_WRITE_RELEASED 2
937
Linus Torvalds1da177e2005-04-16 15:20:36 -0700938struct file {
Eric Dumazet2f512012005-10-30 15:02:16 -0800939 /*
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;
Josef "Jeff" Sipek0f7fc9e2006-12-08 02:36:35 -0800947 struct path f_path;
948#define f_dentry f_path.dentry
949#define f_vfsmnt f_path.mnt
Arjan van de Ven99ac48f2006-03-28 01:56:41 -0800950 const struct file_operations *f_op;
Jonathan Corbet4a6a4492009-03-27 12:24:31 -0600951 spinlock_t f_lock; /* f_ep_links, f_flags, no IRQ */
Nick Piggin6416ccb2010-08-18 04:37:38 +1000952#ifdef CONFIG_SMP
953 int f_sb_list_cpu;
954#endif
Al Viro516e0cc2008-07-26 00:39:17 -0400955 atomic_long_t f_count;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700956 unsigned int f_flags;
Al Viroaeb5d722008-09-02 15:28:45 -0400957 fmode_t f_mode;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700958 loff_t f_pos;
959 struct fown_struct f_owner;
David Howellsd76b0d92008-11-14 10:39:25 +1100960 const struct cred *f_cred;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700961 struct file_ra_state f_ra;
962
Mathieu Desnoyers2b47c362007-10-16 23:27:21 -0700963 u64 f_version;
Alexey Dobriyan50462062006-09-29 02:00:01 -0700964#ifdef CONFIG_SECURITY
Linus Torvalds1da177e2005-04-16 15:20:36 -0700965 void *f_security;
Alexey Dobriyan50462062006-09-29 02:00:01 -0700966#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700967 /* 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;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700973#endif /* #ifdef CONFIG_EPOLL */
974 struct address_space *f_mapping;
Dave Hansenad775f52008-02-15 14:38:01 -0800975#ifdef CONFIG_DEBUG_WRITECOUNT
976 unsigned long f_mnt_write_state;
977#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700978};
Linus Torvalds1da177e2005-04-16 15:20:36 -0700979
Aneesh Kumar K.V990d6c22011-01-29 18:43:26 +0530980struct file_handle {
981 __u32 handle_bytes;
982 int handle_type;
983 /* file identifier */
984 unsigned char f_handle[0];
985};
986
Al Viro516e0cc2008-07-26 00:39:17 -0400987#define get_file(x) atomic_long_inc(&(x)->f_count)
Al Virod7065da2010-05-26 15:13:55 -0400988#define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
Al Viro516e0cc2008-07-26 00:39:17 -0400989#define file_count(x) atomic_long_read(&(x)->f_count)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700990
Dave Hansenad775f52008-02-15 14:38:01 -0800991#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
Linus Torvalds1da177e2005-04-16 15:20:36 -07001034#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 */
Trond Myklebustf475ae92006-06-29 16:38:32 -04001047#define FL_EXISTS 16 /* when unlocking, test for existence */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001048#define FL_LEASE 32 /* lease held on this file */
Miklos Szeredi75e1fcc2006-06-23 02:05:12 -07001049#define FL_CLOSE 64 /* unlock on close */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001050#define FL_SLEEP 128 /* A blocking lock */
1051
1052/*
Miklos Szeredibde74e42008-07-25 01:48:57 -07001053 * Special return value from posix_lock_file() and vfs_lock_file() for
1054 * asynchronous locking.
1055 */
1056#define FILE_LOCK_DEFERRED 1
1057
1058/*
Linus Torvalds1da177e2005-04-16 15:20:36 -07001059 * 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 {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001068 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 */
Marc Eshel2beb6612006-12-05 23:31:28 -05001075 int (*fl_grant)(struct file_lock *, struct file_lock *, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001076 void (*fl_release_private)(struct file_lock *);
1077 void (*fl_break)(struct file_lock *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001078 int (*fl_change)(struct file_lock **, int);
1079};
1080
J. Bruce Fieldsaf558e32007-09-06 12:34:25 -04001081struct 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
Linus Torvalds1da177e2005-04-16 15:20:36 -07001089/* 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;
Richard Kennedyafc12462008-07-11 17:20:49 -07001097 unsigned char fl_flags;
1098 unsigned char fl_type;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001099 unsigned int fl_pid;
Vitaliy Gusevab1f1612008-01-17 00:07:08 +00001100 struct pid *fl_nspid;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001101 wait_queue_head_t fl_wait;
1102 struct file *fl_file;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001103 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
Alexey Dobriyan6aed6282009-09-21 17:01:11 -07001109 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
Alexey Dobriyan7b021962009-09-21 17:01:12 -07001110 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001111 union {
1112 struct nfs_lock_info nfs_fl;
Trond Myklebust8d0a8a92005-06-22 17:16:32 +00001113 struct nfs4_lock_info nfs4_fl;
David Howellse8d6c552007-07-15 23:40:12 -07001114 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;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001118 } 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
Linus Torvalds1da177e2005-04-16 15:20:36 -07001128#include <linux/fcntl.h>
1129
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001130extern void send_sigio(struct fown_struct *fown, int fd, int band);
1131
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001132#ifdef CONFIG_FILE_LOCKING
Linus Torvalds1da177e2005-04-16 15:20:36 -07001133extern int fcntl_getlk(struct file *, struct flock __user *);
Peter Staubachc2936212005-07-27 11:45:09 -07001134extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1135 struct flock __user *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001136
1137#if BITS_PER_LONG == 32
1138extern int fcntl_getlk64(struct file *, struct flock64 __user *);
Peter Staubachc2936212005-07-27 11:45:09 -07001139extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1140 struct flock64 __user *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001141#endif
1142
Linus Torvalds1da177e2005-04-16 15:20:36 -07001143extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1144extern int fcntl_getlease(struct file *filp);
1145
1146/* fs/locks.c */
J. Bruce Fields05fa3132010-10-30 17:31:15 -04001147void locks_free_lock(struct file_lock *fl);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001148extern void locks_init_lock(struct file_lock *);
Arnd Bergmannc5b1f0d2010-10-27 15:46:08 +02001149extern struct file_lock * locks_alloc_lock(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001150extern void locks_copy_lock(struct file_lock *, struct file_lock *);
J. Bruce Fields1a747ee2008-04-24 10:08:22 -04001151extern void __locks_copy_lock(struct file_lock *, const struct file_lock *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001152extern void locks_remove_posix(struct file *, fl_owner_t);
1153extern void locks_remove_flock(struct file *);
Felix Blyakhera9e61e22009-03-31 15:12:56 -05001154extern void locks_release_private(struct file_lock *);
J. Bruce Fields6d34ac12007-05-11 16:09:32 -04001155extern void posix_test_lock(struct file *, struct file_lock *);
Marc Eshel150b3932007-01-18 16:15:35 -05001156extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001157extern int posix_lock_file_wait(struct file *, struct file_lock *);
J. Bruce Fields64a318e2006-01-03 09:55:46 +01001158extern int posix_unblock_lock(struct file *, struct file_lock *);
J. Bruce Fields3ee17ab2007-02-21 00:58:50 -05001159extern int vfs_test_lock(struct file *, struct file_lock *);
Marc Eshel150b3932007-01-18 16:15:35 -05001160extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
Marc Eshel9b9d2ab2007-01-18 17:52:58 -05001161extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001162extern 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);
Christoph Hellwig0af1a452007-07-31 00:39:22 -07001165extern int generic_setlease(struct file *, long, struct file_lock **);
J. Bruce Fieldsa9933ce2007-06-07 17:09:49 -04001166extern int vfs_setlease(struct file *, long, struct file_lock **);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001167extern 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);
Arnd Bergmannb89f4322010-09-18 15:09:31 +02001170extern void lock_flocks(void);
1171extern void unlock_flocks(void);
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001172#else /* !CONFIG_FILE_LOCKING */
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001173static 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
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001184#if BITS_PER_LONG == 32
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001185static 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}
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001195#endif
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001196static 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
Arnd Bergmannb89f4322010-09-18 15:09:31 +02001314static inline void lock_flocks(void)
1315{
1316}
1317
1318static inline void unlock_flocks(void)
1319{
1320}
1321
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001322#endif /* !CONFIG_FILE_LOCKING */
1323
Linus Torvalds1da177e2005-04-16 15:20:36 -07001324
1325struct fasync_struct {
Eric Dumazet989a2972010-04-14 09:55:35 +00001326 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;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001332};
1333
1334#define FASYNC_MAGIC 0x4601
1335
1336/* SMP safe fasync helpers: */
1337extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
Linus Torvaldsf7347ce2010-10-27 12:38:12 -04001338extern 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
Linus Torvalds1da177e2005-04-16 15:20:36 -07001343/* can be called from interrupts */
1344extern void kill_fasync(struct fasync_struct **, int, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001345
Eric W. Biederman609d7fa2006-10-02 02:17:15 -07001346extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001347extern int f_setown(struct file *filp, unsigned long arg, int force);
1348extern void f_delown(struct file *filp);
Eric W. Biederman609d7fa2006-10-02 02:17:15 -07001349extern pid_t f_getown(struct file *filp);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001350extern 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 */
Miklos Szeredidb1f05b2010-02-10 12:15:53 +01001359#define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1360#define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001361
1362extern struct list_head super_blocks;
1363extern spinlock_t sb_lock;
1364
Linus Torvalds1da177e2005-04-16 15:20:36 -07001365struct super_block {
1366 struct list_head s_list; /* Keep this first */
1367 dev_t s_dev; /* search index; _not_ kdev_t */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001368 unsigned char s_dirt;
Richard Kennedy270ba5f2010-01-26 14:12:43 +00001369 unsigned char s_blocksize_bits;
1370 unsigned long s_blocksize;
Jeff Layton42cb56a2009-09-18 13:05:53 -07001371 loff_t s_maxbytes; /* Max file size */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001372 struct file_system_type *s_type;
Josef 'Jeff' Sipekee9b6d62007-02-12 00:55:41 -08001373 const struct super_operations *s_op;
Alexey Dobriyan61e225d2009-09-21 17:01:08 -07001374 const struct dquot_operations *dq_op;
Alexey Dobriyan0d54b212009-09-21 17:01:09 -07001375 const struct quotactl_ops *s_qcop;
Christoph Hellwig39655162007-10-21 16:42:17 -07001376 const struct export_operations *s_export_op;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001377 unsigned long s_flags;
1378 unsigned long s_magic;
1379 struct dentry *s_root;
1380 struct rw_semaphore s_umount;
Ingo Molnar7892f2f2006-01-09 15:59:25 -08001381 struct mutex s_lock;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001382 int s_count;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001383 atomic_t s_active;
Alexey Dobriyan50462062006-09-29 02:00:01 -07001384#ifdef CONFIG_SECURITY
Linus Torvalds1da177e2005-04-16 15:20:36 -07001385 void *s_security;
Alexey Dobriyan50462062006-09-29 02:00:01 -07001386#endif
Stephen Hemmingerbb435452010-05-13 17:53:14 -07001387 const struct xattr_handler **s_xattr;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001388
1389 struct list_head s_inodes; /* all inodes */
Nick Pigginceb5bdc2011-01-07 17:50:05 +11001390 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
Nick Piggin6416ccb2010-08-18 04:37:38 +10001391#ifdef CONFIG_SMP
1392 struct list_head __percpu *s_files;
1393#else
Linus Torvalds1da177e2005-04-16 15:20:36 -07001394 struct list_head s_files;
Nick Piggin6416ccb2010-08-18 04:37:38 +10001395#endif
Nick Pigginb5c84bf2011-01-07 17:49:38 +11001396 /* s_dentry_lru, s_nr_dentry_unused protected by dcache.c lru locks */
Kentaro Makitada3bbdd2008-07-23 21:27:13 -07001397 struct list_head s_dentry_lru; /* unused dentry lru */
1398 int s_nr_dentry_unused; /* # of dentry on lru */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001399
Dave Chinner09cc9fc2011-07-08 14:14:40 +10001400 /* s_inode_lru_lock protects s_inode_lru and s_nr_inodes_unused */
1401 spinlock_t s_inode_lru_lock ____cacheline_aligned_in_smp;
Dave Chinner98b745c2011-07-08 14:14:39 +10001402 struct list_head s_inode_lru; /* unused inode lru */
1403 int s_nr_inodes_unused; /* # of inodes on lru */
1404
Linus Torvalds1da177e2005-04-16 15:20:36 -07001405 struct block_device *s_bdev;
Jens Axboe32a88aa2009-09-16 15:02:33 +02001406 struct backing_dev_info *s_bdi;
David Howellsacaebfd2007-05-10 22:51:50 -07001407 struct mtd_info *s_mtd;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001408 struct list_head s_instances;
1409 struct quota_info s_dquot; /* Diskquota specific options */
1410
1411 int s_frozen;
1412 wait_queue_head_t s_wait_unfrozen;
1413
1414 char s_id[32]; /* Informational name */
Linus Torvalds0bba0162011-04-12 15:21:04 -07001415 u8 s_uuid[16]; /* UUID */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001416
1417 void *s_fs_info; /* Filesystem private info */
Al Viro30c40d22008-02-22 19:50:45 -05001418 fmode_t s_mode;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001419
Richard Kennedy270ba5f2010-01-26 14:12:43 +00001420 /* Granularity of c/m/atime in ns.
1421 Cannot be worse than a second */
1422 u32 s_time_gran;
1423
Linus Torvalds1da177e2005-04-16 15:20:36 -07001424 /*
1425 * The next field is for VFS *only*. No filesystems have any business
1426 * even looking at it. You had been warned.
1427 */
Arjan van de Vena11f3a02006-03-23 03:00:33 -08001428 struct mutex s_vfs_rename_mutex; /* Kludge */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001429
Miklos Szeredi79c0b2d2007-05-08 00:25:43 -07001430 /*
1431 * Filesystem subtype. If non-empty the filesystem type field
1432 * in /proc/mounts will be "type.subtype"
1433 */
1434 char *s_subtype;
Miklos Szeredib3b304a2008-02-08 04:21:35 -08001435
1436 /*
1437 * Saved mount options for lazy filesystems using
1438 * generic_show_options()
1439 */
Arnd Bergmann4d2deb42010-02-24 20:01:56 +01001440 char __rcu *s_options;
Al Viroc8aebb02010-12-18 10:22:30 -05001441 const struct dentry_operations *s_d_op; /* default d_op for dentries */
Dan Magenheimer9fdfdcf2011-05-26 10:01:19 -06001442
1443 /*
1444 * Saved pool identifier for cleancache (-1 means none)
1445 */
1446 int cleancache_poolid;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001447};
1448
1449extern struct timespec current_fs_time(struct super_block *sb);
1450
1451/*
1452 * Snapshotting support.
1453 */
1454enum {
1455 SB_UNFROZEN = 0,
1456 SB_FREEZE_WRITE = 1,
1457 SB_FREEZE_TRANS = 2,
1458};
1459
1460#define vfs_check_frozen(sb, level) \
1461 wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
1462
Al Viro914e2632006-10-18 13:55:46 -04001463#define get_fs_excl() atomic_inc(&current->fs_excl)
1464#define put_fs_excl() atomic_dec(&current->fs_excl)
1465#define has_fs_excl() atomic_read(&current->fs_excl)
Jens Axboe22e2c502005-06-27 10:55:12 +02001466
Serge E. Hallyne795b712011-03-23 16:43:25 -07001467/*
1468 * until VFS tracks user namespaces for inodes, just make all files
1469 * belong to init_user_ns
1470 */
1471extern struct user_namespace init_user_ns;
1472#define inode_userns(inode) (&init_user_ns)
Serge E. Hallyn2e149672011-03-23 16:43:26 -07001473extern bool inode_owner_or_capable(const struct inode *inode);
Satyam Sharma3bd858a2007-07-17 15:00:08 +05301474
Al Viro914e2632006-10-18 13:55:46 -04001475/* not quite ready to be deprecated, but... */
1476extern void lock_super(struct super_block *);
1477extern void unlock_super(struct super_block *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001478
1479/*
1480 * VFS helper functions..
1481 */
1482extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *);
1483extern int vfs_mkdir(struct inode *, struct dentry *, int);
1484extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t);
Miklos Szeredidb2e7472008-06-24 16:50:16 +02001485extern int vfs_symlink(struct inode *, struct dentry *, const char *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001486extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
1487extern int vfs_rmdir(struct inode *, struct dentry *);
1488extern int vfs_unlink(struct inode *, struct dentry *);
1489extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1490
1491/*
1492 * VFS dentry helper functions.
1493 */
1494extern void dentry_unhash(struct dentry *dentry);
1495
1496/*
Christoph Hellwig8c744fb2005-11-08 21:35:04 -08001497 * VFS file helper functions.
1498 */
Dmitry Monakhova1bd1202010-03-04 17:29:14 +03001499extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1500 mode_t mode);
Christoph Hellwig8c744fb2005-11-08 21:35:04 -08001501/*
Mark Fashehc4b929b2008-10-08 19:44:18 -04001502 * VFS FS_IOC_FIEMAP helper definitions.
1503 */
1504struct fiemap_extent_info {
1505 unsigned int fi_flags; /* Flags as passed from user */
1506 unsigned int fi_extents_mapped; /* Number of mapped extents */
1507 unsigned int fi_extents_max; /* Size of fiemap_extent array */
Namhyung Kimecf56322011-01-16 23:28:17 +09001508 struct fiemap_extent __user *fi_extents_start; /* Start of
1509 fiemap_extent array */
Mark Fashehc4b929b2008-10-08 19:44:18 -04001510};
1511int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1512 u64 phys, u64 len, u32 flags);
1513int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1514
1515/*
Linus Torvalds1da177e2005-04-16 15:20:36 -07001516 * File types
1517 *
1518 * NOTE! These match bits 12..15 of stat.st_mode
1519 * (ie "(i_mode >> 12) & 15").
1520 */
1521#define DT_UNKNOWN 0
1522#define DT_FIFO 1
1523#define DT_CHR 2
1524#define DT_DIR 4
1525#define DT_BLK 6
1526#define DT_REG 8
1527#define DT_LNK 10
1528#define DT_SOCK 12
1529#define DT_WHT 14
1530
Linus Torvalds1da177e2005-04-16 15:20:36 -07001531/*
1532 * This is the "filldir" function type, used by readdir() to let
1533 * the kernel specify what kind of dirent layout it wants to have.
1534 * This allows the kernel to read directories into kernel space or
1535 * to have different dirent layouts depending on the binary type.
1536 */
David Howellsafefdbb2006-10-03 01:13:46 -07001537typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
Al Viro08f85852007-10-08 13:26:20 -04001538struct block_device_operations;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001539
Linus Torvalds1da177e2005-04-16 15:20:36 -07001540/* These macros are for out of kernel modules to test that
1541 * the kernel supports the unlocked_ioctl and compat_ioctl
1542 * fields in struct file_operations. */
1543#define HAVE_COMPAT_IOCTL 1
1544#define HAVE_UNLOCKED_IOCTL 1
1545
1546/*
1547 * NOTE:
Arnd Bergmannb19dd422010-07-04 00:15:10 +02001548 * all file operations except setlease can be called without
1549 * the big kernel lock held in all filesystems.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001550 */
1551struct file_operations {
1552 struct module *owner;
1553 loff_t (*llseek) (struct file *, loff_t, int);
1554 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001555 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
Badari Pulavarty027445c2006-09-30 23:28:46 -07001556 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1557 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001558 int (*readdir) (struct file *, void *, filldir_t);
1559 unsigned int (*poll) (struct file *, struct poll_table_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001560 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1561 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1562 int (*mmap) (struct file *, struct vm_area_struct *);
1563 int (*open) (struct inode *, struct file *);
Miklos Szeredi75e1fcc2006-06-23 02:05:12 -07001564 int (*flush) (struct file *, fl_owner_t id);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001565 int (*release) (struct inode *, struct file *);
Christoph Hellwig7ea80852010-05-26 17:53:25 +02001566 int (*fsync) (struct file *, int datasync);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001567 int (*aio_fsync) (struct kiocb *, int datasync);
1568 int (*fasync) (int, struct file *, int);
1569 int (*lock) (struct file *, int, struct file_lock *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001570 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1571 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1572 int (*check_flags)(int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001573 int (*flock) (struct file *, int, struct file_lock *);
Jens Axboecbb7e572006-04-11 14:57:50 +02001574 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1575 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
J. Bruce Fieldsf9ffed22006-11-14 15:51:40 -05001576 int (*setlease)(struct file *, long, struct file_lock **);
Christoph Hellwig2fe17c12011-01-14 13:07:43 +01001577 long (*fallocate)(struct file *file, int mode, loff_t offset,
1578 loff_t len);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001579};
1580
1581struct inode_operations {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001582 struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
Nick Piggin44a7d7a2011-01-07 17:49:56 +11001583 void * (*follow_link) (struct dentry *, struct nameidata *);
Al Viro10556cb2011-06-20 19:28:19 -04001584 int (*permission) (struct inode *, int);
Al Viro7e401452011-06-20 19:12:17 -04001585 int (*check_acl)(struct inode *, int);
Nick Piggin44a7d7a2011-01-07 17:49:56 +11001586
1587 int (*readlink) (struct dentry *, char __user *,int);
1588 void (*put_link) (struct dentry *, struct nameidata *, void *);
1589
1590 int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001591 int (*link) (struct dentry *,struct inode *,struct dentry *);
1592 int (*unlink) (struct inode *,struct dentry *);
1593 int (*symlink) (struct inode *,struct dentry *,const char *);
1594 int (*mkdir) (struct inode *,struct dentry *,int);
1595 int (*rmdir) (struct inode *,struct dentry *);
1596 int (*mknod) (struct inode *,struct dentry *,int,dev_t);
1597 int (*rename) (struct inode *, struct dentry *,
1598 struct inode *, struct dentry *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001599 void (*truncate) (struct inode *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001600 int (*setattr) (struct dentry *, struct iattr *);
1601 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1602 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1603 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1604 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1605 int (*removexattr) (struct dentry *, const char *);
Badari Pulavartyf6b3ec22006-01-06 00:10:38 -08001606 void (*truncate_range)(struct inode *, loff_t, loff_t);
Mark Fashehc4b929b2008-10-08 19:44:18 -04001607 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1608 u64 len);
Nick Piggin44a7d7a2011-01-07 17:49:56 +11001609} ____cacheline_aligned;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001610
1611struct seq_file;
1612
Badari Pulavartyeed4e512006-09-30 23:28:49 -07001613ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1614 unsigned long nr_segs, unsigned long fast_segs,
1615 struct iovec *fast_pointer,
1616 struct iovec **ret_pointer);
1617
Linus Torvalds1da177e2005-04-16 15:20:36 -07001618extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1619extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1620extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1621 unsigned long, loff_t *);
1622extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1623 unsigned long, loff_t *);
1624
Linus Torvalds1da177e2005-04-16 15:20:36 -07001625struct super_operations {
1626 struct inode *(*alloc_inode)(struct super_block *sb);
1627 void (*destroy_inode)(struct inode *);
1628
Christoph Hellwigaa385722011-05-27 06:53:02 -04001629 void (*dirty_inode) (struct inode *, int flags);
Christoph Hellwiga9185b42010-03-05 09:21:37 +01001630 int (*write_inode) (struct inode *, struct writeback_control *wbc);
Al Viro45321ac2010-06-07 13:43:19 -04001631 int (*drop_inode) (struct inode *);
Al Virobe7ce412010-06-04 19:40:39 -04001632 void (*evict_inode) (struct inode *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001633 void (*put_super) (struct super_block *);
1634 void (*write_super) (struct super_block *);
1635 int (*sync_fs)(struct super_block *sb, int wait);
Takashi Satoc4be0c12009-01-09 16:40:58 -08001636 int (*freeze_fs) (struct super_block *);
1637 int (*unfreeze_fs) (struct super_block *);
David Howells726c3342006-06-23 02:02:58 -07001638 int (*statfs) (struct dentry *, struct kstatfs *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001639 int (*remount_fs) (struct super_block *, int *, char *);
Al Viro42faad92008-04-24 07:21:56 -04001640 void (*umount_begin) (struct super_block *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001641
1642 int (*show_options)(struct seq_file *, struct vfsmount *);
Al Viroc7f404b2011-03-16 06:59:40 -04001643 int (*show_devname)(struct seq_file *, struct vfsmount *);
1644 int (*show_path)(struct seq_file *, struct vfsmount *);
Chuck Leverb4629fe2006-03-20 13:44:12 -05001645 int (*show_stats)(struct seq_file *, struct vfsmount *);
Alexey Dobriyan0e51a722006-09-29 01:59:56 -07001646#ifdef CONFIG_QUOTA
Linus Torvalds1da177e2005-04-16 15:20:36 -07001647 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1648 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
Alexey Dobriyan0e51a722006-09-29 01:59:56 -07001649#endif
Theodore Ts'o87d8fe12009-01-03 09:47:09 -05001650 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001651};
1652
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001653/*
Dave Chinner250df6e2011-03-22 22:23:36 +11001654 * Inode state bits. Protected by inode->i_lock
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001655 *
1656 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1657 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1658 *
1659 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1660 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1661 * various stages of removing an inode.
1662 *
Christoph Hellwigeaff8072009-12-17 14:25:01 +01001663 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001664 *
Joern Engele7ca2d42008-02-06 01:36:59 -08001665 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1666 * fdatasync(). i_atime is the usual cause.
Jan Kara000cb482008-02-14 19:31:32 -08001667 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1668 * these changes separately from I_DIRTY_SYNC so that we
1669 * don't have to write inode on fdatasync() when only
1670 * mtime has changed in it.
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001671 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
Christoph Hellwigeaff8072009-12-17 14:25:01 +01001672 * I_NEW Serves as both a mutex and completion notification.
1673 * New inodes set I_NEW. If two processes both create
1674 * the same inode, one of them will release its inode and
1675 * wait for I_NEW to be released before returning.
1676 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1677 * also cause waiting on I_NEW, without I_NEW actually
1678 * being set. find_inode() uses this to prevent returning
1679 * nearly-dead inodes.
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001680 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1681 * is zero. I_FREEING must be set when I_WILL_FREE is
1682 * cleared.
1683 * I_FREEING Set when inode is about to be freed but still has dirty
1684 * pages or buffers attached or the inode itself is still
1685 * dirty.
Al Virob57922d2010-06-07 14:34:48 -04001686 * I_CLEAR Added by end_writeback(). In this state the inode is clean
Al Viroa4ffdde2010-06-02 17:38:30 -04001687 * and can be destroyed. Inode keeps I_FREEING.
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001688 *
1689 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1690 * prohibited for many purposes. iget() must wait for
1691 * the inode to be completely released, then create it
1692 * anew. Other functions will just ignore such inodes,
Christoph Hellwigeaff8072009-12-17 14:25:01 +01001693 * if appropriate. I_NEW is used for waiting.
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001694 *
Christoph Hellwigeaff8072009-12-17 14:25:01 +01001695 * I_SYNC Synchonized write of dirty inode data. The bits is
1696 * set during data writeback, and cleared with a wakeup
1697 * on the bit address once it is done.
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001698 *
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001699 * Q: What is the difference between I_WILL_FREE and I_FREEING?
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001700 */
Nick Piggin9e38d862010-10-23 06:55:17 -04001701#define I_DIRTY_SYNC (1 << 0)
1702#define I_DIRTY_DATASYNC (1 << 1)
1703#define I_DIRTY_PAGES (1 << 2)
Christoph Hellwigeaff8072009-12-17 14:25:01 +01001704#define __I_NEW 3
1705#define I_NEW (1 << __I_NEW)
Nick Piggin9e38d862010-10-23 06:55:17 -04001706#define I_WILL_FREE (1 << 4)
1707#define I_FREEING (1 << 5)
1708#define I_CLEAR (1 << 6)
Christoph Hellwigeaff8072009-12-17 14:25:01 +01001709#define __I_SYNC 7
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001710#define I_SYNC (1 << __I_SYNC)
Nick Piggin9e38d862010-10-23 06:55:17 -04001711#define I_REFERENCED (1 << 8)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001712
1713#define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1714
1715extern void __mark_inode_dirty(struct inode *, int);
1716static inline void mark_inode_dirty(struct inode *inode)
1717{
1718 __mark_inode_dirty(inode, I_DIRTY);
1719}
1720
1721static inline void mark_inode_dirty_sync(struct inode *inode)
1722{
1723 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1724}
1725
Dave Hansen71c42152007-06-23 17:16:42 -07001726/**
1727 * inc_nlink - directly increment an inode's link count
1728 * @inode: inode
1729 *
1730 * This is a low-level filesystem helper to replace any
1731 * direct filesystem manipulation of i_nlink. Currently,
1732 * it is only here for parity with dec_nlink().
1733 */
Dave Hansend8c76e62006-09-30 23:29:04 -07001734static inline void inc_nlink(struct inode *inode)
Alexey Dobriyan66d21732006-03-23 03:00:51 -08001735{
1736 inode->i_nlink++;
Dave Hansend8c76e62006-09-30 23:29:04 -07001737}
1738
1739static inline void inode_inc_link_count(struct inode *inode)
1740{
1741 inc_nlink(inode);
Alexey Dobriyan66d21732006-03-23 03:00:51 -08001742 mark_inode_dirty(inode);
1743}
1744
Dave Hansen71c42152007-06-23 17:16:42 -07001745/**
1746 * drop_nlink - directly drop an inode's link count
1747 * @inode: inode
1748 *
1749 * This is a low-level filesystem helper to replace any
1750 * direct filesystem manipulation of i_nlink. In cases
1751 * where we are attempting to track writes to the
1752 * filesystem, a decrement to zero means an imminent
1753 * write when the file is truncated and actually unlinked
1754 * on the filesystem.
1755 */
Dave Hansen9a53c3a2006-09-30 23:29:03 -07001756static inline void drop_nlink(struct inode *inode)
Alexey Dobriyan66d21732006-03-23 03:00:51 -08001757{
1758 inode->i_nlink--;
Dave Hansen9a53c3a2006-09-30 23:29:03 -07001759}
1760
Dave Hansen71c42152007-06-23 17:16:42 -07001761/**
1762 * clear_nlink - directly zero an inode's link count
1763 * @inode: inode
1764 *
1765 * This is a low-level filesystem helper to replace any
1766 * direct filesystem manipulation of i_nlink. See
1767 * drop_nlink() for why we care about i_nlink hitting zero.
1768 */
Dave Hansence71ec32006-09-30 23:29:06 -07001769static inline void clear_nlink(struct inode *inode)
1770{
1771 inode->i_nlink = 0;
1772}
1773
Dave Hansen9a53c3a2006-09-30 23:29:03 -07001774static inline void inode_dec_link_count(struct inode *inode)
1775{
1776 drop_nlink(inode);
Alexey Dobriyan66d21732006-03-23 03:00:51 -08001777 mark_inode_dirty(inode);
1778}
1779
Jean Noel Cordenner25ec56b2008-01-28 23:58:27 -05001780/**
1781 * inode_inc_iversion - increments i_version
1782 * @inode: inode that need to be updated
1783 *
1784 * Every time the inode is modified, the i_version field will be incremented.
1785 * The filesystem has to be mounted with i_version flag
1786 */
1787
1788static inline void inode_inc_iversion(struct inode *inode)
1789{
1790 spin_lock(&inode->i_lock);
1791 inode->i_version++;
1792 spin_unlock(&inode->i_lock);
1793}
1794
Christoph Hellwig869243a2006-01-09 20:52:03 -08001795extern void touch_atime(struct vfsmount *mnt, struct dentry *dentry);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001796static inline void file_accessed(struct file *file)
1797{
1798 if (!(file->f_flags & O_NOATIME))
Josef "Jeff" Sipek0f7fc9e2006-12-08 02:36:35 -08001799 touch_atime(file->f_path.mnt, file->f_path.dentry);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001800}
1801
1802int sync_inode(struct inode *inode, struct writeback_control *wbc);
Christoph Hellwigc3765012010-10-06 10:48:20 +02001803int sync_inode_metadata(struct inode *inode, int wait);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001804
Linus Torvalds1da177e2005-04-16 15:20:36 -07001805struct file_system_type {
1806 const char *name;
1807 int fs_flags;
Al Viroc96e41e2010-07-25 00:17:56 +04001808 struct dentry *(*mount) (struct file_system_type *, int,
1809 const char *, void *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001810 void (*kill_sb) (struct super_block *);
1811 struct module *owner;
1812 struct file_system_type * next;
1813 struct list_head fs_supers;
Peter Zijlstrad475fd42007-10-15 14:51:31 +02001814
Ingo Molnarcf516242006-07-03 00:25:27 -07001815 struct lock_class_key s_lock_key;
Arjan van de Ven897c6ff2006-07-03 00:25:28 -07001816 struct lock_class_key s_umount_key;
Roland Dreier51ee0492010-04-27 14:23:57 -07001817 struct lock_class_key s_vfs_rename_key;
Peter Zijlstrad475fd42007-10-15 14:51:31 +02001818
1819 struct lock_class_key i_lock_key;
1820 struct lock_class_key i_mutex_key;
Peter Zijlstra14358e62007-10-14 01:38:33 +02001821 struct lock_class_key i_mutex_dir_key;
Peter Zijlstrad475fd42007-10-15 14:51:31 +02001822 struct lock_class_key i_alloc_sem_key;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001823};
1824
Al Viroceefda62010-07-26 13:16:50 +04001825extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1826 void *data, int (*fill_super)(struct super_block *, void *, int));
Al Viro152a0832010-07-25 00:46:55 +04001827extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1828 int flags, const char *dev_name, void *data,
1829 int (*fill_super)(struct super_block *, void *, int));
Al Virofc14f2f2010-07-25 01:48:30 +04001830extern struct dentry *mount_single(struct file_system_type *fs_type,
1831 int flags, void *data,
1832 int (*fill_super)(struct super_block *, void *, int));
Al Viro3c26ff62010-07-25 11:46:36 +04001833extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1834 int flags, void *data,
1835 int (*fill_super)(struct super_block *, void *, int));
Linus Torvalds1da177e2005-04-16 15:20:36 -07001836void generic_shutdown_super(struct super_block *sb);
1837void kill_block_super(struct super_block *sb);
1838void kill_anon_super(struct super_block *sb);
1839void kill_litter_super(struct super_block *sb);
1840void deactivate_super(struct super_block *sb);
Al Viro74dbbdd2009-05-06 01:07:50 -04001841void deactivate_locked_super(struct super_block *sb);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001842int set_anon_super(struct super_block *s, void *data);
Al Viro0ee5dc62011-07-07 15:44:25 -04001843int get_anon_bdev(dev_t *);
1844void free_anon_bdev(dev_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001845struct super_block *sget(struct file_system_type *type,
1846 int (*test)(struct super_block *,void *),
1847 int (*set)(struct super_block *,void *),
1848 void *data);
Al Viro51139ad2010-07-25 23:47:46 +04001849extern struct dentry *mount_pseudo(struct file_system_type *, char *,
Al Viroc74a1cb2011-01-12 16:59:34 -05001850 const struct super_operations *ops,
1851 const struct dentry_operations *dops,
1852 unsigned long);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001853
Artem Bityutskiy140236b2010-06-10 13:56:33 +03001854static inline void sb_mark_dirty(struct super_block *sb)
1855{
1856 sb->s_dirt = 1;
1857}
1858static inline void sb_mark_clean(struct super_block *sb)
1859{
1860 sb->s_dirt = 0;
1861}
1862static inline int sb_is_dirty(struct super_block *sb)
1863{
1864 return sb->s_dirt;
1865}
1866
Linus Torvalds1da177e2005-04-16 15:20:36 -07001867/* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1868#define fops_get(fops) \
1869 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1870#define fops_put(fops) \
1871 do { if (fops) module_put((fops)->owner); } while(0)
1872
1873extern int register_filesystem(struct file_system_type *);
1874extern int unregister_filesystem(struct file_system_type *);
Pavel Emelyanov8bf97252007-10-18 23:40:02 -07001875extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1876#define kern_mount(type) kern_mount_data(type, NULL)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001877extern int may_umount_tree(struct vfsmount *);
1878extern int may_umount(struct vfsmount *);
1879extern long do_mount(char *, char *, char *, unsigned long, void *);
Al Viro589ff872009-04-18 03:28:19 -04001880extern struct vfsmount *collect_mounts(struct path *);
Al Viro8aec0802007-06-07 12:20:32 -04001881extern void drop_collected_mounts(struct vfsmount *);
Al Viro1f707132010-01-30 22:51:25 -05001882extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1883 struct vfsmount *);
Christoph Hellwigebabe9a2010-07-07 18:53:11 +02001884extern int vfs_statfs(struct path *, struct kstatfs *);
Al Viroc8b91ac2011-03-12 10:41:39 -05001885extern int user_statfs(const char __user *, struct kstatfs *);
1886extern int fd_statfs(int, struct kstatfs *);
Christoph Hellwigebabe9a2010-07-07 18:53:11 +02001887extern int statfs_by_dentry(struct dentry *, struct kstatfs *);
Josef Bacik18e9e512010-03-23 10:34:56 -04001888extern int freeze_super(struct super_block *super);
1889extern int thaw_super(struct super_block *super);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001890
Al Viroce3b0f82009-03-29 19:08:22 -04001891extern int current_umask(void);
1892
Miklos Szeredif87fd4c2006-01-16 22:14:23 -08001893/* /sys/fs */
Greg Kroah-Hartman00d26662007-10-29 14:17:23 -06001894extern struct kobject *fs_kobj;
Miklos Szeredif87fd4c2006-01-16 22:14:23 -08001895
Linus Torvalds435f49a2010-10-29 10:36:49 -07001896#define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001897extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1898
Linus Torvalds1da177e2005-04-16 15:20:36 -07001899#define FLOCK_VERIFY_READ 1
1900#define FLOCK_VERIFY_WRITE 2
1901
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001902#ifdef CONFIG_FILE_LOCKING
Linus Torvalds1da177e2005-04-16 15:20:36 -07001903extern int locks_mandatory_locked(struct inode *);
1904extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1905
1906/*
1907 * Candidates for mandatory locking have the setgid bit set
1908 * but no group execute bit - an otherwise meaningless combination.
1909 */
Pavel Emelyanova16877c2007-10-01 14:41:11 -07001910
1911static inline int __mandatory_lock(struct inode *ino)
1912{
1913 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1914}
1915
1916/*
1917 * ... and these candidates should be on MS_MANDLOCK mounted fs,
1918 * otherwise these will be advisory locks
1919 */
1920
1921static inline int mandatory_lock(struct inode *ino)
1922{
1923 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1924}
Linus Torvalds1da177e2005-04-16 15:20:36 -07001925
1926static inline int locks_verify_locked(struct inode *inode)
1927{
Pavel Emelyanova16877c2007-10-01 14:41:11 -07001928 if (mandatory_lock(inode))
Linus Torvalds1da177e2005-04-16 15:20:36 -07001929 return locks_mandatory_locked(inode);
1930 return 0;
1931}
1932
Linus Torvalds1da177e2005-04-16 15:20:36 -07001933static inline int locks_verify_truncate(struct inode *inode,
1934 struct file *filp,
1935 loff_t size)
1936{
Pavel Emelyanova16877c2007-10-01 14:41:11 -07001937 if (inode->i_flock && mandatory_lock(inode))
Linus Torvalds1da177e2005-04-16 15:20:36 -07001938 return locks_mandatory_area(
1939 FLOCK_VERIFY_WRITE, inode, filp,
1940 size < inode->i_size ? size : inode->i_size,
1941 (size < inode->i_size ? inode->i_size - size
1942 : size - inode->i_size)
1943 );
1944 return 0;
1945}
1946
1947static inline int break_lease(struct inode *inode, unsigned int mode)
1948{
1949 if (inode->i_flock)
1950 return __break_lease(inode, mode);
1951 return 0;
1952}
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001953#else /* !CONFIG_FILE_LOCKING */
Steven Whitehouseaf5df562009-01-20 10:29:46 +00001954static inline int locks_mandatory_locked(struct inode *inode)
1955{
1956 return 0;
1957}
1958
1959static inline int locks_mandatory_area(int rw, struct inode *inode,
1960 struct file *filp, loff_t offset,
1961 size_t count)
1962{
1963 return 0;
1964}
1965
1966static inline int __mandatory_lock(struct inode *inode)
1967{
1968 return 0;
1969}
1970
1971static inline int mandatory_lock(struct inode *inode)
1972{
1973 return 0;
1974}
1975
1976static inline int locks_verify_locked(struct inode *inode)
1977{
1978 return 0;
1979}
1980
1981static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
1982 size_t size)
1983{
1984 return 0;
1985}
1986
1987static inline int break_lease(struct inode *inode, unsigned int mode)
1988{
1989 return 0;
1990}
1991
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001992#endif /* CONFIG_FILE_LOCKING */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001993
1994/* fs/open.c */
1995
NeilBrown4a301312006-01-08 01:02:39 -08001996extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
1997 struct file *filp);
Ankit Jain3e63cbb2009-06-19 14:28:07 -04001998extern int do_fallocate(struct file *file, int mode, loff_t offset,
1999 loff_t len);
Jason Uhlenkott8e8a1402007-10-20 03:16:18 +02002000extern long do_sys_open(int dfd, const char __user *filename, int flags,
Ulrich Drepper5590ff02006-01-18 17:43:53 -08002001 int mode);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002002extern struct file *filp_open(const char *, int, int);
Al Viro73d049a2011-03-11 12:08:24 -05002003extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2004 const char *, int);
David Howells745ca242008-11-14 10:39:22 +11002005extern struct file * dentry_open(struct dentry *, struct vfsmount *, int,
2006 const struct cred *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002007extern int filp_close(struct file *, fl_owner_t id);
2008extern char * getname(const char __user *);
2009
Ankit Jain3e63cbb2009-06-19 14:28:07 -04002010/* fs/ioctl.c */
2011
2012extern int ioctl_preallocate(struct file *filp, void __user *argp);
2013
Linus Torvalds1da177e2005-04-16 15:20:36 -07002014/* fs/dcache.c */
2015extern void __init vfs_caches_init_early(void);
2016extern void __init vfs_caches_init(unsigned long);
2017
Christoph Lameterb86c0892006-12-06 20:32:57 -08002018extern struct kmem_cache *names_cachep;
2019
Vegard Nossum3446a8a2009-05-16 11:22:14 +02002020#define __getname_gfp(gfp) kmem_cache_alloc(names_cachep, (gfp))
2021#define __getname() __getname_gfp(GFP_KERNEL)
2022#define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002023#ifndef CONFIG_AUDITSYSCALL
2024#define putname(name) __putname(name)
2025#else
2026extern void putname(const char *name);
2027#endif
2028
David Howells93614012006-09-30 20:45:40 +02002029#ifdef CONFIG_BLOCK
Linus Torvalds1da177e2005-04-16 15:20:36 -07002030extern int register_blkdev(unsigned int, const char *);
Akinobu Mitaf4480242007-07-17 04:03:47 -07002031extern void unregister_blkdev(unsigned int, const char *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002032extern struct block_device *bdget(dev_t);
Alan Jenkinsdddac6a2009-07-29 21:07:55 +02002033extern struct block_device *bdgrab(struct block_device *bdev);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002034extern void bd_set_size(struct block_device *, loff_t size);
2035extern void bd_forget(struct inode *inode);
2036extern void bdput(struct block_device *);
Nick Piggin585d3bc2009-02-25 10:44:19 +01002037extern void invalidate_bdev(struct block_device *);
2038extern int sync_blockdev(struct block_device *bdev);
2039extern struct super_block *freeze_bdev(struct block_device *);
Eric Sandeenc2d75432009-03-31 15:23:46 -07002040extern void emergency_thaw_all(void);
Nick Piggin585d3bc2009-02-25 10:44:19 +01002041extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2042extern int fsync_bdev(struct block_device *);
David Howells93614012006-09-30 20:45:40 +02002043#else
2044static inline void bd_forget(struct inode *inode) {}
Al Viro47e44912009-04-01 07:07:16 -04002045static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2046static inline void invalidate_bdev(struct block_device *bdev) {}
2047
2048static inline struct super_block *freeze_bdev(struct block_device *sb)
2049{
2050 return NULL;
2051}
2052
2053static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2054{
2055 return 0;
2056}
David Howells93614012006-09-30 20:45:40 +02002057#endif
Jan Kara60b06802009-04-27 16:43:53 +02002058extern int sync_filesystem(struct super_block *);
David Howells93614012006-09-30 20:45:40 +02002059extern const struct file_operations def_blk_fops;
Arjan van de Ven4b6f5d22006-03-28 01:56:42 -08002060extern const struct file_operations def_chr_fops;
2061extern const struct file_operations bad_sock_fops;
2062extern const struct file_operations def_fifo_fops;
David Howells93614012006-09-30 20:45:40 +02002063#ifdef CONFIG_BLOCK
Linus Torvalds1da177e2005-04-16 15:20:36 -07002064extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
Al Viro56b26ad2008-09-19 03:17:36 -04002065extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002066extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
Tejun Heoe525fd82010-11-13 11:55:17 +01002067extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
Tejun Heod4d77622010-11-13 11:55:18 +01002068extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2069 void *holder);
2070extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2071 void *holder);
Tejun Heoe525fd82010-11-13 11:55:17 +01002072extern int blkdev_put(struct block_device *bdev, fmode_t mode);
Jun'ichi Nomura641dc632006-03-27 01:17:57 -08002073#ifdef CONFIG_SYSFS
Tejun Heoe09b4572010-11-13 11:55:17 +01002074extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
Tejun Heo49731ba2011-01-14 18:43:57 +01002075extern void bd_unlink_disk_holder(struct block_device *bdev,
2076 struct gendisk *disk);
Jun'ichi Nomura641dc632006-03-27 01:17:57 -08002077#else
Tejun Heoe09b4572010-11-13 11:55:17 +01002078static inline int bd_link_disk_holder(struct block_device *bdev,
2079 struct gendisk *disk)
2080{
2081 return 0;
2082}
Tejun Heo49731ba2011-01-14 18:43:57 +01002083static inline void bd_unlink_disk_holder(struct block_device *bdev,
2084 struct gendisk *disk)
2085{
2086}
Jun'ichi Nomura641dc632006-03-27 01:17:57 -08002087#endif
David Howells93614012006-09-30 20:45:40 +02002088#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002089
2090/* fs/char_dev.c */
Joe Korty68eef3b2006-03-31 02:30:32 -08002091#define CHRDEV_MAJOR_HASH_SIZE 255
Linus Torvalds1da177e2005-04-16 15:20:36 -07002092extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2093extern int register_chrdev_region(dev_t, unsigned, const char *);
Tejun Heo1905b1b2009-08-06 18:13:23 +09002094extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2095 unsigned int count, const char *name,
2096 const struct file_operations *fops);
2097extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2098 unsigned int count, const char *name);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002099extern void unregister_chrdev_region(dev_t, unsigned);
Joe Korty68eef3b2006-03-31 02:30:32 -08002100extern void chrdev_show(struct seq_file *,off_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002101
Tejun Heo1905b1b2009-08-06 18:13:23 +09002102static inline int register_chrdev(unsigned int major, const char *name,
2103 const struct file_operations *fops)
2104{
2105 return __register_chrdev(major, 0, 256, name, fops);
2106}
2107
2108static inline void unregister_chrdev(unsigned int major, const char *name)
2109{
2110 __unregister_chrdev(major, 0, 256, name);
2111}
2112
Linus Torvalds1da177e2005-04-16 15:20:36 -07002113/* fs/block_dev.c */
2114#define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
Tejun Heo1f014292008-08-25 19:47:23 +09002115#define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
David Howells93614012006-09-30 20:45:40 +02002116
2117#ifdef CONFIG_BLOCK
2118#define BLKDEV_MAJOR_HASH_SIZE 255
Linus Torvalds1da177e2005-04-16 15:20:36 -07002119extern const char *__bdevname(dev_t, char *buffer);
2120extern const char *bdevname(struct block_device *bdev, char *buffer);
2121extern struct block_device *lookup_bdev(const char *);
Joe Korty68eef3b2006-03-31 02:30:32 -08002122extern void blkdev_show(struct seq_file *,off_t);
Al Viro30c40d22008-02-22 19:50:45 -05002123
David Howells93614012006-09-30 20:45:40 +02002124#else
2125#define BLKDEV_MAJOR_HASH_SIZE 0
2126#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002127
2128extern void init_special_inode(struct inode *, umode_t, dev_t);
2129
2130/* Invalid inode operations -- fs/bad_inode.c */
2131extern void make_bad_inode(struct inode *);
2132extern int is_bad_inode(struct inode *);
2133
Denys Vlasenkod2d96482008-07-01 14:16:09 +02002134extern const struct file_operations read_pipefifo_fops;
2135extern const struct file_operations write_pipefifo_fops;
2136extern const struct file_operations rdwr_pipefifo_fops;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002137
2138extern int fs_may_remount_ro(struct super_block *);
2139
David Howells93614012006-09-30 20:45:40 +02002140#ifdef CONFIG_BLOCK
Linus Torvalds1da177e2005-04-16 15:20:36 -07002141/*
2142 * return READ, READA, or WRITE
2143 */
2144#define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2145
2146/*
2147 * return data direction, READ or WRITE
2148 */
2149#define bio_data_dir(bio) ((bio)->bi_rw & 1)
2150
Andrew Pattersonc3279d12008-09-04 14:27:25 -06002151extern void check_disk_size_change(struct gendisk *disk,
2152 struct block_device *bdev);
Andrew Patterson0c002c22008-09-04 14:27:20 -06002153extern int revalidate_disk(struct gendisk *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002154extern int check_disk_change(struct block_device *);
NeilBrown93b270f2011-02-24 17:25:47 +11002155extern int __invalidate_device(struct block_device *, bool);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002156extern int invalidate_partition(struct gendisk *, int);
David Howells93614012006-09-30 20:45:40 +02002157#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002158unsigned long invalidate_mapping_pages(struct address_space *mapping,
2159 pgoff_t start, pgoff_t end);
Anton Altaparmakov54bc4852007-02-10 01:45:38 -08002160
Linus Torvalds1da177e2005-04-16 15:20:36 -07002161static inline void invalidate_remote_inode(struct inode *inode)
2162{
2163 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2164 S_ISLNK(inode->i_mode))
Andrew Mortonfc0ecff2007-02-10 01:45:39 -08002165 invalidate_mapping_pages(inode->i_mapping, 0, -1);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002166}
2167extern int invalidate_inode_pages2(struct address_space *mapping);
2168extern int invalidate_inode_pages2_range(struct address_space *mapping,
2169 pgoff_t start, pgoff_t end);
2170extern int write_inode_now(struct inode *, int);
2171extern int filemap_fdatawrite(struct address_space *);
2172extern int filemap_flush(struct address_space *);
2173extern int filemap_fdatawait(struct address_space *);
Jan Karad3bccb62009-08-17 19:30:27 +02002174extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2175 loff_t lend);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002176extern int filemap_write_and_wait(struct address_space *mapping);
2177extern int filemap_write_and_wait_range(struct address_space *mapping,
2178 loff_t lstart, loff_t lend);
Andrew Mortonebcf28e2006-03-24 03:18:04 -08002179extern int __filemap_fdatawrite_range(struct address_space *mapping,
2180 loff_t start, loff_t end, int sync_mode);
Jan Karaf4c0a0f2008-07-11 19:27:31 -04002181extern int filemap_fdatawrite_range(struct address_space *mapping,
2182 loff_t start, loff_t end);
Andrew Mortonebcf28e2006-03-24 03:18:04 -08002183
Christoph Hellwig8018ab02010-03-22 17:32:25 +01002184extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2185 int datasync);
2186extern int vfs_fsync(struct file *file, int datasync);
Jan Kara148f9482009-08-17 19:52:36 +02002187extern int generic_write_sync(struct file *file, loff_t pos, loff_t count);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002188extern void sync_supers(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002189extern void emergency_sync(void);
2190extern void emergency_remount(void);
David Howells93614012006-09-30 20:45:40 +02002191#ifdef CONFIG_BLOCK
Linus Torvalds1da177e2005-04-16 15:20:36 -07002192extern sector_t bmap(struct inode *, sector_t);
David Howells93614012006-09-30 20:45:40 +02002193#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002194extern int notify_change(struct dentry *, struct iattr *);
Al Virof419a2e2008-07-22 00:07:17 -04002195extern int inode_permission(struct inode *, int);
Al Viro2830ba72011-06-20 19:16:29 -04002196extern int generic_permission(struct inode *, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002197
Miklos Szeredif696a362008-07-31 13:41:58 +02002198static inline bool execute_ok(struct inode *inode)
2199{
2200 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2201}
2202
Al Viro07b8ce12011-06-20 10:52:57 -04002203/*
2204 * get_write_access() gets write permission for a file.
2205 * put_write_access() releases this write permission.
2206 * This is used for regular files.
2207 * We cannot support write (and maybe mmap read-write shared) accesses and
2208 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2209 * can have the following values:
2210 * 0: no writers, no VM_DENYWRITE mappings
2211 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2212 * > 0: (i_writecount) users are writing to the file.
2213 *
2214 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2215 * except for the cases where we don't hold i_writecount yet. Then we need to
2216 * use {get,deny}_write_access() - these functions check the sign and refuse
2217 * to do the change if sign is wrong.
2218 */
2219static inline int get_write_access(struct inode *inode)
2220{
2221 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2222}
2223static inline int deny_write_access(struct file *file)
2224{
2225 struct inode *inode = file->f_path.dentry->d_inode;
2226 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2227}
Linus Torvalds1da177e2005-04-16 15:20:36 -07002228static inline void put_write_access(struct inode * inode)
2229{
2230 atomic_dec(&inode->i_writecount);
2231}
2232static inline void allow_write_access(struct file *file)
2233{
2234 if (file)
Josef "Jeff" Sipek0f7fc9e2006-12-08 02:36:35 -08002235 atomic_inc(&file->f_path.dentry->d_inode->i_writecount);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002236}
Mimi Zohara5c96eb2010-11-02 10:11:37 -04002237#ifdef CONFIG_IMA
2238static inline void i_readcount_dec(struct inode *inode)
2239{
2240 BUG_ON(!atomic_read(&inode->i_readcount));
2241 atomic_dec(&inode->i_readcount);
2242}
2243static inline void i_readcount_inc(struct inode *inode)
2244{
2245 atomic_inc(&inode->i_readcount);
2246}
2247#else
2248static inline void i_readcount_dec(struct inode *inode)
2249{
2250 return;
2251}
2252static inline void i_readcount_inc(struct inode *inode)
2253{
2254 return;
2255}
2256#endif
Ulrich Dreppered8cae82008-07-23 21:29:30 -07002257extern int do_pipe_flags(int *, int);
Ulrich Drepperbe61a862008-07-23 21:29:40 -07002258extern struct file *create_read_pipe(struct file *f, int flags);
2259extern struct file *create_write_pipe(int flags);
Andi Kleend6cbd282006-09-30 23:29:26 -07002260extern void free_write_pipe(struct file *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002261
Mimi Zohar6777d772009-08-21 14:32:48 -04002262extern int kernel_read(struct file *, loff_t, char *, unsigned long);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002263extern struct file * open_exec(const char *);
2264
2265/* fs/dcache.c -- generic fs support functions */
2266extern int is_subdir(struct dentry *, struct dentry *);
Al Viro2096f752010-01-30 13:16:21 -05002267extern int path_is_under(struct path *, struct path *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002268extern ino_t find_inode_number(struct dentry *, struct qstr *);
2269
2270#include <linux/err.h>
2271
2272/* needed for stackable file system support */
2273extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
2274
2275extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
2276
Christoph Hellwig54e34622009-08-07 14:38:25 -03002277extern int inode_init_always(struct super_block *, struct inode *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002278extern void inode_init_once(struct inode *);
Miklos Szeredi2aa15892011-02-23 13:49:47 +01002279extern void address_space_init_once(struct address_space *mapping);
Al Viro7de9c6e2010-10-23 11:11:40 -04002280extern void ihold(struct inode * inode);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002281extern void iput(struct inode *);
2282extern struct inode * igrab(struct inode *);
2283extern ino_t iunique(struct super_block *, ino_t);
2284extern int inode_needs_sync(struct inode *inode);
Al Viro45321ac2010-06-07 13:43:19 -04002285extern int generic_delete_inode(struct inode *inode);
2286extern int generic_drop_inode(struct inode *inode);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002287
Anton Altaparmakov88bd5122005-07-13 01:10:44 -07002288extern struct inode *ilookup5_nowait(struct super_block *sb,
2289 unsigned long hashval, int (*test)(struct inode *, void *),
2290 void *data);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002291extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2292 int (*test)(struct inode *, void *), void *data);
2293extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2294
2295extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2296extern struct inode * iget_locked(struct super_block *, unsigned long);
Al Viro261bca82008-12-30 01:48:21 -05002297extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2298extern int insert_inode_locked(struct inode *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002299extern void unlock_new_inode(struct inode *);
Christoph Hellwig85fe4022010-10-23 11:19:54 -04002300extern unsigned int get_next_ino(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002301
Linus Torvalds1da177e2005-04-16 15:20:36 -07002302extern void __iget(struct inode * inode);
David Howellsb46980f2008-02-07 00:15:27 -08002303extern void iget_failed(struct inode *);
Al Virob0683aa2010-06-04 20:55:25 -04002304extern void end_writeback(struct inode *);
Christoph Hellwig2e00c972009-08-07 14:38:29 -03002305extern void __destroy_inode(struct inode *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002306extern struct inode *new_inode(struct super_block *);
Nick Pigginff0c7d12011-01-07 17:49:50 +11002307extern void free_inode_nonrcu(struct inode *inode);
Jens Axboe01de85e2006-10-17 19:50:36 +02002308extern int should_remove_suid(struct dentry *);
Miklos Szeredi2f1936b2008-06-24 16:50:14 +02002309extern int file_remove_suid(struct file *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002310
2311extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2312extern void remove_inode_hash(struct inode *);
Christoph Hellwig646ec462010-10-23 07:15:32 -04002313static inline void insert_inode_hash(struct inode *inode)
2314{
Linus Torvalds1da177e2005-04-16 15:20:36 -07002315 __insert_inode_hash(inode, inode->i_ino);
2316}
Christoph Hellwig646ec462010-10-23 07:15:32 -04002317extern void inode_sb_list_add(struct inode *inode);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002318
David Howells93614012006-09-30 20:45:40 +02002319#ifdef CONFIG_BLOCK
Linus Torvalds1da177e2005-04-16 15:20:36 -07002320extern void submit_bio(int, struct bio *);
2321extern int bdev_read_only(struct block_device *);
David Howells93614012006-09-30 20:45:40 +02002322#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002323extern int set_blocksize(struct block_device *, int);
2324extern int sb_set_blocksize(struct super_block *, int);
2325extern int sb_min_blocksize(struct super_block *, int);
2326
2327extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2328extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2329extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002330int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
Badari Pulavarty027445c2006-09-30 23:28:46 -07002331extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
Jan Karae4dd9de2009-08-17 18:10:06 +02002332extern ssize_t __generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long,
2333 loff_t *);
Badari Pulavarty027445c2006-09-30 23:28:46 -07002334extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002335extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
2336 unsigned long *, loff_t, loff_t *, size_t, size_t);
2337extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
2338 unsigned long, loff_t, loff_t *, size_t, ssize_t);
2339extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2340extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
Dmitriy Monakhov0ceb3312007-05-08 00:23:02 -07002341extern int generic_segment_checks(const struct iovec *iov,
2342 unsigned long *nr_segs, size_t *count, int access_flags);
Andrew Morton88e6fae2006-04-11 13:59:36 +02002343
Christoph Hellwigeef99382009-08-20 17:43:41 +02002344/* fs/block_dev.c */
2345extern ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov,
2346 unsigned long nr_segs, loff_t pos);
Christoph Hellwig7ea80852010-05-26 17:53:25 +02002347extern int blkdev_fsync(struct file *filp, int datasync);
Christoph Hellwigeef99382009-08-20 17:43:41 +02002348
Andrew Morton88e6fae2006-04-11 13:59:36 +02002349/* fs/splice.c */
Jens Axboecbb7e572006-04-11 14:57:50 +02002350extern ssize_t generic_file_splice_read(struct file *, loff_t *,
Andrew Morton88e6fae2006-04-11 13:59:36 +02002351 struct pipe_inode_info *, size_t, unsigned int);
Miklos Szeredi68181732009-05-07 15:37:36 +02002352extern ssize_t default_file_splice_read(struct file *, loff_t *,
2353 struct pipe_inode_info *, size_t, unsigned int);
Andrew Morton88e6fae2006-04-11 13:59:36 +02002354extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
Jens Axboecbb7e572006-04-11 14:57:50 +02002355 struct file *, loff_t *, size_t, unsigned int);
Andrew Morton88e6fae2006-04-11 13:59:36 +02002356extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
Jens Axboecbb7e572006-04-11 14:57:50 +02002357 struct file *out, loff_t *, size_t len, unsigned int flags);
2358extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
Andrew Morton88e6fae2006-04-11 13:59:36 +02002359 size_t len, unsigned int flags);
2360
Linus Torvalds1da177e2005-04-16 15:20:36 -07002361extern void
2362file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
jan Blunckae6afc32010-05-26 14:44:48 -07002363extern loff_t noop_llseek(struct file *file, loff_t offset, int origin);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002364extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
2365extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
Andi Kleen9465efc2008-06-27 11:05:24 +02002366extern loff_t generic_file_llseek_unlocked(struct file *file, loff_t offset,
2367 int origin);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002368extern int generic_file_open(struct inode * inode, struct file * filp);
2369extern int nonseekable_open(struct inode * inode, struct file * filp);
2370
Carsten Otteceffc072005-06-23 22:05:25 -07002371#ifdef CONFIG_FS_XIP
Carsten Otteeb6fe0c2005-06-23 22:05:28 -07002372extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2373 loff_t *ppos);
Carsten Otteceffc072005-06-23 22:05:25 -07002374extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
Carsten Otteeb6fe0c2005-06-23 22:05:28 -07002375extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2376 size_t len, loff_t *ppos);
Carsten Otteceffc072005-06-23 22:05:25 -07002377extern int xip_truncate_page(struct address_space *mapping, loff_t from);
Carsten Otte6d791252005-06-23 22:05:26 -07002378#else
2379static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2380{
2381 return 0;
2382}
Carsten Otteceffc072005-06-23 22:05:25 -07002383#endif
2384
David Howells93614012006-09-30 20:45:40 +02002385#ifdef CONFIG_BLOCK
Josef Bacikfacd07b2010-05-23 11:00:55 -04002386typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2387 loff_t file_offset);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002388
2389enum {
Christoph Hellwig1e431f52009-11-03 16:44:53 +01002390 /* need locking between buffered and direct access */
2391 DIO_LOCKING = 0x01,
2392
2393 /* filesystem does not support filling holes */
2394 DIO_SKIP_HOLES = 0x02,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002395};
2396
Christoph Hellwigeafdc7d2010-06-04 11:29:53 +02002397void dio_end_io(struct bio *bio, int error);
npiggin@suse.de7bb46a62010-05-27 01:05:33 +10002398
Christoph Hellwigeafdc7d2010-06-04 11:29:53 +02002399ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2400 struct block_device *bdev, const struct iovec *iov, loff_t offset,
2401 unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
2402 dio_submit_t submit_io, int flags);
2403
Linus Torvalds1da177e2005-04-16 15:20:36 -07002404static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
2405 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
Badari Pulavarty1d8fa7a2006-03-26 01:38:02 -08002406 loff_t offset, unsigned long nr_segs, get_block_t get_block,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002407 dio_iodone_t end_io)
2408{
2409 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
Josef Bacikfacd07b2010-05-23 11:00:55 -04002410 nr_segs, get_block, end_io, NULL,
Christoph Hellwig1e431f52009-11-03 16:44:53 +01002411 DIO_LOCKING | DIO_SKIP_HOLES);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002412}
David Howells93614012006-09-30 20:45:40 +02002413#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002414
Arjan van de Ven4b6f5d22006-03-28 01:56:42 -08002415extern const struct file_operations generic_ro_fops;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002416
2417#define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2418
2419extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
2420extern int vfs_follow_link(struct nameidata *, const char *);
2421extern int page_readlink(struct dentry *, char __user *, int);
Linus Torvaldscc314ee2005-08-19 18:02:56 -07002422extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2423extern void page_put_link(struct dentry *, struct nameidata *, void *);
Kirill Korotaev0adb25d2006-03-11 03:27:13 -08002424extern int __page_symlink(struct inode *inode, const char *symname, int len,
Nick Piggin54566b22009-01-04 12:00:53 -08002425 int nofs);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002426extern int page_symlink(struct inode *inode, const char *symname, int len);
Arjan van de Venc5ef1c42007-02-12 00:55:40 -08002427extern const struct inode_operations page_symlink_inode_operations;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002428extern int generic_readlink(struct dentry *, char __user *, int);
2429extern void generic_fillattr(struct inode *, struct kstat *);
2430extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
Dmitry Monakhovb4627072009-12-14 15:21:12 +03002431void __inode_add_bytes(struct inode *inode, loff_t bytes);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002432void inode_add_bytes(struct inode *inode, loff_t bytes);
2433void inode_sub_bytes(struct inode *inode, loff_t bytes);
2434loff_t inode_get_bytes(struct inode *inode);
2435void inode_set_bytes(struct inode *inode, loff_t bytes);
2436
2437extern int vfs_readdir(struct file *, filldir_t, void *);
2438
David Howellsc7887322010-08-11 11:26:22 +01002439extern int vfs_stat(const char __user *, struct kstat *);
2440extern int vfs_lstat(const char __user *, struct kstat *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002441extern int vfs_fstat(unsigned int, struct kstat *);
David Howellsc7887322010-08-11 11:26:22 +01002442extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002443
Erez Zadokdeb21db2008-02-07 00:13:25 -08002444extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2445 unsigned long arg);
Steven Whitehousee9079cc2008-10-14 14:43:29 +01002446extern int __generic_block_fiemap(struct inode *inode,
Josef Bacik3a3076f2010-04-23 12:17:17 -04002447 struct fiemap_extent_info *fieinfo,
2448 loff_t start, loff_t len,
2449 get_block_t *get_block);
Josef Bacik68c9d702008-10-03 17:32:43 -04002450extern int generic_block_fiemap(struct inode *inode,
2451 struct fiemap_extent_info *fieinfo, u64 start,
2452 u64 len, get_block_t *get_block);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002453
Miklos Szeredic18479f2007-10-18 23:39:11 -07002454extern void get_filesystem(struct file_system_type *fs);
2455extern void put_filesystem(struct file_system_type *fs);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002456extern struct file_system_type *get_fs_type(const char *name);
2457extern struct super_block *get_super(struct block_device *);
Christoph Hellwig45042302009-08-03 23:28:35 +02002458extern struct super_block *get_active_super(struct block_device *bdev);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002459extern struct super_block *user_get_super(dev_t);
2460extern void drop_super(struct super_block *sb);
Al Viro01a05b32010-03-23 06:06:58 -04002461extern void iterate_supers(void (*)(struct super_block *, void *), void *);
Al Viro43e15cd2011-06-03 20:16:57 -04002462extern void iterate_supers_type(struct file_system_type *,
2463 void (*)(struct super_block *, void *), void *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002464
2465extern int dcache_dir_open(struct inode *, struct file *);
2466extern int dcache_dir_close(struct inode *, struct file *);
2467extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2468extern int dcache_readdir(struct file *, void *, filldir_t);
npiggin@suse.de7bb46a62010-05-27 01:05:33 +10002469extern int simple_setattr(struct dentry *, struct iattr *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002470extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
David Howells726c3342006-06-23 02:02:58 -07002471extern int simple_statfs(struct dentry *, struct kstatfs *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002472extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2473extern int simple_unlink(struct inode *, struct dentry *);
2474extern int simple_rmdir(struct inode *, struct dentry *);
2475extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
Christoph Hellwig1b061d92010-05-26 17:53:41 +02002476extern int noop_fsync(struct file *, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002477extern int simple_empty(struct dentry *);
2478extern int simple_readpage(struct file *file, struct page *page);
Nick Pigginafddba42007-10-16 01:25:01 -07002479extern int simple_write_begin(struct file *file, struct address_space *mapping,
2480 loff_t pos, unsigned len, unsigned flags,
2481 struct page **pagep, void **fsdata);
2482extern int simple_write_end(struct file *file, struct address_space *mapping,
2483 loff_t pos, unsigned len, unsigned copied,
2484 struct page *page, void *fsdata);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002485
2486extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
2487extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
Arjan van de Ven4b6f5d22006-03-28 01:56:42 -08002488extern const struct file_operations simple_dir_operations;
Arjan van de Venc5ef1c42007-02-12 00:55:40 -08002489extern const struct inode_operations simple_dir_inode_operations;
Arjan van de Ven4b6f5d22006-03-28 01:56:42 -08002490struct tree_descr { char *name; const struct file_operations *ops; int mode; };
Linus Torvalds1da177e2005-04-16 15:20:36 -07002491struct dentry *d_alloc_name(struct dentry *, const char *);
Roberto Sassu7d683a02010-06-03 11:58:28 +02002492extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
Trond Myklebust1f5ce9e2006-06-09 09:34:16 -04002493extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002494extern void simple_release_fs(struct vfsmount **mount, int *count);
2495
Akinobu Mita93b07112008-06-05 22:46:21 -07002496extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2497 loff_t *ppos, const void *from, size_t available);
Jiri Slaby6a727b42010-05-01 23:51:22 +02002498extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2499 const void __user *from, size_t count);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002500
Christoph Hellwig1b061d92010-05-26 17:53:41 +02002501extern int generic_file_fsync(struct file *, int);
Al Virod5aacad2009-06-07 14:56:44 -04002502
Patrick J. LoPresti30ca22c2010-07-22 15:03:41 -07002503extern int generic_check_addressable(unsigned, u64);
2504
Christoph Lametere965f962006-02-01 03:05:41 -08002505#ifdef CONFIG_MIGRATION
Christoph Lameter2d1db3b2006-06-23 02:03:33 -07002506extern int buffer_migrate_page(struct address_space *,
2507 struct page *, struct page *);
Christoph Lametere965f962006-02-01 03:05:41 -08002508#else
2509#define buffer_migrate_page NULL
2510#endif
2511
npiggin@suse.de25d9e2d2009-08-21 02:35:05 +10002512extern int inode_change_ok(const struct inode *, struct iattr *);
2513extern int inode_newsize_ok(const struct inode *, loff_t offset);
Christoph Hellwig6a1a90a2010-06-04 11:30:00 +02002514extern void setattr_copy(struct inode *inode, const struct iattr *attr);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002515
Christoph Hellwig870f4812006-01-09 20:52:01 -08002516extern void file_update_time(struct file *file);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002517
Miklos Szeredib3b304a2008-02-08 04:21:35 -08002518extern int generic_show_options(struct seq_file *m, struct vfsmount *mnt);
2519extern void save_mount_options(struct super_block *sb, char *options);
Al Viro2a32ceb2009-05-08 16:05:57 -04002520extern void replace_mount_options(struct super_block *sb, char *options);
Miklos Szeredib3b304a2008-02-08 04:21:35 -08002521
Linus Torvalds1da177e2005-04-16 15:20:36 -07002522static inline ino_t parent_ino(struct dentry *dentry)
2523{
2524 ino_t res;
2525
Nick Pigginb5c84bf2011-01-07 17:49:38 +11002526 /*
2527 * Don't strictly need d_lock here? If the parent ino could change
2528 * then surely we'd have a deeper race in the caller?
2529 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07002530 spin_lock(&dentry->d_lock);
2531 res = dentry->d_parent->d_inode->i_ino;
2532 spin_unlock(&dentry->d_lock);
2533 return res;
2534}
2535
Linus Torvalds1da177e2005-04-16 15:20:36 -07002536/* Transaction based IO helpers */
2537
2538/*
2539 * An argresp is stored in an allocated page and holds the
2540 * size of the argument or response, along with its content
2541 */
2542struct simple_transaction_argresp {
2543 ssize_t size;
2544 char data[0];
2545};
2546
2547#define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2548
2549char *simple_transaction_get(struct file *file, const char __user *buf,
2550 size_t size);
2551ssize_t simple_transaction_read(struct file *file, char __user *buf,
2552 size_t size, loff_t *pos);
2553int simple_transaction_release(struct inode *inode, struct file *file);
2554
Ingo Molnar76791ab2009-03-25 16:48:35 +01002555void simple_transaction_set(struct file *file, size_t n);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002556
Arnd Bergmannacaefc22005-05-18 14:40:59 +02002557/*
2558 * simple attribute files
2559 *
2560 * These attributes behave similar to those in sysfs:
2561 *
2562 * Writing to an attribute immediately sets a value, an open file can be
2563 * written to multiple times.
2564 *
2565 * Reading from an attribute creates a buffer from the value that might get
2566 * read with multiple read calls. When the attribute has been read
2567 * completely, no further read calls are possible until the file is opened
2568 * again.
2569 *
2570 * All attributes contain a text representation of a numeric value
2571 * that are accessed with the get() and set() functions.
2572 */
2573#define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2574static int __fops ## _open(struct inode *inode, struct file *file) \
2575{ \
2576 __simple_attr_check_format(__fmt, 0ull); \
2577 return simple_attr_open(inode, file, __get, __set, __fmt); \
2578} \
Alexey Dobriyan828c0952009-10-01 15:43:56 -07002579static const struct file_operations __fops = { \
Arnd Bergmannacaefc22005-05-18 14:40:59 +02002580 .owner = THIS_MODULE, \
2581 .open = __fops ## _open, \
Christoph Hellwig74bedc42008-02-08 04:20:28 -08002582 .release = simple_attr_release, \
Arnd Bergmannacaefc22005-05-18 14:40:59 +02002583 .read = simple_attr_read, \
2584 .write = simple_attr_write, \
Arnd Bergmann1ec55842010-08-15 21:50:52 +02002585 .llseek = generic_file_llseek, \
Arnd Bergmannacaefc22005-05-18 14:40:59 +02002586};
2587
2588static inline void __attribute__((format(printf, 1, 2)))
2589__simple_attr_check_format(const char *fmt, ...)
2590{
2591 /* don't do anything, just let the compiler check the arguments; */
2592}
2593
2594int simple_attr_open(struct inode *inode, struct file *file,
Christoph Hellwig8b88b092008-02-08 04:20:26 -08002595 int (*get)(void *, u64 *), int (*set)(void *, u64),
Arnd Bergmannacaefc22005-05-18 14:40:59 +02002596 const char *fmt);
Christoph Hellwig74bedc42008-02-08 04:20:28 -08002597int simple_attr_release(struct inode *inode, struct file *file);
Arnd Bergmannacaefc22005-05-18 14:40:59 +02002598ssize_t simple_attr_read(struct file *file, char __user *buf,
2599 size_t len, loff_t *ppos);
2600ssize_t simple_attr_write(struct file *file, const char __user *buf,
2601 size_t len, loff_t *ppos);
2602
Alexey Dobriyan4be28542007-10-16 23:26:21 -07002603struct ctl_table;
Alexey Dobriyan8d65af72009-09-23 15:57:19 -07002604int proc_nr_files(struct ctl_table *table, int write,
Adrian Bunk62239ac2007-07-17 04:03:45 -07002605 void __user *buffer, size_t *lenp, loff_t *ppos);
Christoph Hellwig312d3ca2010-10-10 05:36:23 -04002606int proc_nr_dentry(struct ctl_table *table, int write,
2607 void __user *buffer, size_t *lenp, loff_t *ppos);
Dave Chinnercffbc8a2010-10-23 05:03:02 -04002608int proc_nr_inodes(struct ctl_table *table, int write,
2609 void __user *buffer, size_t *lenp, loff_t *ppos);
Tetsuo Handa38e23c92009-04-09 20:17:52 +09002610int __init get_filesystem_list(char *buf);
Adrian Bunk62239ac2007-07-17 04:03:45 -07002611
Namhyung Kim3cd90ea2011-02-01 15:52:46 -08002612#define __FMODE_EXEC ((__force int) FMODE_EXEC)
Namhyung Kim1a44bc82011-02-01 15:52:46 -08002613#define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
2614
Al Viro6d125522009-12-24 06:58:56 -05002615#define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
Eric Parisecf081d2009-12-17 21:24:25 -05002616#define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
Namhyung Kim1a44bc82011-02-01 15:52:46 -08002617 (flag & __FMODE_NONOTIFY)))
Al Viro53009902009-12-19 10:15:07 -05002618
Andi Kleen69b45732011-05-28 08:25:51 -07002619static inline int is_sxid(mode_t mode)
2620{
2621 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
2622}
2623
2624static inline void inode_has_no_xattr(struct inode *inode)
2625{
Al Viro9e1f1de2011-06-03 18:24:58 -04002626 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
Andi Kleen69b45732011-05-28 08:25:51 -07002627 inode->i_flags |= S_NOSEC;
2628}
2629
Linus Torvalds1da177e2005-04-16 15:20:36 -07002630#endif /* __KERNEL__ */
2631#endif /* _LINUX_FS_H */