Linux kernel mirror (for testing)
git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel
os
linux
1#ifndef _LINUX_FS_H
2#define _LINUX_FS_H
3
4/*
5 * This file has definitions for some important file table
6 * structures etc.
7 */
8
9#include <linux/limits.h>
10#include <linux/ioctl.h>
11#include <linux/blk_types.h>
12#include <linux/types.h>
13
14/*
15 * It's silly to have NR_OPEN bigger than NR_FILE, but you can change
16 * the file limit at runtime and only root can increase the per-process
17 * nr_file rlimit, so it's safe to set up a ridiculously high absolute
18 * upper limit on files-per-process.
19 *
20 * Some programs (notably those using select()) may have to be
21 * recompiled to take full advantage of the new limits..
22 */
23
24/* Fixed constants first: */
25#undef NR_OPEN
26#define INR_OPEN_CUR 1024 /* Initial setting for nfile rlimits */
27#define INR_OPEN_MAX 4096 /* Hard limit for nfile rlimits */
28
29#define BLOCK_SIZE_BITS 10
30#define BLOCK_SIZE (1<<BLOCK_SIZE_BITS)
31
32#define SEEK_SET 0 /* seek relative to beginning of file */
33#define SEEK_CUR 1 /* seek relative to current file position */
34#define SEEK_END 2 /* seek relative to end of file */
35#define SEEK_DATA 3 /* seek to the next data */
36#define SEEK_HOLE 4 /* seek to the next hole */
37#define SEEK_MAX SEEK_HOLE
38
39struct fstrim_range {
40 __u64 start;
41 __u64 len;
42 __u64 minlen;
43};
44
45/* And dynamically-tunable limits and defaults: */
46struct files_stat_struct {
47 unsigned long nr_files; /* read only */
48 unsigned long nr_free_files; /* read only */
49 unsigned long max_files; /* tunable */
50};
51
52struct inodes_stat_t {
53 int nr_inodes;
54 int nr_unused;
55 int dummy[5]; /* padding for sysctl ABI compatibility */
56};
57
58
59#define NR_FILE 8192 /* this can well be larger on a larger system */
60
61#define MAY_EXEC 0x00000001
62#define MAY_WRITE 0x00000002
63#define MAY_READ 0x00000004
64#define MAY_APPEND 0x00000008
65#define MAY_ACCESS 0x00000010
66#define MAY_OPEN 0x00000020
67#define MAY_CHDIR 0x00000040
68/* called from RCU mode, don't block */
69#define MAY_NOT_BLOCK 0x00000080
70
71/*
72 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
73 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
74 */
75
76/* file is open for reading */
77#define FMODE_READ ((__force fmode_t)0x1)
78/* file is open for writing */
79#define FMODE_WRITE ((__force fmode_t)0x2)
80/* file is seekable */
81#define FMODE_LSEEK ((__force fmode_t)0x4)
82/* file can be accessed using pread */
83#define FMODE_PREAD ((__force fmode_t)0x8)
84/* file can be accessed using pwrite */
85#define FMODE_PWRITE ((__force fmode_t)0x10)
86/* File is opened for execution with sys_execve / sys_uselib */
87#define FMODE_EXEC ((__force fmode_t)0x20)
88/* File is opened with O_NDELAY (only set for block devices) */
89#define FMODE_NDELAY ((__force fmode_t)0x40)
90/* File is opened with O_EXCL (only set for block devices) */
91#define FMODE_EXCL ((__force fmode_t)0x80)
92/* File is opened using open(.., 3, ..) and is writeable only for ioctls
93 (specialy hack for floppy.c) */
94#define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
95/* 32bit hashes as llseek() offset (for directories) */
96#define FMODE_32BITHASH ((__force fmode_t)0x200)
97/* 64bit hashes as llseek() offset (for directories) */
98#define FMODE_64BITHASH ((__force fmode_t)0x400)
99
100/*
101 * Don't update ctime and mtime.
102 *
103 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
104 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
105 */
106#define FMODE_NOCMTIME ((__force fmode_t)0x800)
107
108/* Expect random access pattern */
109#define FMODE_RANDOM ((__force fmode_t)0x1000)
110
111/* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
112#define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
113
114/* File is opened with O_PATH; almost nothing can be done with it */
115#define FMODE_PATH ((__force fmode_t)0x4000)
116
117/* File was opened by fanotify and shouldn't generate fanotify events */
118#define FMODE_NONOTIFY ((__force fmode_t)0x1000000)
119
120/*
121 * The below are the various read and write types that we support. Some of
122 * them include behavioral modifiers that send information down to the
123 * block layer and IO scheduler. Terminology:
124 *
125 * The block layer uses device plugging to defer IO a little bit, in
126 * the hope that we will see more IO very shortly. This increases
127 * coalescing of adjacent IO and thus reduces the number of IOs we
128 * have to send to the device. It also allows for better queuing,
129 * if the IO isn't mergeable. If the caller is going to be waiting
130 * for the IO, then he must ensure that the device is unplugged so
131 * that the IO is dispatched to the driver.
132 *
133 * All IO is handled async in Linux. This is fine for background
134 * writes, but for reads or writes that someone waits for completion
135 * on, we want to notify the block layer and IO scheduler so that they
136 * know about it. That allows them to make better scheduling
137 * decisions. So when the below references 'sync' and 'async', it
138 * is referencing this priority hint.
139 *
140 * With that in mind, the available types are:
141 *
142 * READ A normal read operation. Device will be plugged.
143 * READ_SYNC A synchronous read. Device is not plugged, caller can
144 * immediately wait on this read without caring about
145 * unplugging.
146 * READA Used for read-ahead operations. Lower priority, and the
147 * block layer could (in theory) choose to ignore this
148 * request if it runs into resource problems.
149 * WRITE A normal async write. Device will be plugged.
150 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
151 * the hint that someone will be waiting on this IO
152 * shortly. The write equivalent of READ_SYNC.
153 * WRITE_ODIRECT Special case write for O_DIRECT only.
154 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
155 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
156 * non-volatile media on completion.
157 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
158 * by a cache flush and data is guaranteed to be on
159 * non-volatile media on completion.
160 *
161 */
162#define RW_MASK REQ_WRITE
163#define RWA_MASK REQ_RAHEAD
164
165#define READ 0
166#define WRITE RW_MASK
167#define READA RWA_MASK
168
169#define READ_SYNC (READ | REQ_SYNC)
170#define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
171#define WRITE_ODIRECT (WRITE | REQ_SYNC)
172#define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
173#define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
174#define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
175
176#define SEL_IN 1
177#define SEL_OUT 2
178#define SEL_EX 4
179
180/* public flags for file_system_type */
181#define FS_REQUIRES_DEV 1
182#define FS_BINARY_MOUNTDATA 2
183#define FS_HAS_SUBTYPE 4
184#define FS_REVAL_DOT 16384 /* Check the paths ".", ".." for staleness */
185#define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move()
186 * during rename() internally.
187 */
188
189/*
190 * These are the fs-independent mount-flags: up to 32 flags are supported
191 */
192#define MS_RDONLY 1 /* Mount read-only */
193#define MS_NOSUID 2 /* Ignore suid and sgid bits */
194#define MS_NODEV 4 /* Disallow access to device special files */
195#define MS_NOEXEC 8 /* Disallow program execution */
196#define MS_SYNCHRONOUS 16 /* Writes are synced at once */
197#define MS_REMOUNT 32 /* Alter flags of a mounted FS */
198#define MS_MANDLOCK 64 /* Allow mandatory locks on an FS */
199#define MS_DIRSYNC 128 /* Directory modifications are synchronous */
200#define MS_NOATIME 1024 /* Do not update access times. */
201#define MS_NODIRATIME 2048 /* Do not update directory access times */
202#define MS_BIND 4096
203#define MS_MOVE 8192
204#define MS_REC 16384
205#define MS_VERBOSE 32768 /* War is peace. Verbosity is silence.
206 MS_VERBOSE is deprecated. */
207#define MS_SILENT 32768
208#define MS_POSIXACL (1<<16) /* VFS does not apply the umask */
209#define MS_UNBINDABLE (1<<17) /* change to unbindable */
210#define MS_PRIVATE (1<<18) /* change to private */
211#define MS_SLAVE (1<<19) /* change to slave */
212#define MS_SHARED (1<<20) /* change to shared */
213#define MS_RELATIME (1<<21) /* Update atime relative to mtime/ctime. */
214#define MS_KERNMOUNT (1<<22) /* this is a kern_mount call */
215#define MS_I_VERSION (1<<23) /* Update inode I_version field */
216#define MS_STRICTATIME (1<<24) /* Always perform atime updates */
217#define MS_NOSEC (1<<28)
218#define MS_BORN (1<<29)
219#define MS_ACTIVE (1<<30)
220#define MS_NOUSER (1<<31)
221
222/*
223 * Superblock flags that can be altered by MS_REMOUNT
224 */
225#define MS_RMT_MASK (MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK|MS_I_VERSION)
226
227/*
228 * Old magic mount flag and mask
229 */
230#define MS_MGC_VAL 0xC0ED0000
231#define MS_MGC_MSK 0xffff0000
232
233/* Inode flags - they have nothing to superblock flags now */
234
235#define S_SYNC 1 /* Writes are synced at once */
236#define S_NOATIME 2 /* Do not update access times */
237#define S_APPEND 4 /* Append-only file */
238#define S_IMMUTABLE 8 /* Immutable file */
239#define S_DEAD 16 /* removed, but still open directory */
240#define S_NOQUOTA 32 /* Inode is not counted to quota */
241#define S_DIRSYNC 64 /* Directory modifications are synchronous */
242#define S_NOCMTIME 128 /* Do not update file c/mtime */
243#define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
244#define S_PRIVATE 512 /* Inode is fs-internal */
245#define S_IMA 1024 /* Inode has an associated IMA struct */
246#define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
247#define S_NOSEC 4096 /* no suid or xattr security attributes */
248
249/*
250 * Note that nosuid etc flags are inode-specific: setting some file-system
251 * flags just means all the inodes inherit those flags by default. It might be
252 * possible to override it selectively if you really wanted to with some
253 * ioctl() that is not currently implemented.
254 *
255 * Exception: MS_RDONLY is always applied to the entire file system.
256 *
257 * Unfortunately, it is possible to change a filesystems flags with it mounted
258 * with files in use. This means that all of the inodes will not have their
259 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
260 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
261 */
262#define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))
263
264#define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
265#define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
266 ((inode)->i_flags & S_SYNC))
267#define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
268 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
269#define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
270#define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
271#define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
272
273#define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
274#define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
275#define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
276#define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
277
278#define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
279#define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
280#define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
281#define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
282#define IS_IMA(inode) ((inode)->i_flags & S_IMA)
283#define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
284#define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
285
286/* the read-only stuff doesn't really belong here, but any other place is
287 probably as bad and I don't want to create yet another include file. */
288
289#define BLKROSET _IO(0x12,93) /* set device read-only (0 = read-write) */
290#define BLKROGET _IO(0x12,94) /* get read-only status (0 = read_write) */
291#define BLKRRPART _IO(0x12,95) /* re-read partition table */
292#define BLKGETSIZE _IO(0x12,96) /* return device size /512 (long *arg) */
293#define BLKFLSBUF _IO(0x12,97) /* flush buffer cache */
294#define BLKRASET _IO(0x12,98) /* set read ahead for block device */
295#define BLKRAGET _IO(0x12,99) /* get current read ahead setting */
296#define BLKFRASET _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
297#define BLKFRAGET _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
298#define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
299#define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
300#define BLKSSZGET _IO(0x12,104)/* get block device sector size */
301#if 0
302#define BLKPG _IO(0x12,105)/* See blkpg.h */
303
304/* Some people are morons. Do not use sizeof! */
305
306#define BLKELVGET _IOR(0x12,106,size_t)/* elevator get */
307#define BLKELVSET _IOW(0x12,107,size_t)/* elevator set */
308/* This was here just to show that the number is taken -
309 probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
310#endif
311/* A jump here: 108-111 have been used for various private purposes. */
312#define BLKBSZGET _IOR(0x12,112,size_t)
313#define BLKBSZSET _IOW(0x12,113,size_t)
314#define BLKGETSIZE64 _IOR(0x12,114,size_t) /* return device size in bytes (u64 *arg) */
315#define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup)
316#define BLKTRACESTART _IO(0x12,116)
317#define BLKTRACESTOP _IO(0x12,117)
318#define BLKTRACETEARDOWN _IO(0x12,118)
319#define BLKDISCARD _IO(0x12,119)
320#define BLKIOMIN _IO(0x12,120)
321#define BLKIOOPT _IO(0x12,121)
322#define BLKALIGNOFF _IO(0x12,122)
323#define BLKPBSZGET _IO(0x12,123)
324#define BLKDISCARDZEROES _IO(0x12,124)
325#define BLKSECDISCARD _IO(0x12,125)
326#define BLKROTATIONAL _IO(0x12,126)
327
328#define BMAP_IOCTL 1 /* obsolete - kept for compatibility */
329#define FIBMAP _IO(0x00,1) /* bmap access */
330#define FIGETBSZ _IO(0x00,2) /* get the block size used for bmap */
331#define FIFREEZE _IOWR('X', 119, int) /* Freeze */
332#define FITHAW _IOWR('X', 120, int) /* Thaw */
333#define FITRIM _IOWR('X', 121, struct fstrim_range) /* Trim */
334
335#define FS_IOC_GETFLAGS _IOR('f', 1, long)
336#define FS_IOC_SETFLAGS _IOW('f', 2, long)
337#define FS_IOC_GETVERSION _IOR('v', 1, long)
338#define FS_IOC_SETVERSION _IOW('v', 2, long)
339#define FS_IOC_FIEMAP _IOWR('f', 11, struct fiemap)
340#define FS_IOC32_GETFLAGS _IOR('f', 1, int)
341#define FS_IOC32_SETFLAGS _IOW('f', 2, int)
342#define FS_IOC32_GETVERSION _IOR('v', 1, int)
343#define FS_IOC32_SETVERSION _IOW('v', 2, int)
344
345/*
346 * Inode flags (FS_IOC_GETFLAGS / FS_IOC_SETFLAGS)
347 */
348#define FS_SECRM_FL 0x00000001 /* Secure deletion */
349#define FS_UNRM_FL 0x00000002 /* Undelete */
350#define FS_COMPR_FL 0x00000004 /* Compress file */
351#define FS_SYNC_FL 0x00000008 /* Synchronous updates */
352#define FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
353#define FS_APPEND_FL 0x00000020 /* writes to file may only append */
354#define FS_NODUMP_FL 0x00000040 /* do not dump file */
355#define FS_NOATIME_FL 0x00000080 /* do not update atime */
356/* Reserved for compression usage... */
357#define FS_DIRTY_FL 0x00000100
358#define FS_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
359#define FS_NOCOMP_FL 0x00000400 /* Don't compress */
360#define FS_ECOMPR_FL 0x00000800 /* Compression error */
361/* End compression flags --- maybe not all used */
362#define FS_BTREE_FL 0x00001000 /* btree format dir */
363#define FS_INDEX_FL 0x00001000 /* hash-indexed directory */
364#define FS_IMAGIC_FL 0x00002000 /* AFS directory */
365#define FS_JOURNAL_DATA_FL 0x00004000 /* Reserved for ext3 */
366#define FS_NOTAIL_FL 0x00008000 /* file tail should not be merged */
367#define FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
368#define FS_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
369#define FS_EXTENT_FL 0x00080000 /* Extents */
370#define FS_DIRECTIO_FL 0x00100000 /* Use direct i/o */
371#define FS_NOCOW_FL 0x00800000 /* Do not cow file */
372#define FS_RESERVED_FL 0x80000000 /* reserved for ext2 lib */
373
374#define FS_FL_USER_VISIBLE 0x0003DFFF /* User visible flags */
375#define FS_FL_USER_MODIFIABLE 0x000380FF /* User modifiable flags */
376
377
378#define SYNC_FILE_RANGE_WAIT_BEFORE 1
379#define SYNC_FILE_RANGE_WRITE 2
380#define SYNC_FILE_RANGE_WAIT_AFTER 4
381
382#ifdef __KERNEL__
383
384#include <linux/linkage.h>
385#include <linux/wait.h>
386#include <linux/kdev_t.h>
387#include <linux/dcache.h>
388#include <linux/path.h>
389#include <linux/stat.h>
390#include <linux/cache.h>
391#include <linux/list.h>
392#include <linux/radix-tree.h>
393#include <linux/prio_tree.h>
394#include <linux/init.h>
395#include <linux/pid.h>
396#include <linux/bug.h>
397#include <linux/mutex.h>
398#include <linux/capability.h>
399#include <linux/semaphore.h>
400#include <linux/fiemap.h>
401#include <linux/rculist_bl.h>
402#include <linux/atomic.h>
403#include <linux/shrinker.h>
404#include <linux/migrate_mode.h>
405
406#include <asm/byteorder.h>
407
408struct export_operations;
409struct hd_geometry;
410struct iovec;
411struct nameidata;
412struct kiocb;
413struct kobject;
414struct pipe_inode_info;
415struct poll_table_struct;
416struct kstatfs;
417struct vm_area_struct;
418struct vfsmount;
419struct cred;
420
421extern void __init inode_init(void);
422extern void __init inode_init_early(void);
423extern void __init files_init(unsigned long);
424
425extern struct files_stat_struct files_stat;
426extern unsigned long get_max_files(void);
427extern int sysctl_nr_open;
428extern struct inodes_stat_t inodes_stat;
429extern int leases_enable, lease_break_time;
430
431struct buffer_head;
432typedef int (get_block_t)(struct inode *inode, sector_t iblock,
433 struct buffer_head *bh_result, int create);
434typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
435 ssize_t bytes, void *private, int ret,
436 bool is_async);
437
438/*
439 * Attribute flags. These should be or-ed together to figure out what
440 * has been changed!
441 */
442#define ATTR_MODE (1 << 0)
443#define ATTR_UID (1 << 1)
444#define ATTR_GID (1 << 2)
445#define ATTR_SIZE (1 << 3)
446#define ATTR_ATIME (1 << 4)
447#define ATTR_MTIME (1 << 5)
448#define ATTR_CTIME (1 << 6)
449#define ATTR_ATIME_SET (1 << 7)
450#define ATTR_MTIME_SET (1 << 8)
451#define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
452#define ATTR_ATTR_FLAG (1 << 10)
453#define ATTR_KILL_SUID (1 << 11)
454#define ATTR_KILL_SGID (1 << 12)
455#define ATTR_FILE (1 << 13)
456#define ATTR_KILL_PRIV (1 << 14)
457#define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
458#define ATTR_TIMES_SET (1 << 16)
459
460/*
461 * This is the Inode Attributes structure, used for notify_change(). It
462 * uses the above definitions as flags, to know which values have changed.
463 * Also, in this manner, a Filesystem can look at only the values it cares
464 * about. Basically, these are the attributes that the VFS layer can
465 * request to change from the FS layer.
466 *
467 * Derek Atkins <warlord@MIT.EDU> 94-10-20
468 */
469struct iattr {
470 unsigned int ia_valid;
471 umode_t ia_mode;
472 uid_t ia_uid;
473 gid_t ia_gid;
474 loff_t ia_size;
475 struct timespec ia_atime;
476 struct timespec ia_mtime;
477 struct timespec ia_ctime;
478
479 /*
480 * Not an attribute, but an auxiliary info for filesystems wanting to
481 * implement an ftruncate() like method. NOTE: filesystem should
482 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
483 */
484 struct file *ia_file;
485};
486
487/*
488 * Includes for diskquotas.
489 */
490#include <linux/quota.h>
491
492/**
493 * enum positive_aop_returns - aop return codes with specific semantics
494 *
495 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
496 * completed, that the page is still locked, and
497 * should be considered active. The VM uses this hint
498 * to return the page to the active list -- it won't
499 * be a candidate for writeback again in the near
500 * future. Other callers must be careful to unlock
501 * the page if they get this return. Returned by
502 * writepage();
503 *
504 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
505 * unlocked it and the page might have been truncated.
506 * The caller should back up to acquiring a new page and
507 * trying again. The aop will be taking reasonable
508 * precautions not to livelock. If the caller held a page
509 * reference, it should drop it before retrying. Returned
510 * by readpage().
511 *
512 * address_space_operation functions return these large constants to indicate
513 * special semantics to the caller. These are much larger than the bytes in a
514 * page to allow for functions that return the number of bytes operated on in a
515 * given page.
516 */
517
518enum positive_aop_returns {
519 AOP_WRITEPAGE_ACTIVATE = 0x80000,
520 AOP_TRUNCATED_PAGE = 0x80001,
521};
522
523#define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
524#define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
525#define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
526 * helper code (eg buffer layer)
527 * to clear GFP_FS from alloc */
528
529/*
530 * oh the beauties of C type declarations.
531 */
532struct page;
533struct address_space;
534struct writeback_control;
535
536struct iov_iter {
537 const struct iovec *iov;
538 unsigned long nr_segs;
539 size_t iov_offset;
540 size_t count;
541};
542
543size_t iov_iter_copy_from_user_atomic(struct page *page,
544 struct iov_iter *i, unsigned long offset, size_t bytes);
545size_t iov_iter_copy_from_user(struct page *page,
546 struct iov_iter *i, unsigned long offset, size_t bytes);
547void iov_iter_advance(struct iov_iter *i, size_t bytes);
548int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes);
549size_t iov_iter_single_seg_count(struct iov_iter *i);
550
551static inline void iov_iter_init(struct iov_iter *i,
552 const struct iovec *iov, unsigned long nr_segs,
553 size_t count, size_t written)
554{
555 i->iov = iov;
556 i->nr_segs = nr_segs;
557 i->iov_offset = 0;
558 i->count = count + written;
559
560 iov_iter_advance(i, written);
561}
562
563static inline size_t iov_iter_count(struct iov_iter *i)
564{
565 return i->count;
566}
567
568/*
569 * "descriptor" for what we're up to with a read.
570 * This allows us to use the same read code yet
571 * have multiple different users of the data that
572 * we read from a file.
573 *
574 * The simplest case just copies the data to user
575 * mode.
576 */
577typedef struct {
578 size_t written;
579 size_t count;
580 union {
581 char __user *buf;
582 void *data;
583 } arg;
584 int error;
585} read_descriptor_t;
586
587typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
588 unsigned long, unsigned long);
589
590struct address_space_operations {
591 int (*writepage)(struct page *page, struct writeback_control *wbc);
592 int (*readpage)(struct file *, struct page *);
593
594 /* Write back some dirty pages from this mapping. */
595 int (*writepages)(struct address_space *, struct writeback_control *);
596
597 /* Set a page dirty. Return true if this dirtied it */
598 int (*set_page_dirty)(struct page *page);
599
600 int (*readpages)(struct file *filp, struct address_space *mapping,
601 struct list_head *pages, unsigned nr_pages);
602
603 int (*write_begin)(struct file *, struct address_space *mapping,
604 loff_t pos, unsigned len, unsigned flags,
605 struct page **pagep, void **fsdata);
606 int (*write_end)(struct file *, struct address_space *mapping,
607 loff_t pos, unsigned len, unsigned copied,
608 struct page *page, void *fsdata);
609
610 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
611 sector_t (*bmap)(struct address_space *, sector_t);
612 void (*invalidatepage) (struct page *, unsigned long);
613 int (*releasepage) (struct page *, gfp_t);
614 void (*freepage)(struct page *);
615 ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
616 loff_t offset, unsigned long nr_segs);
617 int (*get_xip_mem)(struct address_space *, pgoff_t, int,
618 void **, unsigned long *);
619 /*
620 * migrate the contents of a page to the specified target. If sync
621 * is false, it must not block.
622 */
623 int (*migratepage) (struct address_space *,
624 struct page *, struct page *, enum migrate_mode);
625 int (*launder_page) (struct page *);
626 int (*is_partially_uptodate) (struct page *, read_descriptor_t *,
627 unsigned long);
628 int (*error_remove_page)(struct address_space *, struct page *);
629};
630
631extern const struct address_space_operations empty_aops;
632
633/*
634 * pagecache_write_begin/pagecache_write_end must be used by general code
635 * to write into the pagecache.
636 */
637int pagecache_write_begin(struct file *, struct address_space *mapping,
638 loff_t pos, unsigned len, unsigned flags,
639 struct page **pagep, void **fsdata);
640
641int pagecache_write_end(struct file *, struct address_space *mapping,
642 loff_t pos, unsigned len, unsigned copied,
643 struct page *page, void *fsdata);
644
645struct backing_dev_info;
646struct address_space {
647 struct inode *host; /* owner: inode, block_device */
648 struct radix_tree_root page_tree; /* radix tree of all pages */
649 spinlock_t tree_lock; /* and lock protecting it */
650 unsigned int i_mmap_writable;/* count VM_SHARED mappings */
651 struct prio_tree_root i_mmap; /* tree of private and shared mappings */
652 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
653 struct mutex i_mmap_mutex; /* protect tree, count, list */
654 /* Protected by tree_lock together with the radix tree */
655 unsigned long nrpages; /* number of total pages */
656 pgoff_t writeback_index;/* writeback starts here */
657 const struct address_space_operations *a_ops; /* methods */
658 unsigned long flags; /* error bits/gfp mask */
659 struct backing_dev_info *backing_dev_info; /* device readahead, etc */
660 spinlock_t private_lock; /* for use by the address_space */
661 struct list_head private_list; /* ditto */
662 struct address_space *assoc_mapping; /* ditto */
663} __attribute__((aligned(sizeof(long))));
664 /*
665 * On most architectures that alignment is already the case; but
666 * must be enforced here for CRIS, to let the least significant bit
667 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
668 */
669struct request_queue;
670
671struct block_device {
672 dev_t bd_dev; /* not a kdev_t - it's a search key */
673 int bd_openers;
674 struct inode * bd_inode; /* will die */
675 struct super_block * bd_super;
676 struct mutex bd_mutex; /* open/close mutex */
677 struct list_head bd_inodes;
678 void * bd_claiming;
679 void * bd_holder;
680 int bd_holders;
681 bool bd_write_holder;
682#ifdef CONFIG_SYSFS
683 struct list_head bd_holder_disks;
684#endif
685 struct block_device * bd_contains;
686 unsigned bd_block_size;
687 struct hd_struct * bd_part;
688 /* number of times partitions within this device have been opened. */
689 unsigned bd_part_count;
690 int bd_invalidated;
691 struct gendisk * bd_disk;
692 struct request_queue * bd_queue;
693 struct list_head bd_list;
694 /*
695 * Private data. You must have bd_claim'ed the block_device
696 * to use this. NOTE: bd_claim allows an owner to claim
697 * the same device multiple times, the owner must take special
698 * care to not mess up bd_private for that case.
699 */
700 unsigned long bd_private;
701
702 /* The counter of freeze processes */
703 int bd_fsfreeze_count;
704 /* Mutex for freeze */
705 struct mutex bd_fsfreeze_mutex;
706};
707
708/*
709 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
710 * radix trees
711 */
712#define PAGECACHE_TAG_DIRTY 0
713#define PAGECACHE_TAG_WRITEBACK 1
714#define PAGECACHE_TAG_TOWRITE 2
715
716int mapping_tagged(struct address_space *mapping, int tag);
717
718/*
719 * Might pages of this file be mapped into userspace?
720 */
721static inline int mapping_mapped(struct address_space *mapping)
722{
723 return !prio_tree_empty(&mapping->i_mmap) ||
724 !list_empty(&mapping->i_mmap_nonlinear);
725}
726
727/*
728 * Might pages of this file have been modified in userspace?
729 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
730 * marks vma as VM_SHARED if it is shared, and the file was opened for
731 * writing i.e. vma may be mprotected writable even if now readonly.
732 */
733static inline int mapping_writably_mapped(struct address_space *mapping)
734{
735 return mapping->i_mmap_writable != 0;
736}
737
738/*
739 * Use sequence counter to get consistent i_size on 32-bit processors.
740 */
741#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
742#include <linux/seqlock.h>
743#define __NEED_I_SIZE_ORDERED
744#define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
745#else
746#define i_size_ordered_init(inode) do { } while (0)
747#endif
748
749struct posix_acl;
750#define ACL_NOT_CACHED ((void *)(-1))
751
752#define IOP_FASTPERM 0x0001
753#define IOP_LOOKUP 0x0002
754#define IOP_NOFOLLOW 0x0004
755
756/*
757 * Keep mostly read-only and often accessed (especially for
758 * the RCU path lookup and 'stat' data) fields at the beginning
759 * of the 'struct inode'
760 */
761struct inode {
762 umode_t i_mode;
763 unsigned short i_opflags;
764 uid_t i_uid;
765 gid_t i_gid;
766 unsigned int i_flags;
767
768#ifdef CONFIG_FS_POSIX_ACL
769 struct posix_acl *i_acl;
770 struct posix_acl *i_default_acl;
771#endif
772
773 const struct inode_operations *i_op;
774 struct super_block *i_sb;
775 struct address_space *i_mapping;
776
777#ifdef CONFIG_SECURITY
778 void *i_security;
779#endif
780
781 /* Stat data, not accessed from path walking */
782 unsigned long i_ino;
783 /*
784 * Filesystems may only read i_nlink directly. They shall use the
785 * following functions for modification:
786 *
787 * (set|clear|inc|drop)_nlink
788 * inode_(inc|dec)_link_count
789 */
790 union {
791 const unsigned int i_nlink;
792 unsigned int __i_nlink;
793 };
794 dev_t i_rdev;
795 struct timespec i_atime;
796 struct timespec i_mtime;
797 struct timespec i_ctime;
798 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
799 unsigned short i_bytes;
800 blkcnt_t i_blocks;
801 loff_t i_size;
802
803#ifdef __NEED_I_SIZE_ORDERED
804 seqcount_t i_size_seqcount;
805#endif
806
807 /* Misc */
808 unsigned long i_state;
809 struct mutex i_mutex;
810
811 unsigned long dirtied_when; /* jiffies of first dirtying */
812
813 struct hlist_node i_hash;
814 struct list_head i_wb_list; /* backing dev IO list */
815 struct list_head i_lru; /* inode LRU list */
816 struct list_head i_sb_list;
817 union {
818 struct list_head i_dentry;
819 struct rcu_head i_rcu;
820 };
821 atomic_t i_count;
822 unsigned int i_blkbits;
823 u64 i_version;
824 atomic_t i_dio_count;
825 atomic_t i_writecount;
826 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
827 struct file_lock *i_flock;
828 struct address_space i_data;
829#ifdef CONFIG_QUOTA
830 struct dquot *i_dquot[MAXQUOTAS];
831#endif
832 struct list_head i_devices;
833 union {
834 struct pipe_inode_info *i_pipe;
835 struct block_device *i_bdev;
836 struct cdev *i_cdev;
837 };
838
839 __u32 i_generation;
840
841#ifdef CONFIG_FSNOTIFY
842 __u32 i_fsnotify_mask; /* all events this inode cares about */
843 struct hlist_head i_fsnotify_marks;
844#endif
845
846#ifdef CONFIG_IMA
847 atomic_t i_readcount; /* struct files open RO */
848#endif
849 void *i_private; /* fs or device private pointer */
850};
851
852static inline int inode_unhashed(struct inode *inode)
853{
854 return hlist_unhashed(&inode->i_hash);
855}
856
857/*
858 * inode->i_mutex nesting subclasses for the lock validator:
859 *
860 * 0: the object of the current VFS operation
861 * 1: parent
862 * 2: child/target
863 * 3: quota file
864 *
865 * The locking order between these classes is
866 * parent -> child -> normal -> xattr -> quota
867 */
868enum inode_i_mutex_lock_class
869{
870 I_MUTEX_NORMAL,
871 I_MUTEX_PARENT,
872 I_MUTEX_CHILD,
873 I_MUTEX_XATTR,
874 I_MUTEX_QUOTA
875};
876
877/*
878 * NOTE: in a 32bit arch with a preemptable kernel and
879 * an UP compile the i_size_read/write must be atomic
880 * with respect to the local cpu (unlike with preempt disabled),
881 * but they don't need to be atomic with respect to other cpus like in
882 * true SMP (so they need either to either locally disable irq around
883 * the read or for example on x86 they can be still implemented as a
884 * cmpxchg8b without the need of the lock prefix). For SMP compiles
885 * and 64bit archs it makes no difference if preempt is enabled or not.
886 */
887static inline loff_t i_size_read(const struct inode *inode)
888{
889#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
890 loff_t i_size;
891 unsigned int seq;
892
893 do {
894 seq = read_seqcount_begin(&inode->i_size_seqcount);
895 i_size = inode->i_size;
896 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
897 return i_size;
898#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
899 loff_t i_size;
900
901 preempt_disable();
902 i_size = inode->i_size;
903 preempt_enable();
904 return i_size;
905#else
906 return inode->i_size;
907#endif
908}
909
910/*
911 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
912 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
913 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
914 */
915static inline void i_size_write(struct inode *inode, loff_t i_size)
916{
917#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
918 write_seqcount_begin(&inode->i_size_seqcount);
919 inode->i_size = i_size;
920 write_seqcount_end(&inode->i_size_seqcount);
921#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
922 preempt_disable();
923 inode->i_size = i_size;
924 preempt_enable();
925#else
926 inode->i_size = i_size;
927#endif
928}
929
930static inline unsigned iminor(const struct inode *inode)
931{
932 return MINOR(inode->i_rdev);
933}
934
935static inline unsigned imajor(const struct inode *inode)
936{
937 return MAJOR(inode->i_rdev);
938}
939
940extern struct block_device *I_BDEV(struct inode *inode);
941
942struct fown_struct {
943 rwlock_t lock; /* protects pid, uid, euid fields */
944 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
945 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
946 uid_t uid, euid; /* uid/euid of process setting the owner */
947 int signum; /* posix.1b rt signal to be delivered on IO */
948};
949
950/*
951 * Track a single file's readahead state
952 */
953struct file_ra_state {
954 pgoff_t start; /* where readahead started */
955 unsigned int size; /* # of readahead pages */
956 unsigned int async_size; /* do asynchronous readahead when
957 there are only # of pages ahead */
958
959 unsigned int ra_pages; /* Maximum readahead window */
960 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
961 loff_t prev_pos; /* Cache last read() position */
962};
963
964/*
965 * Check if @index falls in the readahead windows.
966 */
967static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
968{
969 return (index >= ra->start &&
970 index < ra->start + ra->size);
971}
972
973#define FILE_MNT_WRITE_TAKEN 1
974#define FILE_MNT_WRITE_RELEASED 2
975
976struct file {
977 /*
978 * fu_list becomes invalid after file_free is called and queued via
979 * fu_rcuhead for RCU freeing
980 */
981 union {
982 struct list_head fu_list;
983 struct rcu_head fu_rcuhead;
984 } f_u;
985 struct path f_path;
986#define f_dentry f_path.dentry
987#define f_vfsmnt f_path.mnt
988 const struct file_operations *f_op;
989
990 /*
991 * Protects f_ep_links, f_flags, f_pos vs i_size in lseek SEEK_CUR.
992 * Must not be taken from IRQ context.
993 */
994 spinlock_t f_lock;
995#ifdef CONFIG_SMP
996 int f_sb_list_cpu;
997#endif
998 atomic_long_t f_count;
999 unsigned int f_flags;
1000 fmode_t f_mode;
1001 loff_t f_pos;
1002 struct fown_struct f_owner;
1003 const struct cred *f_cred;
1004 struct file_ra_state f_ra;
1005
1006 u64 f_version;
1007#ifdef CONFIG_SECURITY
1008 void *f_security;
1009#endif
1010 /* needed for tty driver, and maybe others */
1011 void *private_data;
1012
1013#ifdef CONFIG_EPOLL
1014 /* Used by fs/eventpoll.c to link all the hooks to this file */
1015 struct list_head f_ep_links;
1016 struct list_head f_tfile_llink;
1017#endif /* #ifdef CONFIG_EPOLL */
1018 struct address_space *f_mapping;
1019#ifdef CONFIG_DEBUG_WRITECOUNT
1020 unsigned long f_mnt_write_state;
1021#endif
1022};
1023
1024struct file_handle {
1025 __u32 handle_bytes;
1026 int handle_type;
1027 /* file identifier */
1028 unsigned char f_handle[0];
1029};
1030
1031#define get_file(x) atomic_long_inc(&(x)->f_count)
1032#define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
1033#define file_count(x) atomic_long_read(&(x)->f_count)
1034
1035#ifdef CONFIG_DEBUG_WRITECOUNT
1036static inline void file_take_write(struct file *f)
1037{
1038 WARN_ON(f->f_mnt_write_state != 0);
1039 f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN;
1040}
1041static inline void file_release_write(struct file *f)
1042{
1043 f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED;
1044}
1045static inline void file_reset_write(struct file *f)
1046{
1047 f->f_mnt_write_state = 0;
1048}
1049static inline void file_check_state(struct file *f)
1050{
1051 /*
1052 * At this point, either both or neither of these bits
1053 * should be set.
1054 */
1055 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN);
1056 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED);
1057}
1058static inline int file_check_writeable(struct file *f)
1059{
1060 if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN)
1061 return 0;
1062 printk(KERN_WARNING "writeable file with no "
1063 "mnt_want_write()\n");
1064 WARN_ON(1);
1065 return -EINVAL;
1066}
1067#else /* !CONFIG_DEBUG_WRITECOUNT */
1068static inline void file_take_write(struct file *filp) {}
1069static inline void file_release_write(struct file *filp) {}
1070static inline void file_reset_write(struct file *filp) {}
1071static inline void file_check_state(struct file *filp) {}
1072static inline int file_check_writeable(struct file *filp)
1073{
1074 return 0;
1075}
1076#endif /* CONFIG_DEBUG_WRITECOUNT */
1077
1078#define MAX_NON_LFS ((1UL<<31) - 1)
1079
1080/* Page cache limit. The filesystems should put that into their s_maxbytes
1081 limits, otherwise bad things can happen in VM. */
1082#if BITS_PER_LONG==32
1083#define MAX_LFS_FILESIZE (((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
1084#elif BITS_PER_LONG==64
1085#define MAX_LFS_FILESIZE 0x7fffffffffffffffUL
1086#endif
1087
1088#define FL_POSIX 1
1089#define FL_FLOCK 2
1090#define FL_ACCESS 8 /* not trying to lock, just looking */
1091#define FL_EXISTS 16 /* when unlocking, test for existence */
1092#define FL_LEASE 32 /* lease held on this file */
1093#define FL_CLOSE 64 /* unlock on close */
1094#define FL_SLEEP 128 /* A blocking lock */
1095#define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
1096#define FL_UNLOCK_PENDING 512 /* Lease is being broken */
1097
1098/*
1099 * Special return value from posix_lock_file() and vfs_lock_file() for
1100 * asynchronous locking.
1101 */
1102#define FILE_LOCK_DEFERRED 1
1103
1104/*
1105 * The POSIX file lock owner is determined by
1106 * the "struct files_struct" in the thread group
1107 * (or NULL for no owner - BSD locks).
1108 *
1109 * Lockd stuffs a "host" pointer into this.
1110 */
1111typedef struct files_struct *fl_owner_t;
1112
1113struct file_lock_operations {
1114 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1115 void (*fl_release_private)(struct file_lock *);
1116};
1117
1118struct lock_manager_operations {
1119 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
1120 void (*lm_notify)(struct file_lock *); /* unblock callback */
1121 int (*lm_grant)(struct file_lock *, struct file_lock *, int);
1122 void (*lm_release_private)(struct file_lock *);
1123 void (*lm_break)(struct file_lock *);
1124 int (*lm_change)(struct file_lock **, int);
1125};
1126
1127struct lock_manager {
1128 struct list_head list;
1129};
1130
1131void locks_start_grace(struct lock_manager *);
1132void locks_end_grace(struct lock_manager *);
1133int locks_in_grace(void);
1134
1135/* that will die - we need it for nfs_lock_info */
1136#include <linux/nfs_fs_i.h>
1137
1138struct file_lock {
1139 struct file_lock *fl_next; /* singly linked list for this inode */
1140 struct list_head fl_link; /* doubly linked list of all locks */
1141 struct list_head fl_block; /* circular list of blocked processes */
1142 fl_owner_t fl_owner;
1143 unsigned int fl_flags;
1144 unsigned char fl_type;
1145 unsigned int fl_pid;
1146 struct pid *fl_nspid;
1147 wait_queue_head_t fl_wait;
1148 struct file *fl_file;
1149 loff_t fl_start;
1150 loff_t fl_end;
1151
1152 struct fasync_struct * fl_fasync; /* for lease break notifications */
1153 /* for lease breaks: */
1154 unsigned long fl_break_time;
1155 unsigned long fl_downgrade_time;
1156
1157 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1158 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1159 union {
1160 struct nfs_lock_info nfs_fl;
1161 struct nfs4_lock_info nfs4_fl;
1162 struct {
1163 struct list_head link; /* link in AFS vnode's pending_locks list */
1164 int state; /* state of grant or error if -ve */
1165 } afs;
1166 } fl_u;
1167};
1168
1169/* The following constant reflects the upper bound of the file/locking space */
1170#ifndef OFFSET_MAX
1171#define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1172#define OFFSET_MAX INT_LIMIT(loff_t)
1173#define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1174#endif
1175
1176#include <linux/fcntl.h>
1177
1178extern void send_sigio(struct fown_struct *fown, int fd, int band);
1179
1180#ifdef CONFIG_FILE_LOCKING
1181extern int fcntl_getlk(struct file *, struct flock __user *);
1182extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1183 struct flock __user *);
1184
1185#if BITS_PER_LONG == 32
1186extern int fcntl_getlk64(struct file *, struct flock64 __user *);
1187extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1188 struct flock64 __user *);
1189#endif
1190
1191extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1192extern int fcntl_getlease(struct file *filp);
1193
1194/* fs/locks.c */
1195void locks_free_lock(struct file_lock *fl);
1196extern void locks_init_lock(struct file_lock *);
1197extern struct file_lock * locks_alloc_lock(void);
1198extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1199extern void __locks_copy_lock(struct file_lock *, const struct file_lock *);
1200extern void locks_remove_posix(struct file *, fl_owner_t);
1201extern void locks_remove_flock(struct file *);
1202extern void locks_release_private(struct file_lock *);
1203extern void posix_test_lock(struct file *, struct file_lock *);
1204extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1205extern int posix_lock_file_wait(struct file *, struct file_lock *);
1206extern int posix_unblock_lock(struct file *, struct file_lock *);
1207extern int vfs_test_lock(struct file *, struct file_lock *);
1208extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1209extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1210extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
1211extern int __break_lease(struct inode *inode, unsigned int flags);
1212extern void lease_get_mtime(struct inode *, struct timespec *time);
1213extern int generic_setlease(struct file *, long, struct file_lock **);
1214extern int vfs_setlease(struct file *, long, struct file_lock **);
1215extern int lease_modify(struct file_lock **, int);
1216extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
1217extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
1218extern void locks_delete_block(struct file_lock *waiter);
1219extern void lock_flocks(void);
1220extern void unlock_flocks(void);
1221#else /* !CONFIG_FILE_LOCKING */
1222static inline int fcntl_getlk(struct file *file, struct flock __user *user)
1223{
1224 return -EINVAL;
1225}
1226
1227static inline int fcntl_setlk(unsigned int fd, struct file *file,
1228 unsigned int cmd, struct flock __user *user)
1229{
1230 return -EACCES;
1231}
1232
1233#if BITS_PER_LONG == 32
1234static inline int fcntl_getlk64(struct file *file, struct flock64 __user *user)
1235{
1236 return -EINVAL;
1237}
1238
1239static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1240 unsigned int cmd, struct flock64 __user *user)
1241{
1242 return -EACCES;
1243}
1244#endif
1245static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1246{
1247 return 0;
1248}
1249
1250static inline int fcntl_getlease(struct file *filp)
1251{
1252 return 0;
1253}
1254
1255static inline void locks_init_lock(struct file_lock *fl)
1256{
1257 return;
1258}
1259
1260static inline void __locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1261{
1262 return;
1263}
1264
1265static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1266{
1267 return;
1268}
1269
1270static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1271{
1272 return;
1273}
1274
1275static inline void locks_remove_flock(struct file *filp)
1276{
1277 return;
1278}
1279
1280static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1281{
1282 return;
1283}
1284
1285static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1286 struct file_lock *conflock)
1287{
1288 return -ENOLCK;
1289}
1290
1291static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1292{
1293 return -ENOLCK;
1294}
1295
1296static inline int posix_unblock_lock(struct file *filp,
1297 struct file_lock *waiter)
1298{
1299 return -ENOENT;
1300}
1301
1302static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1303{
1304 return 0;
1305}
1306
1307static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1308 struct file_lock *fl, struct file_lock *conf)
1309{
1310 return -ENOLCK;
1311}
1312
1313static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1314{
1315 return 0;
1316}
1317
1318static inline int flock_lock_file_wait(struct file *filp,
1319 struct file_lock *request)
1320{
1321 return -ENOLCK;
1322}
1323
1324static inline int __break_lease(struct inode *inode, unsigned int mode)
1325{
1326 return 0;
1327}
1328
1329static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1330{
1331 return;
1332}
1333
1334static inline int generic_setlease(struct file *filp, long arg,
1335 struct file_lock **flp)
1336{
1337 return -EINVAL;
1338}
1339
1340static inline int vfs_setlease(struct file *filp, long arg,
1341 struct file_lock **lease)
1342{
1343 return -EINVAL;
1344}
1345
1346static inline int lease_modify(struct file_lock **before, int arg)
1347{
1348 return -EINVAL;
1349}
1350
1351static inline int lock_may_read(struct inode *inode, loff_t start,
1352 unsigned long len)
1353{
1354 return 1;
1355}
1356
1357static inline int lock_may_write(struct inode *inode, loff_t start,
1358 unsigned long len)
1359{
1360 return 1;
1361}
1362
1363static inline void locks_delete_block(struct file_lock *waiter)
1364{
1365}
1366
1367static inline void lock_flocks(void)
1368{
1369}
1370
1371static inline void unlock_flocks(void)
1372{
1373}
1374
1375#endif /* !CONFIG_FILE_LOCKING */
1376
1377
1378struct fasync_struct {
1379 spinlock_t fa_lock;
1380 int magic;
1381 int fa_fd;
1382 struct fasync_struct *fa_next; /* singly linked list */
1383 struct file *fa_file;
1384 struct rcu_head fa_rcu;
1385};
1386
1387#define FASYNC_MAGIC 0x4601
1388
1389/* SMP safe fasync helpers: */
1390extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1391extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1392extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1393extern struct fasync_struct *fasync_alloc(void);
1394extern void fasync_free(struct fasync_struct *);
1395
1396/* can be called from interrupts */
1397extern void kill_fasync(struct fasync_struct **, int, int);
1398
1399extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1400extern int f_setown(struct file *filp, unsigned long arg, int force);
1401extern void f_delown(struct file *filp);
1402extern pid_t f_getown(struct file *filp);
1403extern int send_sigurg(struct fown_struct *fown);
1404
1405/*
1406 * Umount options
1407 */
1408
1409#define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1410#define MNT_DETACH 0x00000002 /* Just detach from the tree */
1411#define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1412#define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1413#define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1414
1415extern struct list_head super_blocks;
1416extern spinlock_t sb_lock;
1417
1418struct super_block {
1419 struct list_head s_list; /* Keep this first */
1420 dev_t s_dev; /* search index; _not_ kdev_t */
1421 unsigned char s_dirt;
1422 unsigned char s_blocksize_bits;
1423 unsigned long s_blocksize;
1424 loff_t s_maxbytes; /* Max file size */
1425 struct file_system_type *s_type;
1426 const struct super_operations *s_op;
1427 const struct dquot_operations *dq_op;
1428 const struct quotactl_ops *s_qcop;
1429 const struct export_operations *s_export_op;
1430 unsigned long s_flags;
1431 unsigned long s_magic;
1432 struct dentry *s_root;
1433 struct rw_semaphore s_umount;
1434 struct mutex s_lock;
1435 int s_count;
1436 atomic_t s_active;
1437#ifdef CONFIG_SECURITY
1438 void *s_security;
1439#endif
1440 const struct xattr_handler **s_xattr;
1441
1442 struct list_head s_inodes; /* all inodes */
1443 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1444#ifdef CONFIG_SMP
1445 struct list_head __percpu *s_files;
1446#else
1447 struct list_head s_files;
1448#endif
1449 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1450 /* s_dentry_lru, s_nr_dentry_unused protected by dcache.c lru locks */
1451 struct list_head s_dentry_lru; /* unused dentry lru */
1452 int s_nr_dentry_unused; /* # of dentry on lru */
1453
1454 /* s_inode_lru_lock protects s_inode_lru and s_nr_inodes_unused */
1455 spinlock_t s_inode_lru_lock ____cacheline_aligned_in_smp;
1456 struct list_head s_inode_lru; /* unused inode lru */
1457 int s_nr_inodes_unused; /* # of inodes on lru */
1458
1459 struct block_device *s_bdev;
1460 struct backing_dev_info *s_bdi;
1461 struct mtd_info *s_mtd;
1462 struct hlist_node s_instances;
1463 struct quota_info s_dquot; /* Diskquota specific options */
1464
1465 int s_frozen;
1466 wait_queue_head_t s_wait_unfrozen;
1467
1468 char s_id[32]; /* Informational name */
1469 u8 s_uuid[16]; /* UUID */
1470
1471 void *s_fs_info; /* Filesystem private info */
1472 unsigned int s_max_links;
1473 fmode_t s_mode;
1474
1475 /* Granularity of c/m/atime in ns.
1476 Cannot be worse than a second */
1477 u32 s_time_gran;
1478
1479 /*
1480 * The next field is for VFS *only*. No filesystems have any business
1481 * even looking at it. You had been warned.
1482 */
1483 struct mutex s_vfs_rename_mutex; /* Kludge */
1484
1485 /*
1486 * Filesystem subtype. If non-empty the filesystem type field
1487 * in /proc/mounts will be "type.subtype"
1488 */
1489 char *s_subtype;
1490
1491 /*
1492 * Saved mount options for lazy filesystems using
1493 * generic_show_options()
1494 */
1495 char __rcu *s_options;
1496 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1497
1498 /*
1499 * Saved pool identifier for cleancache (-1 means none)
1500 */
1501 int cleancache_poolid;
1502
1503 struct shrinker s_shrink; /* per-sb shrinker handle */
1504
1505 /* Number of inodes with nlink == 0 but still referenced */
1506 atomic_long_t s_remove_count;
1507
1508 /* Being remounted read-only */
1509 int s_readonly_remount;
1510};
1511
1512/* superblock cache pruning functions */
1513extern void prune_icache_sb(struct super_block *sb, int nr_to_scan);
1514extern void prune_dcache_sb(struct super_block *sb, int nr_to_scan);
1515
1516extern struct timespec current_fs_time(struct super_block *sb);
1517
1518/*
1519 * Snapshotting support.
1520 */
1521enum {
1522 SB_UNFROZEN = 0,
1523 SB_FREEZE_WRITE = 1,
1524 SB_FREEZE_TRANS = 2,
1525};
1526
1527#define vfs_check_frozen(sb, level) \
1528 wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
1529
1530/*
1531 * until VFS tracks user namespaces for inodes, just make all files
1532 * belong to init_user_ns
1533 */
1534extern struct user_namespace init_user_ns;
1535#define inode_userns(inode) (&init_user_ns)
1536extern bool inode_owner_or_capable(const struct inode *inode);
1537
1538/* not quite ready to be deprecated, but... */
1539extern void lock_super(struct super_block *);
1540extern void unlock_super(struct super_block *);
1541
1542/*
1543 * VFS helper functions..
1544 */
1545extern int vfs_create(struct inode *, struct dentry *, umode_t, struct nameidata *);
1546extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1547extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1548extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1549extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
1550extern int vfs_rmdir(struct inode *, struct dentry *);
1551extern int vfs_unlink(struct inode *, struct dentry *);
1552extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1553
1554/*
1555 * VFS dentry helper functions.
1556 */
1557extern void dentry_unhash(struct dentry *dentry);
1558
1559/*
1560 * VFS file helper functions.
1561 */
1562extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1563 umode_t mode);
1564/*
1565 * VFS FS_IOC_FIEMAP helper definitions.
1566 */
1567struct fiemap_extent_info {
1568 unsigned int fi_flags; /* Flags as passed from user */
1569 unsigned int fi_extents_mapped; /* Number of mapped extents */
1570 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1571 struct fiemap_extent __user *fi_extents_start; /* Start of
1572 fiemap_extent array */
1573};
1574int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1575 u64 phys, u64 len, u32 flags);
1576int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1577
1578/*
1579 * File types
1580 *
1581 * NOTE! These match bits 12..15 of stat.st_mode
1582 * (ie "(i_mode >> 12) & 15").
1583 */
1584#define DT_UNKNOWN 0
1585#define DT_FIFO 1
1586#define DT_CHR 2
1587#define DT_DIR 4
1588#define DT_BLK 6
1589#define DT_REG 8
1590#define DT_LNK 10
1591#define DT_SOCK 12
1592#define DT_WHT 14
1593
1594/*
1595 * This is the "filldir" function type, used by readdir() to let
1596 * the kernel specify what kind of dirent layout it wants to have.
1597 * This allows the kernel to read directories into kernel space or
1598 * to have different dirent layouts depending on the binary type.
1599 */
1600typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
1601struct block_device_operations;
1602
1603/* These macros are for out of kernel modules to test that
1604 * the kernel supports the unlocked_ioctl and compat_ioctl
1605 * fields in struct file_operations. */
1606#define HAVE_COMPAT_IOCTL 1
1607#define HAVE_UNLOCKED_IOCTL 1
1608
1609struct file_operations {
1610 struct module *owner;
1611 loff_t (*llseek) (struct file *, loff_t, int);
1612 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1613 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1614 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1615 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1616 int (*readdir) (struct file *, void *, filldir_t);
1617 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1618 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1619 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1620 int (*mmap) (struct file *, struct vm_area_struct *);
1621 int (*open) (struct inode *, struct file *);
1622 int (*flush) (struct file *, fl_owner_t id);
1623 int (*release) (struct inode *, struct file *);
1624 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1625 int (*aio_fsync) (struct kiocb *, int datasync);
1626 int (*fasync) (int, struct file *, int);
1627 int (*lock) (struct file *, int, struct file_lock *);
1628 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1629 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1630 int (*check_flags)(int);
1631 int (*flock) (struct file *, int, struct file_lock *);
1632 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1633 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1634 int (*setlease)(struct file *, long, struct file_lock **);
1635 long (*fallocate)(struct file *file, int mode, loff_t offset,
1636 loff_t len);
1637};
1638
1639struct inode_operations {
1640 struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
1641 void * (*follow_link) (struct dentry *, struct nameidata *);
1642 int (*permission) (struct inode *, int);
1643 struct posix_acl * (*get_acl)(struct inode *, int);
1644
1645 int (*readlink) (struct dentry *, char __user *,int);
1646 void (*put_link) (struct dentry *, struct nameidata *, void *);
1647
1648 int (*create) (struct inode *,struct dentry *,umode_t,struct nameidata *);
1649 int (*link) (struct dentry *,struct inode *,struct dentry *);
1650 int (*unlink) (struct inode *,struct dentry *);
1651 int (*symlink) (struct inode *,struct dentry *,const char *);
1652 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1653 int (*rmdir) (struct inode *,struct dentry *);
1654 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1655 int (*rename) (struct inode *, struct dentry *,
1656 struct inode *, struct dentry *);
1657 void (*truncate) (struct inode *);
1658 int (*setattr) (struct dentry *, struct iattr *);
1659 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1660 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1661 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1662 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1663 int (*removexattr) (struct dentry *, const char *);
1664 void (*truncate_range)(struct inode *, loff_t, loff_t);
1665 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1666 u64 len);
1667} ____cacheline_aligned;
1668
1669struct seq_file;
1670
1671ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1672 unsigned long nr_segs, unsigned long fast_segs,
1673 struct iovec *fast_pointer,
1674 struct iovec **ret_pointer,
1675 int check_access);
1676
1677extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1678extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1679extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1680 unsigned long, loff_t *);
1681extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1682 unsigned long, loff_t *);
1683
1684struct super_operations {
1685 struct inode *(*alloc_inode)(struct super_block *sb);
1686 void (*destroy_inode)(struct inode *);
1687
1688 void (*dirty_inode) (struct inode *, int flags);
1689 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1690 int (*drop_inode) (struct inode *);
1691 void (*evict_inode) (struct inode *);
1692 void (*put_super) (struct super_block *);
1693 void (*write_super) (struct super_block *);
1694 int (*sync_fs)(struct super_block *sb, int wait);
1695 int (*freeze_fs) (struct super_block *);
1696 int (*unfreeze_fs) (struct super_block *);
1697 int (*statfs) (struct dentry *, struct kstatfs *);
1698 int (*remount_fs) (struct super_block *, int *, char *);
1699 void (*umount_begin) (struct super_block *);
1700
1701 int (*show_options)(struct seq_file *, struct dentry *);
1702 int (*show_devname)(struct seq_file *, struct dentry *);
1703 int (*show_path)(struct seq_file *, struct dentry *);
1704 int (*show_stats)(struct seq_file *, struct dentry *);
1705#ifdef CONFIG_QUOTA
1706 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1707 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1708#endif
1709 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1710 int (*nr_cached_objects)(struct super_block *);
1711 void (*free_cached_objects)(struct super_block *, int);
1712};
1713
1714/*
1715 * Inode state bits. Protected by inode->i_lock
1716 *
1717 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1718 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1719 *
1720 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1721 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1722 * various stages of removing an inode.
1723 *
1724 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1725 *
1726 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1727 * fdatasync(). i_atime is the usual cause.
1728 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1729 * these changes separately from I_DIRTY_SYNC so that we
1730 * don't have to write inode on fdatasync() when only
1731 * mtime has changed in it.
1732 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1733 * I_NEW Serves as both a mutex and completion notification.
1734 * New inodes set I_NEW. If two processes both create
1735 * the same inode, one of them will release its inode and
1736 * wait for I_NEW to be released before returning.
1737 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1738 * also cause waiting on I_NEW, without I_NEW actually
1739 * being set. find_inode() uses this to prevent returning
1740 * nearly-dead inodes.
1741 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1742 * is zero. I_FREEING must be set when I_WILL_FREE is
1743 * cleared.
1744 * I_FREEING Set when inode is about to be freed but still has dirty
1745 * pages or buffers attached or the inode itself is still
1746 * dirty.
1747 * I_CLEAR Added by end_writeback(). In this state the inode is clean
1748 * and can be destroyed. Inode keeps I_FREEING.
1749 *
1750 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1751 * prohibited for many purposes. iget() must wait for
1752 * the inode to be completely released, then create it
1753 * anew. Other functions will just ignore such inodes,
1754 * if appropriate. I_NEW is used for waiting.
1755 *
1756 * I_SYNC Synchonized write of dirty inode data. The bits is
1757 * set during data writeback, and cleared with a wakeup
1758 * on the bit address once it is done.
1759 *
1760 * I_REFERENCED Marks the inode as recently references on the LRU list.
1761 *
1762 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1763 *
1764 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1765 */
1766#define I_DIRTY_SYNC (1 << 0)
1767#define I_DIRTY_DATASYNC (1 << 1)
1768#define I_DIRTY_PAGES (1 << 2)
1769#define __I_NEW 3
1770#define I_NEW (1 << __I_NEW)
1771#define I_WILL_FREE (1 << 4)
1772#define I_FREEING (1 << 5)
1773#define I_CLEAR (1 << 6)
1774#define __I_SYNC 7
1775#define I_SYNC (1 << __I_SYNC)
1776#define I_REFERENCED (1 << 8)
1777#define __I_DIO_WAKEUP 9
1778#define I_DIO_WAKEUP (1 << I_DIO_WAKEUP)
1779
1780#define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1781
1782extern void __mark_inode_dirty(struct inode *, int);
1783static inline void mark_inode_dirty(struct inode *inode)
1784{
1785 __mark_inode_dirty(inode, I_DIRTY);
1786}
1787
1788static inline void mark_inode_dirty_sync(struct inode *inode)
1789{
1790 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1791}
1792
1793extern void inc_nlink(struct inode *inode);
1794extern void drop_nlink(struct inode *inode);
1795extern void clear_nlink(struct inode *inode);
1796extern void set_nlink(struct inode *inode, unsigned int nlink);
1797
1798static inline void inode_inc_link_count(struct inode *inode)
1799{
1800 inc_nlink(inode);
1801 mark_inode_dirty(inode);
1802}
1803
1804static inline void inode_dec_link_count(struct inode *inode)
1805{
1806 drop_nlink(inode);
1807 mark_inode_dirty(inode);
1808}
1809
1810/**
1811 * inode_inc_iversion - increments i_version
1812 * @inode: inode that need to be updated
1813 *
1814 * Every time the inode is modified, the i_version field will be incremented.
1815 * The filesystem has to be mounted with i_version flag
1816 */
1817
1818static inline void inode_inc_iversion(struct inode *inode)
1819{
1820 spin_lock(&inode->i_lock);
1821 inode->i_version++;
1822 spin_unlock(&inode->i_lock);
1823}
1824
1825extern void touch_atime(struct path *);
1826static inline void file_accessed(struct file *file)
1827{
1828 if (!(file->f_flags & O_NOATIME))
1829 touch_atime(&file->f_path);
1830}
1831
1832int sync_inode(struct inode *inode, struct writeback_control *wbc);
1833int sync_inode_metadata(struct inode *inode, int wait);
1834
1835struct file_system_type {
1836 const char *name;
1837 int fs_flags;
1838 struct dentry *(*mount) (struct file_system_type *, int,
1839 const char *, void *);
1840 void (*kill_sb) (struct super_block *);
1841 struct module *owner;
1842 struct file_system_type * next;
1843 struct hlist_head fs_supers;
1844
1845 struct lock_class_key s_lock_key;
1846 struct lock_class_key s_umount_key;
1847 struct lock_class_key s_vfs_rename_key;
1848
1849 struct lock_class_key i_lock_key;
1850 struct lock_class_key i_mutex_key;
1851 struct lock_class_key i_mutex_dir_key;
1852};
1853
1854extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1855 void *data, int (*fill_super)(struct super_block *, void *, int));
1856extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1857 int flags, const char *dev_name, void *data,
1858 int (*fill_super)(struct super_block *, void *, int));
1859extern struct dentry *mount_single(struct file_system_type *fs_type,
1860 int flags, void *data,
1861 int (*fill_super)(struct super_block *, void *, int));
1862extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1863 int flags, void *data,
1864 int (*fill_super)(struct super_block *, void *, int));
1865extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
1866void generic_shutdown_super(struct super_block *sb);
1867void kill_block_super(struct super_block *sb);
1868void kill_anon_super(struct super_block *sb);
1869void kill_litter_super(struct super_block *sb);
1870void deactivate_super(struct super_block *sb);
1871void deactivate_locked_super(struct super_block *sb);
1872int set_anon_super(struct super_block *s, void *data);
1873int get_anon_bdev(dev_t *);
1874void free_anon_bdev(dev_t);
1875struct super_block *sget(struct file_system_type *type,
1876 int (*test)(struct super_block *,void *),
1877 int (*set)(struct super_block *,void *),
1878 void *data);
1879extern struct dentry *mount_pseudo(struct file_system_type *, char *,
1880 const struct super_operations *ops,
1881 const struct dentry_operations *dops,
1882 unsigned long);
1883
1884/* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1885#define fops_get(fops) \
1886 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1887#define fops_put(fops) \
1888 do { if (fops) module_put((fops)->owner); } while(0)
1889
1890extern int register_filesystem(struct file_system_type *);
1891extern int unregister_filesystem(struct file_system_type *);
1892extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1893#define kern_mount(type) kern_mount_data(type, NULL)
1894extern void kern_unmount(struct vfsmount *mnt);
1895extern int may_umount_tree(struct vfsmount *);
1896extern int may_umount(struct vfsmount *);
1897extern long do_mount(char *, char *, char *, unsigned long, void *);
1898extern struct vfsmount *collect_mounts(struct path *);
1899extern void drop_collected_mounts(struct vfsmount *);
1900extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1901 struct vfsmount *);
1902extern int vfs_statfs(struct path *, struct kstatfs *);
1903extern int user_statfs(const char __user *, struct kstatfs *);
1904extern int fd_statfs(int, struct kstatfs *);
1905extern int vfs_ustat(dev_t, struct kstatfs *);
1906extern int freeze_super(struct super_block *super);
1907extern int thaw_super(struct super_block *super);
1908extern bool our_mnt(struct vfsmount *mnt);
1909
1910extern int current_umask(void);
1911
1912/* /sys/fs */
1913extern struct kobject *fs_kobj;
1914
1915#define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
1916extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1917
1918#define FLOCK_VERIFY_READ 1
1919#define FLOCK_VERIFY_WRITE 2
1920
1921#ifdef CONFIG_FILE_LOCKING
1922extern int locks_mandatory_locked(struct inode *);
1923extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1924
1925/*
1926 * Candidates for mandatory locking have the setgid bit set
1927 * but no group execute bit - an otherwise meaningless combination.
1928 */
1929
1930static inline int __mandatory_lock(struct inode *ino)
1931{
1932 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1933}
1934
1935/*
1936 * ... and these candidates should be on MS_MANDLOCK mounted fs,
1937 * otherwise these will be advisory locks
1938 */
1939
1940static inline int mandatory_lock(struct inode *ino)
1941{
1942 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1943}
1944
1945static inline int locks_verify_locked(struct inode *inode)
1946{
1947 if (mandatory_lock(inode))
1948 return locks_mandatory_locked(inode);
1949 return 0;
1950}
1951
1952static inline int locks_verify_truncate(struct inode *inode,
1953 struct file *filp,
1954 loff_t size)
1955{
1956 if (inode->i_flock && mandatory_lock(inode))
1957 return locks_mandatory_area(
1958 FLOCK_VERIFY_WRITE, inode, filp,
1959 size < inode->i_size ? size : inode->i_size,
1960 (size < inode->i_size ? inode->i_size - size
1961 : size - inode->i_size)
1962 );
1963 return 0;
1964}
1965
1966static inline int break_lease(struct inode *inode, unsigned int mode)
1967{
1968 if (inode->i_flock)
1969 return __break_lease(inode, mode);
1970 return 0;
1971}
1972#else /* !CONFIG_FILE_LOCKING */
1973static inline int locks_mandatory_locked(struct inode *inode)
1974{
1975 return 0;
1976}
1977
1978static inline int locks_mandatory_area(int rw, struct inode *inode,
1979 struct file *filp, loff_t offset,
1980 size_t count)
1981{
1982 return 0;
1983}
1984
1985static inline int __mandatory_lock(struct inode *inode)
1986{
1987 return 0;
1988}
1989
1990static inline int mandatory_lock(struct inode *inode)
1991{
1992 return 0;
1993}
1994
1995static inline int locks_verify_locked(struct inode *inode)
1996{
1997 return 0;
1998}
1999
2000static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2001 size_t size)
2002{
2003 return 0;
2004}
2005
2006static inline int break_lease(struct inode *inode, unsigned int mode)
2007{
2008 return 0;
2009}
2010
2011#endif /* CONFIG_FILE_LOCKING */
2012
2013/* fs/open.c */
2014
2015extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2016 struct file *filp);
2017extern int do_fallocate(struct file *file, int mode, loff_t offset,
2018 loff_t len);
2019extern long do_sys_open(int dfd, const char __user *filename, int flags,
2020 umode_t mode);
2021extern struct file *filp_open(const char *, int, umode_t);
2022extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2023 const char *, int);
2024extern struct file * dentry_open(struct dentry *, struct vfsmount *, int,
2025 const struct cred *);
2026extern int filp_close(struct file *, fl_owner_t id);
2027extern char * getname(const char __user *);
2028
2029/* fs/ioctl.c */
2030
2031extern int ioctl_preallocate(struct file *filp, void __user *argp);
2032
2033/* fs/dcache.c */
2034extern void __init vfs_caches_init_early(void);
2035extern void __init vfs_caches_init(unsigned long);
2036
2037extern struct kmem_cache *names_cachep;
2038
2039#define __getname_gfp(gfp) kmem_cache_alloc(names_cachep, (gfp))
2040#define __getname() __getname_gfp(GFP_KERNEL)
2041#define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2042#ifndef CONFIG_AUDITSYSCALL
2043#define putname(name) __putname(name)
2044#else
2045extern void putname(const char *name);
2046#endif
2047
2048#ifdef CONFIG_BLOCK
2049extern int register_blkdev(unsigned int, const char *);
2050extern void unregister_blkdev(unsigned int, const char *);
2051extern struct block_device *bdget(dev_t);
2052extern struct block_device *bdgrab(struct block_device *bdev);
2053extern void bd_set_size(struct block_device *, loff_t size);
2054extern sector_t blkdev_max_block(struct block_device *bdev);
2055extern void bd_forget(struct inode *inode);
2056extern void bdput(struct block_device *);
2057extern void invalidate_bdev(struct block_device *);
2058extern int sync_blockdev(struct block_device *bdev);
2059extern void kill_bdev(struct block_device *);
2060extern struct super_block *freeze_bdev(struct block_device *);
2061extern void emergency_thaw_all(void);
2062extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2063extern int fsync_bdev(struct block_device *);
2064#else
2065static inline void bd_forget(struct inode *inode) {}
2066static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2067static inline void kill_bdev(struct block_device *bdev) {}
2068static inline void invalidate_bdev(struct block_device *bdev) {}
2069
2070static inline struct super_block *freeze_bdev(struct block_device *sb)
2071{
2072 return NULL;
2073}
2074
2075static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2076{
2077 return 0;
2078}
2079#endif
2080extern int sync_filesystem(struct super_block *);
2081extern const struct file_operations def_blk_fops;
2082extern const struct file_operations def_chr_fops;
2083extern const struct file_operations bad_sock_fops;
2084extern const struct file_operations def_fifo_fops;
2085#ifdef CONFIG_BLOCK
2086extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2087extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2088extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2089extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2090extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2091 void *holder);
2092extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2093 void *holder);
2094extern int blkdev_put(struct block_device *bdev, fmode_t mode);
2095#ifdef CONFIG_SYSFS
2096extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2097extern void bd_unlink_disk_holder(struct block_device *bdev,
2098 struct gendisk *disk);
2099#else
2100static inline int bd_link_disk_holder(struct block_device *bdev,
2101 struct gendisk *disk)
2102{
2103 return 0;
2104}
2105static inline void bd_unlink_disk_holder(struct block_device *bdev,
2106 struct gendisk *disk)
2107{
2108}
2109#endif
2110#endif
2111
2112/* fs/char_dev.c */
2113#define CHRDEV_MAJOR_HASH_SIZE 255
2114extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2115extern int register_chrdev_region(dev_t, unsigned, const char *);
2116extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2117 unsigned int count, const char *name,
2118 const struct file_operations *fops);
2119extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2120 unsigned int count, const char *name);
2121extern void unregister_chrdev_region(dev_t, unsigned);
2122extern void chrdev_show(struct seq_file *,off_t);
2123
2124static inline int register_chrdev(unsigned int major, const char *name,
2125 const struct file_operations *fops)
2126{
2127 return __register_chrdev(major, 0, 256, name, fops);
2128}
2129
2130static inline void unregister_chrdev(unsigned int major, const char *name)
2131{
2132 __unregister_chrdev(major, 0, 256, name);
2133}
2134
2135/* fs/block_dev.c */
2136#define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2137#define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2138
2139#ifdef CONFIG_BLOCK
2140#define BLKDEV_MAJOR_HASH_SIZE 255
2141extern const char *__bdevname(dev_t, char *buffer);
2142extern const char *bdevname(struct block_device *bdev, char *buffer);
2143extern struct block_device *lookup_bdev(const char *);
2144extern void blkdev_show(struct seq_file *,off_t);
2145
2146#else
2147#define BLKDEV_MAJOR_HASH_SIZE 0
2148#endif
2149
2150extern void init_special_inode(struct inode *, umode_t, dev_t);
2151
2152/* Invalid inode operations -- fs/bad_inode.c */
2153extern void make_bad_inode(struct inode *);
2154extern int is_bad_inode(struct inode *);
2155
2156extern const struct file_operations read_pipefifo_fops;
2157extern const struct file_operations write_pipefifo_fops;
2158extern const struct file_operations rdwr_pipefifo_fops;
2159
2160#ifdef CONFIG_BLOCK
2161/*
2162 * return READ, READA, or WRITE
2163 */
2164#define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2165
2166/*
2167 * return data direction, READ or WRITE
2168 */
2169#define bio_data_dir(bio) ((bio)->bi_rw & 1)
2170
2171extern void check_disk_size_change(struct gendisk *disk,
2172 struct block_device *bdev);
2173extern int revalidate_disk(struct gendisk *);
2174extern int check_disk_change(struct block_device *);
2175extern int __invalidate_device(struct block_device *, bool);
2176extern int invalidate_partition(struct gendisk *, int);
2177#endif
2178unsigned long invalidate_mapping_pages(struct address_space *mapping,
2179 pgoff_t start, pgoff_t end);
2180
2181static inline void invalidate_remote_inode(struct inode *inode)
2182{
2183 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2184 S_ISLNK(inode->i_mode))
2185 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2186}
2187extern int invalidate_inode_pages2(struct address_space *mapping);
2188extern int invalidate_inode_pages2_range(struct address_space *mapping,
2189 pgoff_t start, pgoff_t end);
2190extern int write_inode_now(struct inode *, int);
2191extern int filemap_fdatawrite(struct address_space *);
2192extern int filemap_flush(struct address_space *);
2193extern int filemap_fdatawait(struct address_space *);
2194extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2195 loff_t lend);
2196extern int filemap_write_and_wait(struct address_space *mapping);
2197extern int filemap_write_and_wait_range(struct address_space *mapping,
2198 loff_t lstart, loff_t lend);
2199extern int __filemap_fdatawrite_range(struct address_space *mapping,
2200 loff_t start, loff_t end, int sync_mode);
2201extern int filemap_fdatawrite_range(struct address_space *mapping,
2202 loff_t start, loff_t end);
2203
2204extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2205 int datasync);
2206extern int vfs_fsync(struct file *file, int datasync);
2207extern int generic_write_sync(struct file *file, loff_t pos, loff_t count);
2208extern void sync_supers(void);
2209extern void emergency_sync(void);
2210extern void emergency_remount(void);
2211#ifdef CONFIG_BLOCK
2212extern sector_t bmap(struct inode *, sector_t);
2213#endif
2214extern int notify_change(struct dentry *, struct iattr *);
2215extern int inode_permission(struct inode *, int);
2216extern int generic_permission(struct inode *, int);
2217
2218static inline bool execute_ok(struct inode *inode)
2219{
2220 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2221}
2222
2223/*
2224 * get_write_access() gets write permission for a file.
2225 * put_write_access() releases this write permission.
2226 * This is used for regular files.
2227 * We cannot support write (and maybe mmap read-write shared) accesses and
2228 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2229 * can have the following values:
2230 * 0: no writers, no VM_DENYWRITE mappings
2231 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2232 * > 0: (i_writecount) users are writing to the file.
2233 *
2234 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2235 * except for the cases where we don't hold i_writecount yet. Then we need to
2236 * use {get,deny}_write_access() - these functions check the sign and refuse
2237 * to do the change if sign is wrong.
2238 */
2239static inline int get_write_access(struct inode *inode)
2240{
2241 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2242}
2243static inline int deny_write_access(struct file *file)
2244{
2245 struct inode *inode = file->f_path.dentry->d_inode;
2246 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2247}
2248static inline void put_write_access(struct inode * inode)
2249{
2250 atomic_dec(&inode->i_writecount);
2251}
2252static inline void allow_write_access(struct file *file)
2253{
2254 if (file)
2255 atomic_inc(&file->f_path.dentry->d_inode->i_writecount);
2256}
2257#ifdef CONFIG_IMA
2258static inline void i_readcount_dec(struct inode *inode)
2259{
2260 BUG_ON(!atomic_read(&inode->i_readcount));
2261 atomic_dec(&inode->i_readcount);
2262}
2263static inline void i_readcount_inc(struct inode *inode)
2264{
2265 atomic_inc(&inode->i_readcount);
2266}
2267#else
2268static inline void i_readcount_dec(struct inode *inode)
2269{
2270 return;
2271}
2272static inline void i_readcount_inc(struct inode *inode)
2273{
2274 return;
2275}
2276#endif
2277extern int do_pipe_flags(int *, int);
2278extern struct file *create_read_pipe(struct file *f, int flags);
2279extern struct file *create_write_pipe(int flags);
2280extern void free_write_pipe(struct file *);
2281
2282extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2283extern struct file * open_exec(const char *);
2284
2285/* fs/dcache.c -- generic fs support functions */
2286extern int is_subdir(struct dentry *, struct dentry *);
2287extern int path_is_under(struct path *, struct path *);
2288extern ino_t find_inode_number(struct dentry *, struct qstr *);
2289
2290#include <linux/err.h>
2291
2292/* needed for stackable file system support */
2293extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
2294
2295extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
2296
2297extern int inode_init_always(struct super_block *, struct inode *);
2298extern void inode_init_once(struct inode *);
2299extern void address_space_init_once(struct address_space *mapping);
2300extern void ihold(struct inode * inode);
2301extern void iput(struct inode *);
2302extern struct inode * igrab(struct inode *);
2303extern ino_t iunique(struct super_block *, ino_t);
2304extern int inode_needs_sync(struct inode *inode);
2305extern int generic_delete_inode(struct inode *inode);
2306static inline int generic_drop_inode(struct inode *inode)
2307{
2308 return !inode->i_nlink || inode_unhashed(inode);
2309}
2310
2311extern struct inode *ilookup5_nowait(struct super_block *sb,
2312 unsigned long hashval, int (*test)(struct inode *, void *),
2313 void *data);
2314extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2315 int (*test)(struct inode *, void *), void *data);
2316extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2317
2318extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2319extern struct inode * iget_locked(struct super_block *, unsigned long);
2320extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2321extern int insert_inode_locked(struct inode *);
2322#ifdef CONFIG_DEBUG_LOCK_ALLOC
2323extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2324#else
2325static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2326#endif
2327extern void unlock_new_inode(struct inode *);
2328extern unsigned int get_next_ino(void);
2329
2330extern void __iget(struct inode * inode);
2331extern void iget_failed(struct inode *);
2332extern void end_writeback(struct inode *);
2333extern void __destroy_inode(struct inode *);
2334extern struct inode *new_inode_pseudo(struct super_block *sb);
2335extern struct inode *new_inode(struct super_block *sb);
2336extern void free_inode_nonrcu(struct inode *inode);
2337extern int should_remove_suid(struct dentry *);
2338extern int file_remove_suid(struct file *);
2339
2340extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2341static inline void insert_inode_hash(struct inode *inode)
2342{
2343 __insert_inode_hash(inode, inode->i_ino);
2344}
2345
2346extern void __remove_inode_hash(struct inode *);
2347static inline void remove_inode_hash(struct inode *inode)
2348{
2349 if (!inode_unhashed(inode))
2350 __remove_inode_hash(inode);
2351}
2352
2353extern void inode_sb_list_add(struct inode *inode);
2354
2355#ifdef CONFIG_BLOCK
2356extern void submit_bio(int, struct bio *);
2357extern int bdev_read_only(struct block_device *);
2358#endif
2359extern int set_blocksize(struct block_device *, int);
2360extern int sb_set_blocksize(struct super_block *, int);
2361extern int sb_min_blocksize(struct super_block *, int);
2362
2363extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2364extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2365extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
2366int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
2367extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2368extern ssize_t __generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long,
2369 loff_t *);
2370extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2371extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
2372 unsigned long *, loff_t, loff_t *, size_t, size_t);
2373extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
2374 unsigned long, loff_t, loff_t *, size_t, ssize_t);
2375extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2376extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2377extern int generic_segment_checks(const struct iovec *iov,
2378 unsigned long *nr_segs, size_t *count, int access_flags);
2379
2380/* fs/block_dev.c */
2381extern ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov,
2382 unsigned long nr_segs, loff_t pos);
2383extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2384 int datasync);
2385extern void block_sync_page(struct page *page);
2386
2387/* fs/splice.c */
2388extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2389 struct pipe_inode_info *, size_t, unsigned int);
2390extern ssize_t default_file_splice_read(struct file *, loff_t *,
2391 struct pipe_inode_info *, size_t, unsigned int);
2392extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
2393 struct file *, loff_t *, size_t, unsigned int);
2394extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2395 struct file *out, loff_t *, size_t len, unsigned int flags);
2396extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2397 size_t len, unsigned int flags);
2398
2399extern void
2400file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2401extern loff_t noop_llseek(struct file *file, loff_t offset, int origin);
2402extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
2403extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
2404extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2405 int origin, loff_t maxsize);
2406extern int generic_file_open(struct inode * inode, struct file * filp);
2407extern int nonseekable_open(struct inode * inode, struct file * filp);
2408
2409#ifdef CONFIG_FS_XIP
2410extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2411 loff_t *ppos);
2412extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
2413extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2414 size_t len, loff_t *ppos);
2415extern int xip_truncate_page(struct address_space *mapping, loff_t from);
2416#else
2417static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2418{
2419 return 0;
2420}
2421#endif
2422
2423#ifdef CONFIG_BLOCK
2424typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2425 loff_t file_offset);
2426
2427enum {
2428 /* need locking between buffered and direct access */
2429 DIO_LOCKING = 0x01,
2430
2431 /* filesystem does not support filling holes */
2432 DIO_SKIP_HOLES = 0x02,
2433};
2434
2435void dio_end_io(struct bio *bio, int error);
2436void inode_dio_wait(struct inode *inode);
2437void inode_dio_done(struct inode *inode);
2438
2439ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2440 struct block_device *bdev, const struct iovec *iov, loff_t offset,
2441 unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
2442 dio_submit_t submit_io, int flags);
2443
2444static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
2445 struct inode *inode, const struct iovec *iov, loff_t offset,
2446 unsigned long nr_segs, get_block_t get_block)
2447{
2448 return __blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov,
2449 offset, nr_segs, get_block, NULL, NULL,
2450 DIO_LOCKING | DIO_SKIP_HOLES);
2451}
2452#else
2453static inline void inode_dio_wait(struct inode *inode)
2454{
2455}
2456#endif
2457
2458extern const struct file_operations generic_ro_fops;
2459
2460#define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2461
2462extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
2463extern int vfs_follow_link(struct nameidata *, const char *);
2464extern int page_readlink(struct dentry *, char __user *, int);
2465extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2466extern void page_put_link(struct dentry *, struct nameidata *, void *);
2467extern int __page_symlink(struct inode *inode, const char *symname, int len,
2468 int nofs);
2469extern int page_symlink(struct inode *inode, const char *symname, int len);
2470extern const struct inode_operations page_symlink_inode_operations;
2471extern int generic_readlink(struct dentry *, char __user *, int);
2472extern void generic_fillattr(struct inode *, struct kstat *);
2473extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2474void __inode_add_bytes(struct inode *inode, loff_t bytes);
2475void inode_add_bytes(struct inode *inode, loff_t bytes);
2476void inode_sub_bytes(struct inode *inode, loff_t bytes);
2477loff_t inode_get_bytes(struct inode *inode);
2478void inode_set_bytes(struct inode *inode, loff_t bytes);
2479
2480extern int vfs_readdir(struct file *, filldir_t, void *);
2481
2482extern int vfs_stat(const char __user *, struct kstat *);
2483extern int vfs_lstat(const char __user *, struct kstat *);
2484extern int vfs_fstat(unsigned int, struct kstat *);
2485extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2486
2487extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2488 unsigned long arg);
2489extern int __generic_block_fiemap(struct inode *inode,
2490 struct fiemap_extent_info *fieinfo,
2491 loff_t start, loff_t len,
2492 get_block_t *get_block);
2493extern int generic_block_fiemap(struct inode *inode,
2494 struct fiemap_extent_info *fieinfo, u64 start,
2495 u64 len, get_block_t *get_block);
2496
2497extern void get_filesystem(struct file_system_type *fs);
2498extern void put_filesystem(struct file_system_type *fs);
2499extern struct file_system_type *get_fs_type(const char *name);
2500extern struct super_block *get_super(struct block_device *);
2501extern struct super_block *get_super_thawed(struct block_device *);
2502extern struct super_block *get_active_super(struct block_device *bdev);
2503extern void drop_super(struct super_block *sb);
2504extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2505extern void iterate_supers_type(struct file_system_type *,
2506 void (*)(struct super_block *, void *), void *);
2507
2508extern int dcache_dir_open(struct inode *, struct file *);
2509extern int dcache_dir_close(struct inode *, struct file *);
2510extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2511extern int dcache_readdir(struct file *, void *, filldir_t);
2512extern int simple_setattr(struct dentry *, struct iattr *);
2513extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2514extern int simple_statfs(struct dentry *, struct kstatfs *);
2515extern int simple_open(struct inode *inode, struct file *file);
2516extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2517extern int simple_unlink(struct inode *, struct dentry *);
2518extern int simple_rmdir(struct inode *, struct dentry *);
2519extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2520extern int noop_fsync(struct file *, loff_t, loff_t, int);
2521extern int simple_empty(struct dentry *);
2522extern int simple_readpage(struct file *file, struct page *page);
2523extern int simple_write_begin(struct file *file, struct address_space *mapping,
2524 loff_t pos, unsigned len, unsigned flags,
2525 struct page **pagep, void **fsdata);
2526extern int simple_write_end(struct file *file, struct address_space *mapping,
2527 loff_t pos, unsigned len, unsigned copied,
2528 struct page *page, void *fsdata);
2529
2530extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
2531extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2532extern const struct file_operations simple_dir_operations;
2533extern const struct inode_operations simple_dir_inode_operations;
2534struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2535struct dentry *d_alloc_name(struct dentry *, const char *);
2536extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2537extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2538extern void simple_release_fs(struct vfsmount **mount, int *count);
2539
2540extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2541 loff_t *ppos, const void *from, size_t available);
2542extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2543 const void __user *from, size_t count);
2544
2545extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2546
2547extern int generic_check_addressable(unsigned, u64);
2548
2549#ifdef CONFIG_MIGRATION
2550extern int buffer_migrate_page(struct address_space *,
2551 struct page *, struct page *,
2552 enum migrate_mode);
2553#else
2554#define buffer_migrate_page NULL
2555#endif
2556
2557extern int inode_change_ok(const struct inode *, struct iattr *);
2558extern int inode_newsize_ok(const struct inode *, loff_t offset);
2559extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2560
2561extern void file_update_time(struct file *file);
2562
2563extern int generic_show_options(struct seq_file *m, struct dentry *root);
2564extern void save_mount_options(struct super_block *sb, char *options);
2565extern void replace_mount_options(struct super_block *sb, char *options);
2566
2567static inline ino_t parent_ino(struct dentry *dentry)
2568{
2569 ino_t res;
2570
2571 /*
2572 * Don't strictly need d_lock here? If the parent ino could change
2573 * then surely we'd have a deeper race in the caller?
2574 */
2575 spin_lock(&dentry->d_lock);
2576 res = dentry->d_parent->d_inode->i_ino;
2577 spin_unlock(&dentry->d_lock);
2578 return res;
2579}
2580
2581/* Transaction based IO helpers */
2582
2583/*
2584 * An argresp is stored in an allocated page and holds the
2585 * size of the argument or response, along with its content
2586 */
2587struct simple_transaction_argresp {
2588 ssize_t size;
2589 char data[0];
2590};
2591
2592#define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2593
2594char *simple_transaction_get(struct file *file, const char __user *buf,
2595 size_t size);
2596ssize_t simple_transaction_read(struct file *file, char __user *buf,
2597 size_t size, loff_t *pos);
2598int simple_transaction_release(struct inode *inode, struct file *file);
2599
2600void simple_transaction_set(struct file *file, size_t n);
2601
2602/*
2603 * simple attribute files
2604 *
2605 * These attributes behave similar to those in sysfs:
2606 *
2607 * Writing to an attribute immediately sets a value, an open file can be
2608 * written to multiple times.
2609 *
2610 * Reading from an attribute creates a buffer from the value that might get
2611 * read with multiple read calls. When the attribute has been read
2612 * completely, no further read calls are possible until the file is opened
2613 * again.
2614 *
2615 * All attributes contain a text representation of a numeric value
2616 * that are accessed with the get() and set() functions.
2617 */
2618#define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2619static int __fops ## _open(struct inode *inode, struct file *file) \
2620{ \
2621 __simple_attr_check_format(__fmt, 0ull); \
2622 return simple_attr_open(inode, file, __get, __set, __fmt); \
2623} \
2624static const struct file_operations __fops = { \
2625 .owner = THIS_MODULE, \
2626 .open = __fops ## _open, \
2627 .release = simple_attr_release, \
2628 .read = simple_attr_read, \
2629 .write = simple_attr_write, \
2630 .llseek = generic_file_llseek, \
2631};
2632
2633static inline __printf(1, 2)
2634void __simple_attr_check_format(const char *fmt, ...)
2635{
2636 /* don't do anything, just let the compiler check the arguments; */
2637}
2638
2639int simple_attr_open(struct inode *inode, struct file *file,
2640 int (*get)(void *, u64 *), int (*set)(void *, u64),
2641 const char *fmt);
2642int simple_attr_release(struct inode *inode, struct file *file);
2643ssize_t simple_attr_read(struct file *file, char __user *buf,
2644 size_t len, loff_t *ppos);
2645ssize_t simple_attr_write(struct file *file, const char __user *buf,
2646 size_t len, loff_t *ppos);
2647
2648struct ctl_table;
2649int proc_nr_files(struct ctl_table *table, int write,
2650 void __user *buffer, size_t *lenp, loff_t *ppos);
2651int proc_nr_dentry(struct ctl_table *table, int write,
2652 void __user *buffer, size_t *lenp, loff_t *ppos);
2653int proc_nr_inodes(struct ctl_table *table, int write,
2654 void __user *buffer, size_t *lenp, loff_t *ppos);
2655int __init get_filesystem_list(char *buf);
2656
2657#define __FMODE_EXEC ((__force int) FMODE_EXEC)
2658#define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
2659
2660#define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2661#define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2662 (flag & __FMODE_NONOTIFY)))
2663
2664static inline int is_sxid(umode_t mode)
2665{
2666 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
2667}
2668
2669static inline void inode_has_no_xattr(struct inode *inode)
2670{
2671 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
2672 inode->i_flags |= S_NOSEC;
2673}
2674
2675#endif /* __KERNEL__ */
2676#endif /* _LINUX_FS_H */