[PATCH] DocBook: fix <void/> xml tag
[GitHub/mt8127/android_kernel_alcatel_ttab.git] / include / linux / fs.h
CommitLineData
1da177e4
LT
1#ifndef _LINUX_FS_H
2#define _LINUX_FS_H
3
4/*
5 * This file has definitions for some important file table
6 * structures etc.
7 */
8
9#include <linux/config.h>
10#include <linux/limits.h>
11#include <linux/ioctl.h>
12
13/*
14 * It's silly to have NR_OPEN bigger than NR_FILE, but you can change
15 * the file limit at runtime and only root can increase the per-process
16 * nr_file rlimit, so it's safe to set up a ridiculously high absolute
17 * upper limit on files-per-process.
18 *
19 * Some programs (notably those using select()) may have to be
20 * recompiled to take full advantage of the new limits..
21 */
22
23/* Fixed constants first: */
24#undef NR_OPEN
25#define NR_OPEN (1024*1024) /* Absolute upper limit on fd num */
26#define INR_OPEN 1024 /* Initial setting for nfile rlimits */
27
28#define BLOCK_SIZE_BITS 10
29#define BLOCK_SIZE (1<<BLOCK_SIZE_BITS)
30
31/* And dynamically-tunable limits and defaults: */
32struct files_stat_struct {
33 int nr_files; /* read only */
34 int nr_free_files; /* read only */
35 int max_files; /* tunable */
36};
37extern struct files_stat_struct files_stat;
38
39struct inodes_stat_t {
40 int nr_inodes;
41 int nr_unused;
42 int dummy[5];
43};
44extern struct inodes_stat_t inodes_stat;
45
46extern int leases_enable, lease_break_time;
47
48#ifdef CONFIG_DNOTIFY
49extern int dir_notify_enable;
50#endif
51
52#define NR_FILE 8192 /* this can well be larger on a larger system */
53
54#define MAY_EXEC 1
55#define MAY_WRITE 2
56#define MAY_READ 4
57#define MAY_APPEND 8
58
59#define FMODE_READ 1
60#define FMODE_WRITE 2
61
62/* Internal kernel extensions */
63#define FMODE_LSEEK 4
64#define FMODE_PREAD 8
65#define FMODE_PWRITE FMODE_PREAD /* These go hand in hand */
66
67#define RW_MASK 1
68#define RWA_MASK 2
69#define READ 0
70#define WRITE 1
71#define READA 2 /* read-ahead - don't block if no resources */
72#define SPECIAL 4 /* For non-blockdevice requests in request queue */
73#define READ_SYNC (READ | (1 << BIO_RW_SYNC))
74#define WRITE_SYNC (WRITE | (1 << BIO_RW_SYNC))
75#define WRITE_BARRIER ((1 << BIO_RW) | (1 << BIO_RW_BARRIER))
76
77#define SEL_IN 1
78#define SEL_OUT 2
79#define SEL_EX 4
80
81/* public flags for file_system_type */
82#define FS_REQUIRES_DEV 1
83#define FS_BINARY_MOUNTDATA 2
84#define FS_REVAL_DOT 16384 /* Check the paths ".", ".." for staleness */
85#define FS_ODD_RENAME 32768 /* Temporary stuff; will go away as soon
86 * as nfs_rename() will be cleaned up
87 */
88/*
89 * These are the fs-independent mount-flags: up to 32 flags are supported
90 */
91#define MS_RDONLY 1 /* Mount read-only */
92#define MS_NOSUID 2 /* Ignore suid and sgid bits */
93#define MS_NODEV 4 /* Disallow access to device special files */
94#define MS_NOEXEC 8 /* Disallow program execution */
95#define MS_SYNCHRONOUS 16 /* Writes are synced at once */
96#define MS_REMOUNT 32 /* Alter flags of a mounted FS */
97#define MS_MANDLOCK 64 /* Allow mandatory locks on an FS */
98#define MS_DIRSYNC 128 /* Directory modifications are synchronous */
99#define MS_NOATIME 1024 /* Do not update access times. */
100#define MS_NODIRATIME 2048 /* Do not update directory access times */
101#define MS_BIND 4096
102#define MS_MOVE 8192
103#define MS_REC 16384
104#define MS_VERBOSE 32768
105#define MS_POSIXACL (1<<16) /* VFS does not apply the umask */
106#define MS_ACTIVE (1<<30)
107#define MS_NOUSER (1<<31)
108
109/*
110 * Superblock flags that can be altered by MS_REMOUNT
111 */
112#define MS_RMT_MASK (MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK|MS_NOATIME|\
113 MS_NODIRATIME)
114
115/*
116 * Old magic mount flag and mask
117 */
118#define MS_MGC_VAL 0xC0ED0000
119#define MS_MGC_MSK 0xffff0000
120
121/* Inode flags - they have nothing to superblock flags now */
122
123#define S_SYNC 1 /* Writes are synced at once */
124#define S_NOATIME 2 /* Do not update access times */
125#define S_APPEND 4 /* Append-only file */
126#define S_IMMUTABLE 8 /* Immutable file */
127#define S_DEAD 16 /* removed, but still open directory */
128#define S_NOQUOTA 32 /* Inode is not counted to quota */
129#define S_DIRSYNC 64 /* Directory modifications are synchronous */
130#define S_NOCMTIME 128 /* Do not update file c/mtime */
131#define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
132#define S_PRIVATE 512 /* Inode is fs-internal */
133
134/*
135 * Note that nosuid etc flags are inode-specific: setting some file-system
136 * flags just means all the inodes inherit those flags by default. It might be
137 * possible to override it selectively if you really wanted to with some
138 * ioctl() that is not currently implemented.
139 *
140 * Exception: MS_RDONLY is always applied to the entire file system.
141 *
142 * Unfortunately, it is possible to change a filesystems flags with it mounted
143 * with files in use. This means that all of the inodes will not have their
144 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
145 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
146 */
147#define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))
148
149#define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
150#define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
151 ((inode)->i_flags & S_SYNC))
152#define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
153 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
154#define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
155
156#define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
157#define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
158#define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
159#define IS_NOATIME(inode) (__IS_FLG(inode, MS_NOATIME) || ((inode)->i_flags & S_NOATIME))
160#define IS_NODIRATIME(inode) __IS_FLG(inode, MS_NODIRATIME)
161#define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
162
163#define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
164#define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
165#define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
166#define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
167
168/* the read-only stuff doesn't really belong here, but any other place is
169 probably as bad and I don't want to create yet another include file. */
170
171#define BLKROSET _IO(0x12,93) /* set device read-only (0 = read-write) */
172#define BLKROGET _IO(0x12,94) /* get read-only status (0 = read_write) */
173#define BLKRRPART _IO(0x12,95) /* re-read partition table */
174#define BLKGETSIZE _IO(0x12,96) /* return device size /512 (long *arg) */
175#define BLKFLSBUF _IO(0x12,97) /* flush buffer cache */
176#define BLKRASET _IO(0x12,98) /* set read ahead for block device */
177#define BLKRAGET _IO(0x12,99) /* get current read ahead setting */
178#define BLKFRASET _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
179#define BLKFRAGET _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
180#define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
181#define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
182#define BLKSSZGET _IO(0x12,104)/* get block device sector size */
183#if 0
184#define BLKPG _IO(0x12,105)/* See blkpg.h */
185
186/* Some people are morons. Do not use sizeof! */
187
188#define BLKELVGET _IOR(0x12,106,size_t)/* elevator get */
189#define BLKELVSET _IOW(0x12,107,size_t)/* elevator set */
190/* This was here just to show that the number is taken -
191 probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
192#endif
193/* A jump here: 108-111 have been used for various private purposes. */
194#define BLKBSZGET _IOR(0x12,112,size_t)
195#define BLKBSZSET _IOW(0x12,113,size_t)
196#define BLKGETSIZE64 _IOR(0x12,114,size_t) /* return device size in bytes (u64 *arg) */
197
198#define BMAP_IOCTL 1 /* obsolete - kept for compatibility */
199#define FIBMAP _IO(0x00,1) /* bmap access */
200#define FIGETBSZ _IO(0x00,2) /* get the block size used for bmap */
201
202#ifdef __KERNEL__
203
204#include <linux/linkage.h>
205#include <linux/wait.h>
206#include <linux/types.h>
207#include <linux/kdev_t.h>
208#include <linux/dcache.h>
209#include <linux/stat.h>
210#include <linux/cache.h>
211#include <linux/kobject.h>
212#include <linux/list.h>
213#include <linux/radix-tree.h>
214#include <linux/prio_tree.h>
215#include <linux/init.h>
216
217#include <asm/atomic.h>
218#include <asm/semaphore.h>
219#include <asm/byteorder.h>
220
221struct iovec;
222struct nameidata;
223struct pipe_inode_info;
224struct poll_table_struct;
225struct kstatfs;
226struct vm_area_struct;
227struct vfsmount;
228
229/* Used to be a macro which just called the function, now just a function */
230extern void update_atime (struct inode *);
231
232extern void __init inode_init(unsigned long);
233extern void __init inode_init_early(void);
234extern void __init mnt_init(unsigned long);
235extern void __init files_init(unsigned long);
236
237struct buffer_head;
238typedef int (get_block_t)(struct inode *inode, sector_t iblock,
239 struct buffer_head *bh_result, int create);
240typedef int (get_blocks_t)(struct inode *inode, sector_t iblock,
241 unsigned long max_blocks,
242 struct buffer_head *bh_result, int create);
243typedef void (dio_iodone_t)(struct inode *inode, loff_t offset,
244 ssize_t bytes, void *private);
245
246/*
247 * Attribute flags. These should be or-ed together to figure out what
248 * has been changed!
249 */
250#define ATTR_MODE 1
251#define ATTR_UID 2
252#define ATTR_GID 4
253#define ATTR_SIZE 8
254#define ATTR_ATIME 16
255#define ATTR_MTIME 32
256#define ATTR_CTIME 64
257#define ATTR_ATIME_SET 128
258#define ATTR_MTIME_SET 256
259#define ATTR_FORCE 512 /* Not a change, but a change it */
260#define ATTR_ATTR_FLAG 1024
261#define ATTR_KILL_SUID 2048
262#define ATTR_KILL_SGID 4096
263
264/*
265 * This is the Inode Attributes structure, used for notify_change(). It
266 * uses the above definitions as flags, to know which values have changed.
267 * Also, in this manner, a Filesystem can look at only the values it cares
268 * about. Basically, these are the attributes that the VFS layer can
269 * request to change from the FS layer.
270 *
271 * Derek Atkins <warlord@MIT.EDU> 94-10-20
272 */
273struct iattr {
274 unsigned int ia_valid;
275 umode_t ia_mode;
276 uid_t ia_uid;
277 gid_t ia_gid;
278 loff_t ia_size;
279 struct timespec ia_atime;
280 struct timespec ia_mtime;
281 struct timespec ia_ctime;
282 unsigned int ia_attr_flags;
283};
284
285/*
286 * This is the inode attributes flag definitions
287 */
288#define ATTR_FLAG_SYNCRONOUS 1 /* Syncronous write */
289#define ATTR_FLAG_NOATIME 2 /* Don't update atime */
290#define ATTR_FLAG_APPEND 4 /* Append-only file */
291#define ATTR_FLAG_IMMUTABLE 8 /* Immutable file */
292#define ATTR_FLAG_NODIRATIME 16 /* Don't update atime for directory */
293
294/*
295 * Includes for diskquotas.
296 */
297#include <linux/quota.h>
298
299/*
300 * oh the beauties of C type declarations.
301 */
302struct page;
303struct address_space;
304struct writeback_control;
305struct kiocb;
306
307struct address_space_operations {
308 int (*writepage)(struct page *page, struct writeback_control *wbc);
309 int (*readpage)(struct file *, struct page *);
310 int (*sync_page)(struct page *);
311
312 /* Write back some dirty pages from this mapping. */
313 int (*writepages)(struct address_space *, struct writeback_control *);
314
315 /* Set a page dirty */
316 int (*set_page_dirty)(struct page *page);
317
318 int (*readpages)(struct file *filp, struct address_space *mapping,
319 struct list_head *pages, unsigned nr_pages);
320
321 /*
322 * ext3 requires that a successful prepare_write() call be followed
323 * by a commit_write() call - they must be balanced
324 */
325 int (*prepare_write)(struct file *, struct page *, unsigned, unsigned);
326 int (*commit_write)(struct file *, struct page *, unsigned, unsigned);
327 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
328 sector_t (*bmap)(struct address_space *, sector_t);
329 int (*invalidatepage) (struct page *, unsigned long);
330 int (*releasepage) (struct page *, int);
331 ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
332 loff_t offset, unsigned long nr_segs);
333};
334
335struct backing_dev_info;
336struct address_space {
337 struct inode *host; /* owner: inode, block_device */
338 struct radix_tree_root page_tree; /* radix tree of all pages */
339 rwlock_t tree_lock; /* and rwlock protecting it */
340 unsigned int i_mmap_writable;/* count VM_SHARED mappings */
341 struct prio_tree_root i_mmap; /* tree of private and shared mappings */
342 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
343 spinlock_t i_mmap_lock; /* protect tree, count, list */
344 unsigned int truncate_count; /* Cover race condition with truncate */
345 unsigned long nrpages; /* number of total pages */
346 pgoff_t writeback_index;/* writeback starts here */
347 struct address_space_operations *a_ops; /* methods */
348 unsigned long flags; /* error bits/gfp mask */
349 struct backing_dev_info *backing_dev_info; /* device readahead, etc */
350 spinlock_t private_lock; /* for use by the address_space */
351 struct list_head private_list; /* ditto */
352 struct address_space *assoc_mapping; /* ditto */
353} __attribute__((aligned(sizeof(long))));
354 /*
355 * On most architectures that alignment is already the case; but
356 * must be enforced here for CRIS, to let the least signficant bit
357 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
358 */
359
360struct block_device {
361 dev_t bd_dev; /* not a kdev_t - it's a search key */
362 struct inode * bd_inode; /* will die */
363 int bd_openers;
364 struct semaphore bd_sem; /* open/close mutex */
365 struct semaphore bd_mount_sem; /* mount mutex */
366 struct list_head bd_inodes;
367 void * bd_holder;
368 int bd_holders;
369 struct block_device * bd_contains;
370 unsigned bd_block_size;
371 struct hd_struct * bd_part;
372 /* number of times partitions within this device have been opened. */
373 unsigned bd_part_count;
374 int bd_invalidated;
375 struct gendisk * bd_disk;
376 struct list_head bd_list;
377 struct backing_dev_info *bd_inode_backing_dev_info;
378 /*
379 * Private data. You must have bd_claim'ed the block_device
380 * to use this. NOTE: bd_claim allows an owner to claim
381 * the same device multiple times, the owner must take special
382 * care to not mess up bd_private for that case.
383 */
384 unsigned long bd_private;
385};
386
387/*
388 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
389 * radix trees
390 */
391#define PAGECACHE_TAG_DIRTY 0
392#define PAGECACHE_TAG_WRITEBACK 1
393
394int mapping_tagged(struct address_space *mapping, int tag);
395
396/*
397 * Might pages of this file be mapped into userspace?
398 */
399static inline int mapping_mapped(struct address_space *mapping)
400{
401 return !prio_tree_empty(&mapping->i_mmap) ||
402 !list_empty(&mapping->i_mmap_nonlinear);
403}
404
405/*
406 * Might pages of this file have been modified in userspace?
407 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
408 * marks vma as VM_SHARED if it is shared, and the file was opened for
409 * writing i.e. vma may be mprotected writable even if now readonly.
410 */
411static inline int mapping_writably_mapped(struct address_space *mapping)
412{
413 return mapping->i_mmap_writable != 0;
414}
415
416/*
417 * Use sequence counter to get consistent i_size on 32-bit processors.
418 */
419#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
420#include <linux/seqlock.h>
421#define __NEED_I_SIZE_ORDERED
422#define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
423#else
424#define i_size_ordered_init(inode) do { } while (0)
425#endif
426
427struct inode {
428 struct hlist_node i_hash;
429 struct list_head i_list;
430 struct list_head i_sb_list;
431 struct list_head i_dentry;
432 unsigned long i_ino;
433 atomic_t i_count;
434 umode_t i_mode;
435 unsigned int i_nlink;
436 uid_t i_uid;
437 gid_t i_gid;
438 dev_t i_rdev;
439 loff_t i_size;
440 struct timespec i_atime;
441 struct timespec i_mtime;
442 struct timespec i_ctime;
443 unsigned int i_blkbits;
444 unsigned long i_blksize;
445 unsigned long i_version;
446 unsigned long i_blocks;
447 unsigned short i_bytes;
448 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
449 struct semaphore i_sem;
450 struct rw_semaphore i_alloc_sem;
451 struct inode_operations *i_op;
452 struct file_operations *i_fop; /* former ->i_op->default_file_ops */
453 struct super_block *i_sb;
454 struct file_lock *i_flock;
455 struct address_space *i_mapping;
456 struct address_space i_data;
457#ifdef CONFIG_QUOTA
458 struct dquot *i_dquot[MAXQUOTAS];
459#endif
460 /* These three should probably be a union */
461 struct list_head i_devices;
462 struct pipe_inode_info *i_pipe;
463 struct block_device *i_bdev;
464 struct cdev *i_cdev;
465 int i_cindex;
466
467 __u32 i_generation;
468
469#ifdef CONFIG_DNOTIFY
470 unsigned long i_dnotify_mask; /* Directory notify events */
471 struct dnotify_struct *i_dnotify; /* for directory notifications */
472#endif
473
474 unsigned long i_state;
475 unsigned long dirtied_when; /* jiffies of first dirtying */
476
477 unsigned int i_flags;
478
479 atomic_t i_writecount;
480 void *i_security;
481 union {
482 void *generic_ip;
483 } u;
484#ifdef __NEED_I_SIZE_ORDERED
485 seqcount_t i_size_seqcount;
486#endif
487};
488
489/*
490 * NOTE: in a 32bit arch with a preemptable kernel and
491 * an UP compile the i_size_read/write must be atomic
492 * with respect to the local cpu (unlike with preempt disabled),
493 * but they don't need to be atomic with respect to other cpus like in
494 * true SMP (so they need either to either locally disable irq around
495 * the read or for example on x86 they can be still implemented as a
496 * cmpxchg8b without the need of the lock prefix). For SMP compiles
497 * and 64bit archs it makes no difference if preempt is enabled or not.
498 */
499static inline loff_t i_size_read(struct inode *inode)
500{
501#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
502 loff_t i_size;
503 unsigned int seq;
504
505 do {
506 seq = read_seqcount_begin(&inode->i_size_seqcount);
507 i_size = inode->i_size;
508 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
509 return i_size;
510#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
511 loff_t i_size;
512
513 preempt_disable();
514 i_size = inode->i_size;
515 preempt_enable();
516 return i_size;
517#else
518 return inode->i_size;
519#endif
520}
521
522
523static inline void i_size_write(struct inode *inode, loff_t i_size)
524{
525#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
526 write_seqcount_begin(&inode->i_size_seqcount);
527 inode->i_size = i_size;
528 write_seqcount_end(&inode->i_size_seqcount);
529#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
530 preempt_disable();
531 inode->i_size = i_size;
532 preempt_enable();
533#else
534 inode->i_size = i_size;
535#endif
536}
537
538static inline unsigned iminor(struct inode *inode)
539{
540 return MINOR(inode->i_rdev);
541}
542
543static inline unsigned imajor(struct inode *inode)
544{
545 return MAJOR(inode->i_rdev);
546}
547
548extern struct block_device *I_BDEV(struct inode *inode);
549
550struct fown_struct {
551 rwlock_t lock; /* protects pid, uid, euid fields */
552 int pid; /* pid or -pgrp where SIGIO should be sent */
553 uid_t uid, euid; /* uid/euid of process setting the owner */
554 void *security;
555 int signum; /* posix.1b rt signal to be delivered on IO */
556};
557
558/*
559 * Track a single file's readahead state
560 */
561struct file_ra_state {
562 unsigned long start; /* Current window */
563 unsigned long size;
564 unsigned long flags; /* ra flags RA_FLAG_xxx*/
565 unsigned long cache_hit; /* cache hit count*/
566 unsigned long prev_page; /* Cache last read() position */
567 unsigned long ahead_start; /* Ahead window */
568 unsigned long ahead_size;
569 unsigned long ra_pages; /* Maximum readahead window */
570 unsigned long mmap_hit; /* Cache hit stat for mmap accesses */
571 unsigned long mmap_miss; /* Cache miss stat for mmap accesses */
572};
573#define RA_FLAG_MISS 0x01 /* a cache miss occured against this file */
574#define RA_FLAG_INCACHE 0x02 /* file is already in cache */
575
576struct file {
577 struct list_head f_list;
578 struct dentry *f_dentry;
579 struct vfsmount *f_vfsmnt;
580 struct file_operations *f_op;
581 atomic_t f_count;
582 unsigned int f_flags;
583 mode_t f_mode;
584 int f_error;
585 loff_t f_pos;
586 struct fown_struct f_owner;
587 unsigned int f_uid, f_gid;
588 struct file_ra_state f_ra;
589
590 size_t f_maxcount;
591 unsigned long f_version;
592 void *f_security;
593
594 /* needed for tty driver, and maybe others */
595 void *private_data;
596
597#ifdef CONFIG_EPOLL
598 /* Used by fs/eventpoll.c to link all the hooks to this file */
599 struct list_head f_ep_links;
600 spinlock_t f_ep_lock;
601#endif /* #ifdef CONFIG_EPOLL */
602 struct address_space *f_mapping;
603};
604extern spinlock_t files_lock;
605#define file_list_lock() spin_lock(&files_lock);
606#define file_list_unlock() spin_unlock(&files_lock);
607
608#define get_file(x) atomic_inc(&(x)->f_count)
609#define file_count(x) atomic_read(&(x)->f_count)
610
611#define MAX_NON_LFS ((1UL<<31) - 1)
612
613/* Page cache limit. The filesystems should put that into their s_maxbytes
614 limits, otherwise bad things can happen in VM. */
615#if BITS_PER_LONG==32
616#define MAX_LFS_FILESIZE (((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
617#elif BITS_PER_LONG==64
618#define MAX_LFS_FILESIZE 0x7fffffffffffffffUL
619#endif
620
621#define FL_POSIX 1
622#define FL_FLOCK 2
623#define FL_ACCESS 8 /* not trying to lock, just looking */
624#define FL_LOCKD 16 /* lock held by rpc.lockd */
625#define FL_LEASE 32 /* lease held on this file */
626#define FL_SLEEP 128 /* A blocking lock */
627
628/*
629 * The POSIX file lock owner is determined by
630 * the "struct files_struct" in the thread group
631 * (or NULL for no owner - BSD locks).
632 *
633 * Lockd stuffs a "host" pointer into this.
634 */
635typedef struct files_struct *fl_owner_t;
636
637struct file_lock_operations {
638 void (*fl_insert)(struct file_lock *); /* lock insertion callback */
639 void (*fl_remove)(struct file_lock *); /* lock removal callback */
640 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
641 void (*fl_release_private)(struct file_lock *);
642};
643
644struct lock_manager_operations {
645 int (*fl_compare_owner)(struct file_lock *, struct file_lock *);
646 void (*fl_notify)(struct file_lock *); /* unblock callback */
647 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
648 void (*fl_release_private)(struct file_lock *);
649 void (*fl_break)(struct file_lock *);
650 int (*fl_mylease)(struct file_lock *, struct file_lock *);
651 int (*fl_change)(struct file_lock **, int);
652};
653
654/* that will die - we need it for nfs_lock_info */
655#include <linux/nfs_fs_i.h>
656
657struct file_lock {
658 struct file_lock *fl_next; /* singly linked list for this inode */
659 struct list_head fl_link; /* doubly linked list of all locks */
660 struct list_head fl_block; /* circular list of blocked processes */
661 fl_owner_t fl_owner;
662 unsigned int fl_pid;
663 wait_queue_head_t fl_wait;
664 struct file *fl_file;
665 unsigned char fl_flags;
666 unsigned char fl_type;
667 loff_t fl_start;
668 loff_t fl_end;
669
670 struct fasync_struct * fl_fasync; /* for lease break notifications */
671 unsigned long fl_break_time; /* for nonblocking lease breaks */
672
673 struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
674 struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
675 union {
676 struct nfs_lock_info nfs_fl;
677 } fl_u;
678};
679
680/* The following constant reflects the upper bound of the file/locking space */
681#ifndef OFFSET_MAX
682#define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
683#define OFFSET_MAX INT_LIMIT(loff_t)
684#define OFFT_OFFSET_MAX INT_LIMIT(off_t)
685#endif
686
687extern struct list_head file_lock_list;
688
689#include <linux/fcntl.h>
690
691extern int fcntl_getlk(struct file *, struct flock __user *);
692extern int fcntl_setlk(struct file *, unsigned int, struct flock __user *);
693
694#if BITS_PER_LONG == 32
695extern int fcntl_getlk64(struct file *, struct flock64 __user *);
696extern int fcntl_setlk64(struct file *, unsigned int, struct flock64 __user *);
697#endif
698
699extern void send_sigio(struct fown_struct *fown, int fd, int band);
700extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
701extern int fcntl_getlease(struct file *filp);
702
703/* fs/locks.c */
704extern void locks_init_lock(struct file_lock *);
705extern void locks_copy_lock(struct file_lock *, struct file_lock *);
706extern void locks_remove_posix(struct file *, fl_owner_t);
707extern void locks_remove_flock(struct file *);
708extern struct file_lock *posix_test_lock(struct file *, struct file_lock *);
709extern int posix_lock_file(struct file *, struct file_lock *);
710extern int posix_lock_file_wait(struct file *, struct file_lock *);
711extern void posix_block_lock(struct file_lock *, struct file_lock *);
712extern void posix_unblock_lock(struct file *, struct file_lock *);
713extern int posix_locks_deadlock(struct file_lock *, struct file_lock *);
714extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
715extern int __break_lease(struct inode *inode, unsigned int flags);
716extern void lease_get_mtime(struct inode *, struct timespec *time);
717extern int setlease(struct file *, long, struct file_lock **);
718extern int lease_modify(struct file_lock **, int);
719extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
720extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
721extern void steal_locks(fl_owner_t from);
722
723struct fasync_struct {
724 int magic;
725 int fa_fd;
726 struct fasync_struct *fa_next; /* singly linked list */
727 struct file *fa_file;
728};
729
730#define FASYNC_MAGIC 0x4601
731
732/* SMP safe fasync helpers: */
733extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
734/* can be called from interrupts */
735extern void kill_fasync(struct fasync_struct **, int, int);
736/* only for net: no internal synchronization */
737extern void __kill_fasync(struct fasync_struct *, int, int);
738
739extern int f_setown(struct file *filp, unsigned long arg, int force);
740extern void f_delown(struct file *filp);
741extern int send_sigurg(struct fown_struct *fown);
742
743/*
744 * Umount options
745 */
746
747#define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
748#define MNT_DETACH 0x00000002 /* Just detach from the tree */
749#define MNT_EXPIRE 0x00000004 /* Mark for expiry */
750
751extern struct list_head super_blocks;
752extern spinlock_t sb_lock;
753
754#define sb_entry(list) list_entry((list), struct super_block, s_list)
755#define S_BIAS (1<<30)
756struct super_block {
757 struct list_head s_list; /* Keep this first */
758 dev_t s_dev; /* search index; _not_ kdev_t */
759 unsigned long s_blocksize;
760 unsigned long s_old_blocksize;
761 unsigned char s_blocksize_bits;
762 unsigned char s_dirt;
763 unsigned long long s_maxbytes; /* Max file size */
764 struct file_system_type *s_type;
765 struct super_operations *s_op;
766 struct dquot_operations *dq_op;
767 struct quotactl_ops *s_qcop;
768 struct export_operations *s_export_op;
769 unsigned long s_flags;
770 unsigned long s_magic;
771 struct dentry *s_root;
772 struct rw_semaphore s_umount;
773 struct semaphore s_lock;
774 int s_count;
775 int s_syncing;
776 int s_need_sync_fs;
777 atomic_t s_active;
778 void *s_security;
779 struct xattr_handler **s_xattr;
780
781 struct list_head s_inodes; /* all inodes */
782 struct list_head s_dirty; /* dirty inodes */
783 struct list_head s_io; /* parked for writeback */
784 struct hlist_head s_anon; /* anonymous dentries for (nfs) exporting */
785 struct list_head s_files;
786
787 struct block_device *s_bdev;
788 struct list_head s_instances;
789 struct quota_info s_dquot; /* Diskquota specific options */
790
791 int s_frozen;
792 wait_queue_head_t s_wait_unfrozen;
793
794 char s_id[32]; /* Informational name */
795
796 void *s_fs_info; /* Filesystem private info */
797
798 /*
799 * The next field is for VFS *only*. No filesystems have any business
800 * even looking at it. You had been warned.
801 */
802 struct semaphore s_vfs_rename_sem; /* Kludge */
803
804 /* Granuality of c/m/atime in ns.
805 Cannot be worse than a second */
806 u32 s_time_gran;
807};
808
809extern struct timespec current_fs_time(struct super_block *sb);
810
811/*
812 * Snapshotting support.
813 */
814enum {
815 SB_UNFROZEN = 0,
816 SB_FREEZE_WRITE = 1,
817 SB_FREEZE_TRANS = 2,
818};
819
820#define vfs_check_frozen(sb, level) \
821 wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
822
823/*
824 * Superblock locking.
825 */
826static inline void lock_super(struct super_block * sb)
827{
828 down(&sb->s_lock);
829}
830
831static inline void unlock_super(struct super_block * sb)
832{
833 up(&sb->s_lock);
834}
835
836/*
837 * VFS helper functions..
838 */
839extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *);
840extern int vfs_mkdir(struct inode *, struct dentry *, int);
841extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t);
842extern int vfs_symlink(struct inode *, struct dentry *, const char *, int);
843extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
844extern int vfs_rmdir(struct inode *, struct dentry *);
845extern int vfs_unlink(struct inode *, struct dentry *);
846extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
847
848/*
849 * VFS dentry helper functions.
850 */
851extern void dentry_unhash(struct dentry *dentry);
852
853/*
854 * File types
855 *
856 * NOTE! These match bits 12..15 of stat.st_mode
857 * (ie "(i_mode >> 12) & 15").
858 */
859#define DT_UNKNOWN 0
860#define DT_FIFO 1
861#define DT_CHR 2
862#define DT_DIR 4
863#define DT_BLK 6
864#define DT_REG 8
865#define DT_LNK 10
866#define DT_SOCK 12
867#define DT_WHT 14
868
869#define OSYNC_METADATA (1<<0)
870#define OSYNC_DATA (1<<1)
871#define OSYNC_INODE (1<<2)
872int generic_osync_inode(struct inode *, struct address_space *, int);
873
874/*
875 * This is the "filldir" function type, used by readdir() to let
876 * the kernel specify what kind of dirent layout it wants to have.
877 * This allows the kernel to read directories into kernel space or
878 * to have different dirent layouts depending on the binary type.
879 */
880typedef int (*filldir_t)(void *, const char *, int, loff_t, ino_t, unsigned);
881
882struct block_device_operations {
883 int (*open) (struct inode *, struct file *);
884 int (*release) (struct inode *, struct file *);
885 int (*ioctl) (struct inode *, struct file *, unsigned, unsigned long);
886 long (*compat_ioctl) (struct file *, unsigned, unsigned long);
887 int (*media_changed) (struct gendisk *);
888 int (*revalidate_disk) (struct gendisk *);
889 struct module *owner;
890};
891
892/*
893 * "descriptor" for what we're up to with a read for sendfile().
894 * This allows us to use the same read code yet
895 * have multiple different users of the data that
896 * we read from a file.
897 *
898 * The simplest case just copies the data to user
899 * mode.
900 */
901typedef struct {
902 size_t written;
903 size_t count;
904 union {
905 char __user * buf;
906 void *data;
907 } arg;
908 int error;
909} read_descriptor_t;
910
911typedef int (*read_actor_t)(read_descriptor_t *, struct page *, unsigned long, unsigned long);
912
913/* These macros are for out of kernel modules to test that
914 * the kernel supports the unlocked_ioctl and compat_ioctl
915 * fields in struct file_operations. */
916#define HAVE_COMPAT_IOCTL 1
917#define HAVE_UNLOCKED_IOCTL 1
918
919/*
920 * NOTE:
921 * read, write, poll, fsync, readv, writev, unlocked_ioctl and compat_ioctl
922 * can be called without the big kernel lock held in all filesystems.
923 */
924struct file_operations {
925 struct module *owner;
926 loff_t (*llseek) (struct file *, loff_t, int);
927 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
928 ssize_t (*aio_read) (struct kiocb *, char __user *, size_t, loff_t);
929 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
930 ssize_t (*aio_write) (struct kiocb *, const char __user *, size_t, loff_t);
931 int (*readdir) (struct file *, void *, filldir_t);
932 unsigned int (*poll) (struct file *, struct poll_table_struct *);
933 int (*ioctl) (struct inode *, struct file *, unsigned int, unsigned long);
934 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
935 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
936 int (*mmap) (struct file *, struct vm_area_struct *);
937 int (*open) (struct inode *, struct file *);
938 int (*flush) (struct file *);
939 int (*release) (struct inode *, struct file *);
940 int (*fsync) (struct file *, struct dentry *, int datasync);
941 int (*aio_fsync) (struct kiocb *, int datasync);
942 int (*fasync) (int, struct file *, int);
943 int (*lock) (struct file *, int, struct file_lock *);
944 ssize_t (*readv) (struct file *, const struct iovec *, unsigned long, loff_t *);
945 ssize_t (*writev) (struct file *, const struct iovec *, unsigned long, loff_t *);
946 ssize_t (*sendfile) (struct file *, loff_t *, size_t, read_actor_t, void *);
947 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
948 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
949 int (*check_flags)(int);
950 int (*dir_notify)(struct file *filp, unsigned long arg);
951 int (*flock) (struct file *, int, struct file_lock *);
952};
953
954struct inode_operations {
955 int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
956 struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
957 int (*link) (struct dentry *,struct inode *,struct dentry *);
958 int (*unlink) (struct inode *,struct dentry *);
959 int (*symlink) (struct inode *,struct dentry *,const char *);
960 int (*mkdir) (struct inode *,struct dentry *,int);
961 int (*rmdir) (struct inode *,struct dentry *);
962 int (*mknod) (struct inode *,struct dentry *,int,dev_t);
963 int (*rename) (struct inode *, struct dentry *,
964 struct inode *, struct dentry *);
965 int (*readlink) (struct dentry *, char __user *,int);
966 int (*follow_link) (struct dentry *, struct nameidata *);
967 void (*put_link) (struct dentry *, struct nameidata *);
968 void (*truncate) (struct inode *);
969 int (*permission) (struct inode *, int, struct nameidata *);
970 int (*setattr) (struct dentry *, struct iattr *);
971 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
972 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
973 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
974 ssize_t (*listxattr) (struct dentry *, char *, size_t);
975 int (*removexattr) (struct dentry *, const char *);
976};
977
978struct seq_file;
979
980extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
981extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
982extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
983 unsigned long, loff_t *);
984extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
985 unsigned long, loff_t *);
986
987/*
988 * NOTE: write_inode, delete_inode, clear_inode, put_inode can be called
989 * without the big kernel lock held in all filesystems.
990 */
991struct super_operations {
992 struct inode *(*alloc_inode)(struct super_block *sb);
993 void (*destroy_inode)(struct inode *);
994
995 void (*read_inode) (struct inode *);
996
997 void (*dirty_inode) (struct inode *);
998 int (*write_inode) (struct inode *, int);
999 void (*put_inode) (struct inode *);
1000 void (*drop_inode) (struct inode *);
1001 void (*delete_inode) (struct inode *);
1002 void (*put_super) (struct super_block *);
1003 void (*write_super) (struct super_block *);
1004 int (*sync_fs)(struct super_block *sb, int wait);
1005 void (*write_super_lockfs) (struct super_block *);
1006 void (*unlockfs) (struct super_block *);
1007 int (*statfs) (struct super_block *, struct kstatfs *);
1008 int (*remount_fs) (struct super_block *, int *, char *);
1009 void (*clear_inode) (struct inode *);
1010 void (*umount_begin) (struct super_block *);
1011
1012 int (*show_options)(struct seq_file *, struct vfsmount *);
1013
1014 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1015 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1016};
1017
1018/* Inode state bits. Protected by inode_lock. */
1019#define I_DIRTY_SYNC 1 /* Not dirty enough for O_DATASYNC */
1020#define I_DIRTY_DATASYNC 2 /* Data-related inode changes pending */
1021#define I_DIRTY_PAGES 4 /* Data-related inode changes pending */
1022#define __I_LOCK 3
1023#define I_LOCK (1 << __I_LOCK)
1024#define I_FREEING 16
1025#define I_CLEAR 32
1026#define I_NEW 64
1027
1028#define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1029
1030extern void __mark_inode_dirty(struct inode *, int);
1031static inline void mark_inode_dirty(struct inode *inode)
1032{
1033 __mark_inode_dirty(inode, I_DIRTY);
1034}
1035
1036static inline void mark_inode_dirty_sync(struct inode *inode)
1037{
1038 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1039}
1040
1041static inline void touch_atime(struct vfsmount *mnt, struct dentry *dentry)
1042{
1043 /* per-mountpoint checks will go here */
1044 update_atime(dentry->d_inode);
1045}
1046
1047static inline void file_accessed(struct file *file)
1048{
1049 if (!(file->f_flags & O_NOATIME))
1050 touch_atime(file->f_vfsmnt, file->f_dentry);
1051}
1052
1053int sync_inode(struct inode *inode, struct writeback_control *wbc);
1054
1055/**
4dc3b16b
PP
1056 * struct export_operations - for nfsd to communicate with file systems
1057 * @decode_fh: decode a file handle fragment and return a &struct dentry
1058 * @encode_fh: encode a file handle fragment from a dentry
1059 * @get_name: find the name for a given inode in a given directory
1060 * @get_parent: find the parent of a given directory
1061 * @get_dentry: find a dentry for the inode given a file handle sub-fragment
1da177e4
LT
1062 *
1063 * Description:
1064 * The export_operations structure provides a means for nfsd to communicate
1065 * with a particular exported file system - particularly enabling nfsd and
1066 * the filesystem to co-operate when dealing with file handles.
1067 *
1068 * export_operations contains two basic operation for dealing with file handles,
1069 * decode_fh() and encode_fh(), and allows for some other operations to be defined
1070 * which standard helper routines use to get specific information from the
1071 * filesystem.
1072 *
1073 * nfsd encodes information use to determine which filesystem a filehandle
1074 * applies to in the initial part of the file handle. The remainder, termed a
1075 * file handle fragment, is controlled completely by the filesystem.
1076 * The standard helper routines assume that this fragment will contain one or two
1077 * sub-fragments, one which identifies the file, and one which may be used to
1078 * identify the (a) directory containing the file.
1079 *
1080 * In some situations, nfsd needs to get a dentry which is connected into a
1081 * specific part of the file tree. To allow for this, it passes the function
1082 * acceptable() together with a @context which can be used to see if the dentry
1083 * is acceptable. As there can be multiple dentrys for a given file, the filesystem
1084 * should check each one for acceptability before looking for the next. As soon
1085 * as an acceptable one is found, it should be returned.
1086 *
1087 * decode_fh:
1088 * @decode_fh is given a &struct super_block (@sb), a file handle fragment (@fh, @fh_len)
1089 * and an acceptability testing function (@acceptable, @context). It should return
1090 * a &struct dentry which refers to the same file that the file handle fragment refers
1091 * to, and which passes the acceptability test. If it cannot, it should return
1092 * a %NULL pointer if the file was found but no acceptable &dentries were available, or
1093 * a %ERR_PTR error code indicating why it couldn't be found (e.g. %ENOENT or %ENOMEM).
1094 *
1095 * encode_fh:
1096 * @encode_fh should store in the file handle fragment @fh (using at most @max_len bytes)
1097 * information that can be used by @decode_fh to recover the file refered to by the
1098 * &struct dentry @de. If the @connectable flag is set, the encode_fh() should store
1099 * sufficient information so that a good attempt can be made to find not only
1100 * the file but also it's place in the filesystem. This typically means storing
1101 * a reference to de->d_parent in the filehandle fragment.
1102 * encode_fh() should return the number of bytes stored or a negative error code
1103 * such as %-ENOSPC
1104 *
1105 * get_name:
1106 * @get_name should find a name for the given @child in the given @parent directory.
1107 * The name should be stored in the @name (with the understanding that it is already
1108 * pointing to a a %NAME_MAX+1 sized buffer. get_name() should return %0 on success,
1109 * a negative error code or error.
1110 * @get_name will be called without @parent->i_sem held.
1111 *
1112 * get_parent:
1113 * @get_parent should find the parent directory for the given @child which is also
1114 * a directory. In the event that it cannot be found, or storage space cannot be
1115 * allocated, a %ERR_PTR should be returned.
1116 *
1117 * get_dentry:
1118 * Given a &super_block (@sb) and a pointer to a file-system specific inode identifier,
1119 * possibly an inode number, (@inump) get_dentry() should find the identified inode and
1120 * return a dentry for that inode.
1121 * Any suitable dentry can be returned including, if necessary, a new dentry created
1122 * with d_alloc_root. The caller can then find any other extant dentrys by following the
1123 * d_alias links. If a new dentry was created using d_alloc_root, DCACHE_NFSD_DISCONNECTED
1124 * should be set, and the dentry should be d_rehash()ed.
1125 *
1126 * If the inode cannot be found, either a %NULL pointer or an %ERR_PTR code can be returned.
1127 * The @inump will be whatever was passed to nfsd_find_fh_dentry() in either the
1128 * @obj or @parent parameters.
1129 *
1130 * Locking rules:
1131 * get_parent is called with child->d_inode->i_sem down
1132 * get_name is not (which is possibly inconsistent)
1133 */
1134
1135struct export_operations {
1136 struct dentry *(*decode_fh)(struct super_block *sb, __u32 *fh, int fh_len, int fh_type,
1137 int (*acceptable)(void *context, struct dentry *de),
1138 void *context);
1139 int (*encode_fh)(struct dentry *de, __u32 *fh, int *max_len,
1140 int connectable);
1141
1142 /* the following are only called from the filesystem itself */
1143 int (*get_name)(struct dentry *parent, char *name,
1144 struct dentry *child);
1145 struct dentry * (*get_parent)(struct dentry *child);
1146 struct dentry * (*get_dentry)(struct super_block *sb, void *inump);
1147
1148 /* This is set by the exporting module to a standard helper */
1149 struct dentry * (*find_exported_dentry)(
1150 struct super_block *sb, void *obj, void *parent,
1151 int (*acceptable)(void *context, struct dentry *de),
1152 void *context);
1153
1154
1155};
1156
1157extern struct dentry *
1158find_exported_dentry(struct super_block *sb, void *obj, void *parent,
1159 int (*acceptable)(void *context, struct dentry *de),
1160 void *context);
1161
1162struct file_system_type {
1163 const char *name;
1164 int fs_flags;
1165 struct super_block *(*get_sb) (struct file_system_type *, int,
1166 const char *, void *);
1167 void (*kill_sb) (struct super_block *);
1168 struct module *owner;
1169 struct file_system_type * next;
1170 struct list_head fs_supers;
1171};
1172
1173struct super_block *get_sb_bdev(struct file_system_type *fs_type,
1174 int flags, const char *dev_name, void *data,
1175 int (*fill_super)(struct super_block *, void *, int));
1176struct super_block *get_sb_single(struct file_system_type *fs_type,
1177 int flags, void *data,
1178 int (*fill_super)(struct super_block *, void *, int));
1179struct super_block *get_sb_nodev(struct file_system_type *fs_type,
1180 int flags, void *data,
1181 int (*fill_super)(struct super_block *, void *, int));
1182void generic_shutdown_super(struct super_block *sb);
1183void kill_block_super(struct super_block *sb);
1184void kill_anon_super(struct super_block *sb);
1185void kill_litter_super(struct super_block *sb);
1186void deactivate_super(struct super_block *sb);
1187int set_anon_super(struct super_block *s, void *data);
1188struct super_block *sget(struct file_system_type *type,
1189 int (*test)(struct super_block *,void *),
1190 int (*set)(struct super_block *,void *),
1191 void *data);
1192struct super_block *get_sb_pseudo(struct file_system_type *, char *,
1193 struct super_operations *ops, unsigned long);
1194int __put_super(struct super_block *sb);
1195int __put_super_and_need_restart(struct super_block *sb);
1196void unnamed_dev_init(void);
1197
1198/* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1199#define fops_get(fops) \
1200 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1201#define fops_put(fops) \
1202 do { if (fops) module_put((fops)->owner); } while(0)
1203
1204extern int register_filesystem(struct file_system_type *);
1205extern int unregister_filesystem(struct file_system_type *);
1206extern struct vfsmount *kern_mount(struct file_system_type *);
1207extern int may_umount_tree(struct vfsmount *);
1208extern int may_umount(struct vfsmount *);
1209extern long do_mount(char *, char *, char *, unsigned long, void *);
1210
1211extern int vfs_statfs(struct super_block *, struct kstatfs *);
1212
1213#define FLOCK_VERIFY_READ 1
1214#define FLOCK_VERIFY_WRITE 2
1215
1216extern int locks_mandatory_locked(struct inode *);
1217extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1218
1219/*
1220 * Candidates for mandatory locking have the setgid bit set
1221 * but no group execute bit - an otherwise meaningless combination.
1222 */
1223#define MANDATORY_LOCK(inode) \
1224 (IS_MANDLOCK(inode) && ((inode)->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID)
1225
1226static inline int locks_verify_locked(struct inode *inode)
1227{
1228 if (MANDATORY_LOCK(inode))
1229 return locks_mandatory_locked(inode);
1230 return 0;
1231}
1232
1233extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1234
1235static inline int locks_verify_truncate(struct inode *inode,
1236 struct file *filp,
1237 loff_t size)
1238{
1239 if (inode->i_flock && MANDATORY_LOCK(inode))
1240 return locks_mandatory_area(
1241 FLOCK_VERIFY_WRITE, inode, filp,
1242 size < inode->i_size ? size : inode->i_size,
1243 (size < inode->i_size ? inode->i_size - size
1244 : size - inode->i_size)
1245 );
1246 return 0;
1247}
1248
1249static inline int break_lease(struct inode *inode, unsigned int mode)
1250{
1251 if (inode->i_flock)
1252 return __break_lease(inode, mode);
1253 return 0;
1254}
1255
1256/* fs/open.c */
1257
1258extern int do_truncate(struct dentry *, loff_t start);
1259extern struct file *filp_open(const char *, int, int);
1260extern struct file * dentry_open(struct dentry *, struct vfsmount *, int);
1261extern int filp_close(struct file *, fl_owner_t id);
1262extern char * getname(const char __user *);
1263
1264/* fs/dcache.c */
1265extern void __init vfs_caches_init_early(void);
1266extern void __init vfs_caches_init(unsigned long);
1267
1268#define __getname() kmem_cache_alloc(names_cachep, SLAB_KERNEL)
1269#define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
1270#ifndef CONFIG_AUDITSYSCALL
1271#define putname(name) __putname(name)
1272#else
1273extern void putname(const char *name);
1274#endif
1275
1276extern int register_blkdev(unsigned int, const char *);
1277extern int unregister_blkdev(unsigned int, const char *);
1278extern struct block_device *bdget(dev_t);
1279extern void bd_set_size(struct block_device *, loff_t size);
1280extern void bd_forget(struct inode *inode);
1281extern void bdput(struct block_device *);
1282extern struct block_device *open_by_devnum(dev_t, unsigned);
1283extern struct file_operations def_blk_fops;
1284extern struct address_space_operations def_blk_aops;
1285extern struct file_operations def_chr_fops;
1286extern struct file_operations bad_sock_fops;
1287extern struct file_operations def_fifo_fops;
1288extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
1289extern int blkdev_ioctl(struct inode *, struct file *, unsigned, unsigned long);
1290extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
1291extern int blkdev_get(struct block_device *, mode_t, unsigned);
1292extern int blkdev_put(struct block_device *);
1293extern int bd_claim(struct block_device *, void *);
1294extern void bd_release(struct block_device *);
1295
1296/* fs/char_dev.c */
1297extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
1298extern int register_chrdev_region(dev_t, unsigned, const char *);
1299extern int register_chrdev(unsigned int, const char *,
1300 struct file_operations *);
1301extern int unregister_chrdev(unsigned int, const char *);
1302extern void unregister_chrdev_region(dev_t, unsigned);
1303extern int chrdev_open(struct inode *, struct file *);
1304
1305/* fs/block_dev.c */
1306#define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
1307extern const char *__bdevname(dev_t, char *buffer);
1308extern const char *bdevname(struct block_device *bdev, char *buffer);
1309extern struct block_device *lookup_bdev(const char *);
1310extern struct block_device *open_bdev_excl(const char *, int, void *);
1311extern void close_bdev_excl(struct block_device *);
1312
1313extern void init_special_inode(struct inode *, umode_t, dev_t);
1314
1315/* Invalid inode operations -- fs/bad_inode.c */
1316extern void make_bad_inode(struct inode *);
1317extern int is_bad_inode(struct inode *);
1318
1319extern struct file_operations read_fifo_fops;
1320extern struct file_operations write_fifo_fops;
1321extern struct file_operations rdwr_fifo_fops;
1322extern struct file_operations read_pipe_fops;
1323extern struct file_operations write_pipe_fops;
1324extern struct file_operations rdwr_pipe_fops;
1325
1326extern int fs_may_remount_ro(struct super_block *);
1327
1328/*
1329 * return READ, READA, or WRITE
1330 */
1331#define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
1332
1333/*
1334 * return data direction, READ or WRITE
1335 */
1336#define bio_data_dir(bio) ((bio)->bi_rw & 1)
1337
1338extern int check_disk_change(struct block_device *);
1339extern int invalidate_inodes(struct super_block *);
1340extern int __invalidate_device(struct block_device *, int);
1341extern int invalidate_partition(struct gendisk *, int);
1342unsigned long invalidate_mapping_pages(struct address_space *mapping,
1343 pgoff_t start, pgoff_t end);
1344unsigned long invalidate_inode_pages(struct address_space *mapping);
1345static inline void invalidate_remote_inode(struct inode *inode)
1346{
1347 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
1348 S_ISLNK(inode->i_mode))
1349 invalidate_inode_pages(inode->i_mapping);
1350}
1351extern int invalidate_inode_pages2(struct address_space *mapping);
1352extern int invalidate_inode_pages2_range(struct address_space *mapping,
1353 pgoff_t start, pgoff_t end);
1354extern int write_inode_now(struct inode *, int);
1355extern int filemap_fdatawrite(struct address_space *);
1356extern int filemap_flush(struct address_space *);
1357extern int filemap_fdatawait(struct address_space *);
1358extern int filemap_write_and_wait(struct address_space *mapping);
1359extern int filemap_write_and_wait_range(struct address_space *mapping,
1360 loff_t lstart, loff_t lend);
1361extern void sync_supers(void);
1362extern void sync_filesystems(int wait);
1363extern void emergency_sync(void);
1364extern void emergency_remount(void);
1365extern int do_remount_sb(struct super_block *sb, int flags,
1366 void *data, int force);
1367extern sector_t bmap(struct inode *, sector_t);
1368extern int setattr_mask(unsigned int);
1369extern int notify_change(struct dentry *, struct iattr *);
1370extern int permission(struct inode *, int, struct nameidata *);
1371extern int generic_permission(struct inode *, int,
1372 int (*check_acl)(struct inode *, int));
1373
1374extern int get_write_access(struct inode *);
1375extern int deny_write_access(struct file *);
1376static inline void put_write_access(struct inode * inode)
1377{
1378 atomic_dec(&inode->i_writecount);
1379}
1380static inline void allow_write_access(struct file *file)
1381{
1382 if (file)
1383 atomic_inc(&file->f_dentry->d_inode->i_writecount);
1384}
1385extern int do_pipe(int *);
1386
1387extern int open_namei(const char *, int, int, struct nameidata *);
1388extern int may_open(struct nameidata *, int, int);
1389
1390extern int kernel_read(struct file *, unsigned long, char *, unsigned long);
1391extern struct file * open_exec(const char *);
1392
1393/* fs/dcache.c -- generic fs support functions */
1394extern int is_subdir(struct dentry *, struct dentry *);
1395extern ino_t find_inode_number(struct dentry *, struct qstr *);
1396
1397#include <linux/err.h>
1398
1399/* needed for stackable file system support */
1400extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
1401
1402extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
1403
1404extern void inode_init_once(struct inode *);
1405extern void iput(struct inode *);
1406extern struct inode * igrab(struct inode *);
1407extern ino_t iunique(struct super_block *, ino_t);
1408extern int inode_needs_sync(struct inode *inode);
1409extern void generic_delete_inode(struct inode *inode);
1410
1411extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
1412 int (*test)(struct inode *, void *), void *data);
1413extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
1414
1415extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
1416extern struct inode * iget_locked(struct super_block *, unsigned long);
1417extern void unlock_new_inode(struct inode *);
1418
1419static inline struct inode *iget(struct super_block *sb, unsigned long ino)
1420{
1421 struct inode *inode = iget_locked(sb, ino);
1422
1423 if (inode && (inode->i_state & I_NEW)) {
1424 sb->s_op->read_inode(inode);
1425 unlock_new_inode(inode);
1426 }
1427
1428 return inode;
1429}
1430
1431extern void __iget(struct inode * inode);
1432extern void clear_inode(struct inode *);
1433extern void destroy_inode(struct inode *);
1434extern struct inode *new_inode(struct super_block *);
1435extern int remove_suid(struct dentry *);
1436extern void remove_dquot_ref(struct super_block *, int, struct list_head *);
1437extern struct semaphore iprune_sem;
1438
1439extern void __insert_inode_hash(struct inode *, unsigned long hashval);
1440extern void remove_inode_hash(struct inode *);
1441static inline void insert_inode_hash(struct inode *inode) {
1442 __insert_inode_hash(inode, inode->i_ino);
1443}
1444
1445extern struct file * get_empty_filp(void);
1446extern void file_move(struct file *f, struct list_head *list);
1447extern void file_kill(struct file *f);
1448struct bio;
1449extern void submit_bio(int, struct bio *);
1450extern int bdev_read_only(struct block_device *);
1451extern int set_blocksize(struct block_device *, int);
1452extern int sb_set_blocksize(struct super_block *, int);
1453extern int sb_min_blocksize(struct super_block *, int);
1454
1455extern int generic_file_mmap(struct file *, struct vm_area_struct *);
1456extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
1457extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
1458extern int file_send_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
1459extern ssize_t generic_file_read(struct file *, char __user *, size_t, loff_t *);
1460int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
1461extern ssize_t generic_file_write(struct file *, const char __user *, size_t, loff_t *);
1462extern ssize_t generic_file_aio_read(struct kiocb *, char __user *, size_t, loff_t);
1463extern ssize_t __generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t *);
1464extern ssize_t generic_file_aio_write(struct kiocb *, const char __user *, size_t, loff_t);
1465extern ssize_t generic_file_aio_write_nolock(struct kiocb *, const struct iovec *,
1466 unsigned long, loff_t *);
1467extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
1468 unsigned long *, loff_t, loff_t *, size_t, size_t);
1469extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
1470 unsigned long, loff_t, loff_t *, size_t, ssize_t);
1471extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
1472extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
1473ssize_t generic_file_write_nolock(struct file *file, const struct iovec *iov,
1474 unsigned long nr_segs, loff_t *ppos);
1475extern ssize_t generic_file_sendfile(struct file *, loff_t *, size_t, read_actor_t, void *);
1476extern void do_generic_mapping_read(struct address_space *mapping,
1477 struct file_ra_state *, struct file *,
1478 loff_t *, read_descriptor_t *, read_actor_t);
1479extern void
1480file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
1481extern ssize_t generic_file_direct_IO(int rw, struct kiocb *iocb,
1482 const struct iovec *iov, loff_t offset, unsigned long nr_segs);
1483extern ssize_t generic_file_readv(struct file *filp, const struct iovec *iov,
1484 unsigned long nr_segs, loff_t *ppos);
1485ssize_t generic_file_writev(struct file *filp, const struct iovec *iov,
1486 unsigned long nr_segs, loff_t *ppos);
1487extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
1488extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
1489extern loff_t remote_llseek(struct file *file, loff_t offset, int origin);
1490extern int generic_file_open(struct inode * inode, struct file * filp);
1491extern int nonseekable_open(struct inode * inode, struct file * filp);
1492
1493static inline void do_generic_file_read(struct file * filp, loff_t *ppos,
1494 read_descriptor_t * desc,
1495 read_actor_t actor)
1496{
1497 do_generic_mapping_read(filp->f_mapping,
1498 &filp->f_ra,
1499 filp,
1500 ppos,
1501 desc,
1502 actor);
1503}
1504
1505ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
1506 struct block_device *bdev, const struct iovec *iov, loff_t offset,
1507 unsigned long nr_segs, get_blocks_t get_blocks, dio_iodone_t end_io,
1508 int lock_type);
1509
1510enum {
1511 DIO_LOCKING = 1, /* need locking between buffered and direct access */
1512 DIO_NO_LOCKING, /* bdev; no locking at all between buffered/direct */
1513 DIO_OWN_LOCKING, /* filesystem locks buffered and direct internally */
1514};
1515
1516static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
1517 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1518 loff_t offset, unsigned long nr_segs, get_blocks_t get_blocks,
1519 dio_iodone_t end_io)
1520{
1521 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1522 nr_segs, get_blocks, end_io, DIO_LOCKING);
1523}
1524
1525static inline ssize_t blockdev_direct_IO_no_locking(int rw, struct kiocb *iocb,
1526 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1527 loff_t offset, unsigned long nr_segs, get_blocks_t get_blocks,
1528 dio_iodone_t end_io)
1529{
1530 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1531 nr_segs, get_blocks, end_io, DIO_NO_LOCKING);
1532}
1533
1534static inline ssize_t blockdev_direct_IO_own_locking(int rw, struct kiocb *iocb,
1535 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1536 loff_t offset, unsigned long nr_segs, get_blocks_t get_blocks,
1537 dio_iodone_t end_io)
1538{
1539 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1540 nr_segs, get_blocks, end_io, DIO_OWN_LOCKING);
1541}
1542
1543extern struct file_operations generic_ro_fops;
1544
1545#define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
1546
1547extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
1548extern int vfs_follow_link(struct nameidata *, const char *);
1549extern int page_readlink(struct dentry *, char __user *, int);
1550extern int page_follow_link_light(struct dentry *, struct nameidata *);
1551extern void page_put_link(struct dentry *, struct nameidata *);
1552extern int page_symlink(struct inode *inode, const char *symname, int len);
1553extern struct inode_operations page_symlink_inode_operations;
1554extern int generic_readlink(struct dentry *, char __user *, int);
1555extern void generic_fillattr(struct inode *, struct kstat *);
1556extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1557void inode_add_bytes(struct inode *inode, loff_t bytes);
1558void inode_sub_bytes(struct inode *inode, loff_t bytes);
1559loff_t inode_get_bytes(struct inode *inode);
1560void inode_set_bytes(struct inode *inode, loff_t bytes);
1561
1562extern int vfs_readdir(struct file *, filldir_t, void *);
1563
1564extern int vfs_stat(char __user *, struct kstat *);
1565extern int vfs_lstat(char __user *, struct kstat *);
1566extern int vfs_fstat(unsigned int, struct kstat *);
1567
1568extern int vfs_ioctl(struct file *, unsigned int, unsigned int, unsigned long);
1569
1570extern struct file_system_type *get_fs_type(const char *name);
1571extern struct super_block *get_super(struct block_device *);
1572extern struct super_block *user_get_super(dev_t);
1573extern void drop_super(struct super_block *sb);
1574
1575extern int dcache_dir_open(struct inode *, struct file *);
1576extern int dcache_dir_close(struct inode *, struct file *);
1577extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
1578extern int dcache_readdir(struct file *, void *, filldir_t);
1579extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1580extern int simple_statfs(struct super_block *, struct kstatfs *);
1581extern int simple_link(struct dentry *, struct inode *, struct dentry *);
1582extern int simple_unlink(struct inode *, struct dentry *);
1583extern int simple_rmdir(struct inode *, struct dentry *);
1584extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1585extern int simple_sync_file(struct file *, struct dentry *, int);
1586extern int simple_empty(struct dentry *);
1587extern int simple_readpage(struct file *file, struct page *page);
1588extern int simple_prepare_write(struct file *file, struct page *page,
1589 unsigned offset, unsigned to);
1590extern int simple_commit_write(struct file *file, struct page *page,
1591 unsigned offset, unsigned to);
1592
1593extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
1594extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
1595extern struct file_operations simple_dir_operations;
1596extern struct inode_operations simple_dir_inode_operations;
1597struct tree_descr { char *name; struct file_operations *ops; int mode; };
1598struct dentry *d_alloc_name(struct dentry *, const char *);
1599extern int simple_fill_super(struct super_block *, int, struct tree_descr *);
1600extern int simple_pin_fs(char *name, struct vfsmount **mount, int *count);
1601extern void simple_release_fs(struct vfsmount **mount, int *count);
1602
1603extern ssize_t simple_read_from_buffer(void __user *, size_t, loff_t *, const void *, size_t);
1604
1605extern int inode_change_ok(struct inode *, struct iattr *);
1606extern int __must_check inode_setattr(struct inode *, struct iattr *);
1607
1608extern void inode_update_time(struct inode *inode, int ctime_too);
1609
1610static inline ino_t parent_ino(struct dentry *dentry)
1611{
1612 ino_t res;
1613
1614 spin_lock(&dentry->d_lock);
1615 res = dentry->d_parent->d_inode->i_ino;
1616 spin_unlock(&dentry->d_lock);
1617 return res;
1618}
1619
1620/* kernel/fork.c */
1621extern int unshare_files(void);
1622
1623/* Transaction based IO helpers */
1624
1625/*
1626 * An argresp is stored in an allocated page and holds the
1627 * size of the argument or response, along with its content
1628 */
1629struct simple_transaction_argresp {
1630 ssize_t size;
1631 char data[0];
1632};
1633
1634#define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
1635
1636char *simple_transaction_get(struct file *file, const char __user *buf,
1637 size_t size);
1638ssize_t simple_transaction_read(struct file *file, char __user *buf,
1639 size_t size, loff_t *pos);
1640int simple_transaction_release(struct inode *inode, struct file *file);
1641
1642static inline void simple_transaction_set(struct file *file, size_t n)
1643{
1644 struct simple_transaction_argresp *ar = file->private_data;
1645
1646 BUG_ON(n > SIMPLE_TRANSACTION_LIMIT);
1647
1648 /*
1649 * The barrier ensures that ar->size will really remain zero until
1650 * ar->data is ready for reading.
1651 */
1652 smp_mb();
1653 ar->size = n;
1654}
1655
1656#ifdef CONFIG_SECURITY
1657static inline char *alloc_secdata(void)
1658{
1659 return (char *)get_zeroed_page(GFP_KERNEL);
1660}
1661
1662static inline void free_secdata(void *secdata)
1663{
1664 free_page((unsigned long)secdata);
1665}
1666#else
1667static inline char *alloc_secdata(void)
1668{
1669 return (char *)1;
1670}
1671
1672static inline void free_secdata(void *secdata)
1673{ }
1674#endif /* CONFIG_SECURITY */
1675
1676#endif /* __KERNEL__ */
1677#endif /* _LINUX_FS_H */