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