blob: 128d0082522c4ad067bbffd4199b879ba79c977b [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001#ifndef _LINUX_FS_H
2#define _LINUX_FS_H
3
4/*
5 * This file has definitions for some important file table
6 * structures etc.
7 */
8
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;
Dipankar Sarma529bf6b2006-03-07 21:55:35 -080038extern int get_max_files(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -070039
40struct inodes_stat_t {
41 int nr_inodes;
42 int nr_unused;
43 int dummy[5];
44};
45extern struct inodes_stat_t inodes_stat;
46
47extern int leases_enable, lease_break_time;
48
49#ifdef CONFIG_DNOTIFY
50extern int dir_notify_enable;
51#endif
52
53#define NR_FILE 8192 /* this can well be larger on a larger system */
54
55#define MAY_EXEC 1
56#define MAY_WRITE 2
57#define MAY_READ 4
58#define MAY_APPEND 8
59
60#define FMODE_READ 1
61#define FMODE_WRITE 2
62
63/* Internal kernel extensions */
64#define FMODE_LSEEK 4
65#define FMODE_PREAD 8
66#define FMODE_PWRITE FMODE_PREAD /* These go hand in hand */
67
68#define RW_MASK 1
69#define RWA_MASK 2
70#define READ 0
71#define WRITE 1
72#define READA 2 /* read-ahead - don't block if no resources */
Jan Karaa7662232005-09-06 15:19:10 -070073#define SWRITE 3 /* for ll_rw_block() - wait for buffer lock */
Linus Torvalds1da177e2005-04-16 15:20:36 -070074#define SPECIAL 4 /* For non-blockdevice requests in request queue */
75#define READ_SYNC (READ | (1 << BIO_RW_SYNC))
76#define WRITE_SYNC (WRITE | (1 << BIO_RW_SYNC))
77#define WRITE_BARRIER ((1 << BIO_RW) | (1 << BIO_RW_BARRIER))
78
79#define SEL_IN 1
80#define SEL_OUT 2
81#define SEL_EX 4
82
83/* public flags for file_system_type */
84#define FS_REQUIRES_DEV 1
85#define FS_BINARY_MOUNTDATA 2
86#define FS_REVAL_DOT 16384 /* Check the paths ".", ".." for staleness */
87#define FS_ODD_RENAME 32768 /* Temporary stuff; will go away as soon
88 * as nfs_rename() will be cleaned up
89 */
90/*
91 * These are the fs-independent mount-flags: up to 32 flags are supported
92 */
93#define MS_RDONLY 1 /* Mount read-only */
94#define MS_NOSUID 2 /* Ignore suid and sgid bits */
95#define MS_NODEV 4 /* Disallow access to device special files */
96#define MS_NOEXEC 8 /* Disallow program execution */
97#define MS_SYNCHRONOUS 16 /* Writes are synced at once */
98#define MS_REMOUNT 32 /* Alter flags of a mounted FS */
99#define MS_MANDLOCK 64 /* Allow mandatory locks on an FS */
100#define MS_DIRSYNC 128 /* Directory modifications are synchronous */
101#define MS_NOATIME 1024 /* Do not update access times. */
102#define MS_NODIRATIME 2048 /* Do not update directory access times */
103#define MS_BIND 4096
104#define MS_MOVE 8192
105#define MS_REC 16384
106#define MS_VERBOSE 32768
Miklos Szeredibf066c72006-01-08 01:03:19 -0800107#define MS_POSIXACL (1<<16) /* VFS does not apply the umask */
Ram Pai9676f0c2005-11-07 17:21:20 -0500108#define MS_UNBINDABLE (1<<17) /* change to unbindable */
Ram Pai07b20882005-11-07 17:19:07 -0500109#define MS_PRIVATE (1<<18) /* change to private */
Ram Paia58b0eb2005-11-07 17:20:48 -0500110#define MS_SLAVE (1<<19) /* change to slave */
Ram Pai03e06e62005-11-07 17:19:33 -0500111#define MS_SHARED (1<<20) /* change to shared */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700112#define MS_ACTIVE (1<<30)
113#define MS_NOUSER (1<<31)
114
115/*
116 * Superblock flags that can be altered by MS_REMOUNT
117 */
Christoph Hellwigfc33a7b2006-01-09 20:52:17 -0800118#define MS_RMT_MASK (MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700119
120/*
121 * Old magic mount flag and mask
122 */
123#define MS_MGC_VAL 0xC0ED0000
124#define MS_MGC_MSK 0xffff0000
125
126/* Inode flags - they have nothing to superblock flags now */
127
128#define S_SYNC 1 /* Writes are synced at once */
129#define S_NOATIME 2 /* Do not update access times */
130#define S_APPEND 4 /* Append-only file */
131#define S_IMMUTABLE 8 /* Immutable file */
132#define S_DEAD 16 /* removed, but still open directory */
133#define S_NOQUOTA 32 /* Inode is not counted to quota */
134#define S_DIRSYNC 64 /* Directory modifications are synchronous */
135#define S_NOCMTIME 128 /* Do not update file c/mtime */
136#define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
137#define S_PRIVATE 512 /* Inode is fs-internal */
138
139/*
140 * Note that nosuid etc flags are inode-specific: setting some file-system
141 * flags just means all the inodes inherit those flags by default. It might be
142 * possible to override it selectively if you really wanted to with some
143 * ioctl() that is not currently implemented.
144 *
145 * Exception: MS_RDONLY is always applied to the entire file system.
146 *
147 * Unfortunately, it is possible to change a filesystems flags with it mounted
148 * with files in use. This means that all of the inodes will not have their
149 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
150 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
151 */
152#define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))
153
154#define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
155#define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
156 ((inode)->i_flags & S_SYNC))
157#define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
158 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
159#define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
160
161#define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
162#define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
163#define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700164#define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
165
166#define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
167#define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
168#define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
169#define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
170
171/* the read-only stuff doesn't really belong here, but any other place is
172 probably as bad and I don't want to create yet another include file. */
173
174#define BLKROSET _IO(0x12,93) /* set device read-only (0 = read-write) */
175#define BLKROGET _IO(0x12,94) /* get read-only status (0 = read_write) */
176#define BLKRRPART _IO(0x12,95) /* re-read partition table */
177#define BLKGETSIZE _IO(0x12,96) /* return device size /512 (long *arg) */
178#define BLKFLSBUF _IO(0x12,97) /* flush buffer cache */
179#define BLKRASET _IO(0x12,98) /* set read ahead for block device */
180#define BLKRAGET _IO(0x12,99) /* get current read ahead setting */
181#define BLKFRASET _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
182#define BLKFRAGET _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
183#define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
184#define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
185#define BLKSSZGET _IO(0x12,104)/* get block device sector size */
186#if 0
187#define BLKPG _IO(0x12,105)/* See blkpg.h */
188
189/* Some people are morons. Do not use sizeof! */
190
191#define BLKELVGET _IOR(0x12,106,size_t)/* elevator get */
192#define BLKELVSET _IOW(0x12,107,size_t)/* elevator set */
193/* This was here just to show that the number is taken -
194 probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
195#endif
196/* A jump here: 108-111 have been used for various private purposes. */
197#define BLKBSZGET _IOR(0x12,112,size_t)
198#define BLKBSZSET _IOW(0x12,113,size_t)
199#define BLKGETSIZE64 _IOR(0x12,114,size_t) /* return device size in bytes (u64 *arg) */
200
201#define BMAP_IOCTL 1 /* obsolete - kept for compatibility */
202#define FIBMAP _IO(0x00,1) /* bmap access */
203#define FIGETBSZ _IO(0x00,2) /* get the block size used for bmap */
204
205#ifdef __KERNEL__
206
207#include <linux/linkage.h>
208#include <linux/wait.h>
209#include <linux/types.h>
210#include <linux/kdev_t.h>
211#include <linux/dcache.h>
212#include <linux/stat.h>
213#include <linux/cache.h>
214#include <linux/kobject.h>
215#include <linux/list.h>
216#include <linux/radix-tree.h>
217#include <linux/prio_tree.h>
218#include <linux/init.h>
Jens Axboe22e2c502005-06-27 10:55:12 +0200219#include <linux/sched.h>
Jes Sorensen1b1dcc12006-01-09 15:59:24 -0800220#include <linux/mutex.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700221
222#include <asm/atomic.h>
223#include <asm/semaphore.h>
224#include <asm/byteorder.h>
225
Christoph Hellwiga885c8c2006-01-08 01:02:50 -0800226struct hd_geometry;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700227struct iovec;
228struct nameidata;
Christoph Hellwig92198f72005-06-23 22:00:59 -0700229struct kiocb;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700230struct pipe_inode_info;
231struct poll_table_struct;
232struct kstatfs;
233struct vm_area_struct;
234struct vfsmount;
235
Linus Torvalds1da177e2005-04-16 15:20:36 -0700236extern void __init inode_init(unsigned long);
237extern void __init inode_init_early(void);
238extern void __init mnt_init(unsigned long);
239extern void __init files_init(unsigned long);
240
241struct buffer_head;
242typedef int (get_block_t)(struct inode *inode, sector_t iblock,
243 struct buffer_head *bh_result, int create);
244typedef int (get_blocks_t)(struct inode *inode, sector_t iblock,
245 unsigned long max_blocks,
246 struct buffer_head *bh_result, int create);
Christoph Hellwig92198f72005-06-23 22:00:59 -0700247typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700248 ssize_t bytes, void *private);
249
250/*
251 * Attribute flags. These should be or-ed together to figure out what
252 * has been changed!
253 */
254#define ATTR_MODE 1
255#define ATTR_UID 2
256#define ATTR_GID 4
257#define ATTR_SIZE 8
258#define ATTR_ATIME 16
259#define ATTR_MTIME 32
260#define ATTR_CTIME 64
261#define ATTR_ATIME_SET 128
262#define ATTR_MTIME_SET 256
263#define ATTR_FORCE 512 /* Not a change, but a change it */
264#define ATTR_ATTR_FLAG 1024
265#define ATTR_KILL_SUID 2048
266#define ATTR_KILL_SGID 4096
Miklos Szeredicc4e69d2005-11-07 00:59:49 -0800267#define ATTR_FILE 8192
Linus Torvalds1da177e2005-04-16 15:20:36 -0700268
269/*
270 * This is the Inode Attributes structure, used for notify_change(). It
271 * uses the above definitions as flags, to know which values have changed.
272 * Also, in this manner, a Filesystem can look at only the values it cares
273 * about. Basically, these are the attributes that the VFS layer can
274 * request to change from the FS layer.
275 *
276 * Derek Atkins <warlord@MIT.EDU> 94-10-20
277 */
278struct iattr {
279 unsigned int ia_valid;
280 umode_t ia_mode;
281 uid_t ia_uid;
282 gid_t ia_gid;
283 loff_t ia_size;
284 struct timespec ia_atime;
285 struct timespec ia_mtime;
286 struct timespec ia_ctime;
Miklos Szeredicc4e69d2005-11-07 00:59:49 -0800287
288 /*
289 * Not an attribute, but an auxilary info for filesystems wanting to
290 * implement an ftruncate() like method. NOTE: filesystem should
291 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
292 */
293 struct file *ia_file;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700294};
295
296/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700297 * Includes for diskquotas.
298 */
299#include <linux/quota.h>
300
Zach Brown994fc28c2005-12-15 14:28:17 -0800301/**
302 * enum positive_aop_returns - aop return codes with specific semantics
303 *
304 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
305 * completed, that the page is still locked, and
306 * should be considered active. The VM uses this hint
307 * to return the page to the active list -- it won't
308 * be a candidate for writeback again in the near
309 * future. Other callers must be careful to unlock
310 * the page if they get this return. Returned by
311 * writepage();
312 *
313 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
314 * unlocked it and the page might have been truncated.
315 * The caller should back up to acquiring a new page and
316 * trying again. The aop will be taking reasonable
317 * precautions not to livelock. If the caller held a page
318 * reference, it should drop it before retrying. Returned
319 * by readpage(), prepare_write(), and commit_write().
320 *
321 * address_space_operation functions return these large constants to indicate
322 * special semantics to the caller. These are much larger than the bytes in a
323 * page to allow for functions that return the number of bytes operated on in a
324 * given page.
325 */
326
327enum positive_aop_returns {
328 AOP_WRITEPAGE_ACTIVATE = 0x80000,
329 AOP_TRUNCATED_PAGE = 0x80001,
330};
331
Linus Torvalds1da177e2005-04-16 15:20:36 -0700332/*
333 * oh the beauties of C type declarations.
334 */
335struct page;
336struct address_space;
337struct writeback_control;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700338
339struct address_space_operations {
340 int (*writepage)(struct page *page, struct writeback_control *wbc);
341 int (*readpage)(struct file *, struct page *);
342 int (*sync_page)(struct page *);
343
344 /* Write back some dirty pages from this mapping. */
345 int (*writepages)(struct address_space *, struct writeback_control *);
346
347 /* Set a page dirty */
348 int (*set_page_dirty)(struct page *page);
349
350 int (*readpages)(struct file *filp, struct address_space *mapping,
351 struct list_head *pages, unsigned nr_pages);
352
353 /*
354 * ext3 requires that a successful prepare_write() call be followed
355 * by a commit_write() call - they must be balanced
356 */
357 int (*prepare_write)(struct file *, struct page *, unsigned, unsigned);
358 int (*commit_write)(struct file *, struct page *, unsigned, unsigned);
359 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
360 sector_t (*bmap)(struct address_space *, sector_t);
361 int (*invalidatepage) (struct page *, unsigned long);
Al Viro27496a82005-10-21 03:20:48 -0400362 int (*releasepage) (struct page *, gfp_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700363 ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
364 loff_t offset, unsigned long nr_segs);
Carsten Otteceffc072005-06-23 22:05:25 -0700365 struct page* (*get_xip_page)(struct address_space *, sector_t,
366 int);
Christoph Lametere965f962006-02-01 03:05:41 -0800367 /* migrate the contents of a page to the specified target */
368 int (*migratepage) (struct page *, struct page *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700369};
370
371struct backing_dev_info;
372struct address_space {
373 struct inode *host; /* owner: inode, block_device */
374 struct radix_tree_root page_tree; /* radix tree of all pages */
375 rwlock_t tree_lock; /* and rwlock protecting it */
376 unsigned int i_mmap_writable;/* count VM_SHARED mappings */
377 struct prio_tree_root i_mmap; /* tree of private and shared mappings */
378 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
379 spinlock_t i_mmap_lock; /* protect tree, count, list */
380 unsigned int truncate_count; /* Cover race condition with truncate */
381 unsigned long nrpages; /* number of total pages */
382 pgoff_t writeback_index;/* writeback starts here */
383 struct address_space_operations *a_ops; /* methods */
384 unsigned long flags; /* error bits/gfp mask */
385 struct backing_dev_info *backing_dev_info; /* device readahead, etc */
386 spinlock_t private_lock; /* for use by the address_space */
387 struct list_head private_list; /* ditto */
388 struct address_space *assoc_mapping; /* ditto */
389} __attribute__((aligned(sizeof(long))));
390 /*
391 * On most architectures that alignment is already the case; but
392 * must be enforced here for CRIS, to let the least signficant bit
393 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
394 */
395
396struct block_device {
397 dev_t bd_dev; /* not a kdev_t - it's a search key */
398 struct inode * bd_inode; /* will die */
399 int bd_openers;
400 struct semaphore bd_sem; /* open/close mutex */
401 struct semaphore bd_mount_sem; /* mount mutex */
402 struct list_head bd_inodes;
403 void * bd_holder;
404 int bd_holders;
405 struct block_device * bd_contains;
406 unsigned bd_block_size;
407 struct hd_struct * bd_part;
408 /* number of times partitions within this device have been opened. */
409 unsigned bd_part_count;
410 int bd_invalidated;
411 struct gendisk * bd_disk;
412 struct list_head bd_list;
413 struct backing_dev_info *bd_inode_backing_dev_info;
414 /*
415 * Private data. You must have bd_claim'ed the block_device
416 * to use this. NOTE: bd_claim allows an owner to claim
417 * the same device multiple times, the owner must take special
418 * care to not mess up bd_private for that case.
419 */
420 unsigned long bd_private;
421};
422
423/*
424 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
425 * radix trees
426 */
427#define PAGECACHE_TAG_DIRTY 0
428#define PAGECACHE_TAG_WRITEBACK 1
429
430int mapping_tagged(struct address_space *mapping, int tag);
431
432/*
433 * Might pages of this file be mapped into userspace?
434 */
435static inline int mapping_mapped(struct address_space *mapping)
436{
437 return !prio_tree_empty(&mapping->i_mmap) ||
438 !list_empty(&mapping->i_mmap_nonlinear);
439}
440
441/*
442 * Might pages of this file have been modified in userspace?
443 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
444 * marks vma as VM_SHARED if it is shared, and the file was opened for
445 * writing i.e. vma may be mprotected writable even if now readonly.
446 */
447static inline int mapping_writably_mapped(struct address_space *mapping)
448{
449 return mapping->i_mmap_writable != 0;
450}
451
452/*
453 * Use sequence counter to get consistent i_size on 32-bit processors.
454 */
455#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
456#include <linux/seqlock.h>
457#define __NEED_I_SIZE_ORDERED
458#define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
459#else
460#define i_size_ordered_init(inode) do { } while (0)
461#endif
462
463struct inode {
464 struct hlist_node i_hash;
465 struct list_head i_list;
466 struct list_head i_sb_list;
467 struct list_head i_dentry;
468 unsigned long i_ino;
469 atomic_t i_count;
470 umode_t i_mode;
471 unsigned int i_nlink;
472 uid_t i_uid;
473 gid_t i_gid;
474 dev_t i_rdev;
475 loff_t i_size;
476 struct timespec i_atime;
477 struct timespec i_mtime;
478 struct timespec i_ctime;
479 unsigned int i_blkbits;
480 unsigned long i_blksize;
481 unsigned long i_version;
482 unsigned long i_blocks;
483 unsigned short i_bytes;
484 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
Jes Sorensen1b1dcc12006-01-09 15:59:24 -0800485 struct mutex i_mutex;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700486 struct rw_semaphore i_alloc_sem;
487 struct inode_operations *i_op;
488 struct file_operations *i_fop; /* former ->i_op->default_file_ops */
489 struct super_block *i_sb;
490 struct file_lock *i_flock;
491 struct address_space *i_mapping;
492 struct address_space i_data;
493#ifdef CONFIG_QUOTA
494 struct dquot *i_dquot[MAXQUOTAS];
495#endif
496 /* These three should probably be a union */
497 struct list_head i_devices;
498 struct pipe_inode_info *i_pipe;
499 struct block_device *i_bdev;
500 struct cdev *i_cdev;
501 int i_cindex;
502
503 __u32 i_generation;
504
505#ifdef CONFIG_DNOTIFY
506 unsigned long i_dnotify_mask; /* Directory notify events */
507 struct dnotify_struct *i_dnotify; /* for directory notifications */
508#endif
509
Robert Love0eeca282005-07-12 17:06:03 -0400510#ifdef CONFIG_INOTIFY
511 struct list_head inotify_watches; /* watches on this inode */
512 struct semaphore inotify_sem; /* protects the watches list */
513#endif
514
Linus Torvalds1da177e2005-04-16 15:20:36 -0700515 unsigned long i_state;
516 unsigned long dirtied_when; /* jiffies of first dirtying */
517
518 unsigned int i_flags;
519
520 atomic_t i_writecount;
521 void *i_security;
522 union {
523 void *generic_ip;
524 } u;
525#ifdef __NEED_I_SIZE_ORDERED
526 seqcount_t i_size_seqcount;
527#endif
528};
529
530/*
531 * NOTE: in a 32bit arch with a preemptable kernel and
532 * an UP compile the i_size_read/write must be atomic
533 * with respect to the local cpu (unlike with preempt disabled),
534 * but they don't need to be atomic with respect to other cpus like in
535 * true SMP (so they need either to either locally disable irq around
536 * the read or for example on x86 they can be still implemented as a
537 * cmpxchg8b without the need of the lock prefix). For SMP compiles
538 * and 64bit archs it makes no difference if preempt is enabled or not.
539 */
540static inline loff_t i_size_read(struct inode *inode)
541{
542#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
543 loff_t i_size;
544 unsigned int seq;
545
546 do {
547 seq = read_seqcount_begin(&inode->i_size_seqcount);
548 i_size = inode->i_size;
549 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
550 return i_size;
551#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
552 loff_t i_size;
553
554 preempt_disable();
555 i_size = inode->i_size;
556 preempt_enable();
557 return i_size;
558#else
559 return inode->i_size;
560#endif
561}
562
563
564static inline void i_size_write(struct inode *inode, loff_t i_size)
565{
566#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
567 write_seqcount_begin(&inode->i_size_seqcount);
568 inode->i_size = i_size;
569 write_seqcount_end(&inode->i_size_seqcount);
570#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
571 preempt_disable();
572 inode->i_size = i_size;
573 preempt_enable();
574#else
575 inode->i_size = i_size;
576#endif
577}
578
579static inline unsigned iminor(struct inode *inode)
580{
581 return MINOR(inode->i_rdev);
582}
583
584static inline unsigned imajor(struct inode *inode)
585{
586 return MAJOR(inode->i_rdev);
587}
588
589extern struct block_device *I_BDEV(struct inode *inode);
590
591struct fown_struct {
592 rwlock_t lock; /* protects pid, uid, euid fields */
593 int pid; /* pid or -pgrp where SIGIO should be sent */
594 uid_t uid, euid; /* uid/euid of process setting the owner */
595 void *security;
596 int signum; /* posix.1b rt signal to be delivered on IO */
597};
598
599/*
600 * Track a single file's readahead state
601 */
602struct file_ra_state {
603 unsigned long start; /* Current window */
604 unsigned long size;
605 unsigned long flags; /* ra flags RA_FLAG_xxx*/
606 unsigned long cache_hit; /* cache hit count*/
607 unsigned long prev_page; /* Cache last read() position */
608 unsigned long ahead_start; /* Ahead window */
609 unsigned long ahead_size;
610 unsigned long ra_pages; /* Maximum readahead window */
611 unsigned long mmap_hit; /* Cache hit stat for mmap accesses */
612 unsigned long mmap_miss; /* Cache miss stat for mmap accesses */
613};
614#define RA_FLAG_MISS 0x01 /* a cache miss occured against this file */
615#define RA_FLAG_INCACHE 0x02 /* file is already in cache */
616
617struct file {
Eric Dumazet2f512012005-10-30 15:02:16 -0800618 /*
619 * fu_list becomes invalid after file_free is called and queued via
620 * fu_rcuhead for RCU freeing
621 */
622 union {
623 struct list_head fu_list;
624 struct rcu_head fu_rcuhead;
625 } f_u;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700626 struct dentry *f_dentry;
627 struct vfsmount *f_vfsmnt;
628 struct file_operations *f_op;
629 atomic_t f_count;
630 unsigned int f_flags;
631 mode_t f_mode;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700632 loff_t f_pos;
633 struct fown_struct f_owner;
634 unsigned int f_uid, f_gid;
635 struct file_ra_state f_ra;
636
Linus Torvalds1da177e2005-04-16 15:20:36 -0700637 unsigned long f_version;
638 void *f_security;
639
640 /* needed for tty driver, and maybe others */
641 void *private_data;
642
643#ifdef CONFIG_EPOLL
644 /* Used by fs/eventpoll.c to link all the hooks to this file */
645 struct list_head f_ep_links;
646 spinlock_t f_ep_lock;
647#endif /* #ifdef CONFIG_EPOLL */
648 struct address_space *f_mapping;
649};
650extern spinlock_t files_lock;
651#define file_list_lock() spin_lock(&files_lock);
652#define file_list_unlock() spin_unlock(&files_lock);
653
Nick Piggin095975d2006-01-08 01:02:19 -0800654#define get_file(x) atomic_inc(&(x)->f_count)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700655#define file_count(x) atomic_read(&(x)->f_count)
656
657#define MAX_NON_LFS ((1UL<<31) - 1)
658
659/* Page cache limit. The filesystems should put that into their s_maxbytes
660 limits, otherwise bad things can happen in VM. */
661#if BITS_PER_LONG==32
662#define MAX_LFS_FILESIZE (((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
663#elif BITS_PER_LONG==64
664#define MAX_LFS_FILESIZE 0x7fffffffffffffffUL
665#endif
666
667#define FL_POSIX 1
668#define FL_FLOCK 2
669#define FL_ACCESS 8 /* not trying to lock, just looking */
670#define FL_LOCKD 16 /* lock held by rpc.lockd */
671#define FL_LEASE 32 /* lease held on this file */
672#define FL_SLEEP 128 /* A blocking lock */
673
674/*
675 * The POSIX file lock owner is determined by
676 * the "struct files_struct" in the thread group
677 * (or NULL for no owner - BSD locks).
678 *
679 * Lockd stuffs a "host" pointer into this.
680 */
681typedef struct files_struct *fl_owner_t;
682
683struct file_lock_operations {
684 void (*fl_insert)(struct file_lock *); /* lock insertion callback */
685 void (*fl_remove)(struct file_lock *); /* lock removal callback */
686 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
687 void (*fl_release_private)(struct file_lock *);
688};
689
690struct lock_manager_operations {
691 int (*fl_compare_owner)(struct file_lock *, struct file_lock *);
692 void (*fl_notify)(struct file_lock *); /* unblock callback */
693 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
694 void (*fl_release_private)(struct file_lock *);
695 void (*fl_break)(struct file_lock *);
696 int (*fl_mylease)(struct file_lock *, struct file_lock *);
697 int (*fl_change)(struct file_lock **, int);
698};
699
700/* that will die - we need it for nfs_lock_info */
701#include <linux/nfs_fs_i.h>
702
703struct file_lock {
704 struct file_lock *fl_next; /* singly linked list for this inode */
705 struct list_head fl_link; /* doubly linked list of all locks */
706 struct list_head fl_block; /* circular list of blocked processes */
707 fl_owner_t fl_owner;
708 unsigned int fl_pid;
709 wait_queue_head_t fl_wait;
710 struct file *fl_file;
711 unsigned char fl_flags;
712 unsigned char fl_type;
713 loff_t fl_start;
714 loff_t fl_end;
715
716 struct fasync_struct * fl_fasync; /* for lease break notifications */
717 unsigned long fl_break_time; /* for nonblocking lease breaks */
718
719 struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
720 struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
721 union {
722 struct nfs_lock_info nfs_fl;
Trond Myklebust8d0a8a92005-06-22 17:16:32 +0000723 struct nfs4_lock_info nfs4_fl;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700724 } fl_u;
725};
726
727/* The following constant reflects the upper bound of the file/locking space */
728#ifndef OFFSET_MAX
729#define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
730#define OFFSET_MAX INT_LIMIT(loff_t)
731#define OFFT_OFFSET_MAX INT_LIMIT(off_t)
732#endif
733
734extern struct list_head file_lock_list;
735
736#include <linux/fcntl.h>
737
738extern int fcntl_getlk(struct file *, struct flock __user *);
Peter Staubachc2936212005-07-27 11:45:09 -0700739extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
740 struct flock __user *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700741
742#if BITS_PER_LONG == 32
743extern int fcntl_getlk64(struct file *, struct flock64 __user *);
Peter Staubachc2936212005-07-27 11:45:09 -0700744extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
745 struct flock64 __user *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700746#endif
747
748extern void send_sigio(struct fown_struct *fown, int fd, int band);
749extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
750extern int fcntl_getlease(struct file *filp);
751
752/* fs/locks.c */
753extern void locks_init_lock(struct file_lock *);
754extern void locks_copy_lock(struct file_lock *, struct file_lock *);
755extern void locks_remove_posix(struct file *, fl_owner_t);
756extern void locks_remove_flock(struct file *);
757extern struct file_lock *posix_test_lock(struct file *, struct file_lock *);
758extern int posix_lock_file(struct file *, struct file_lock *);
759extern int posix_lock_file_wait(struct file *, struct file_lock *);
760extern void posix_block_lock(struct file_lock *, struct file_lock *);
J. Bruce Fields64a318e2006-01-03 09:55:46 +0100761extern int posix_unblock_lock(struct file *, struct file_lock *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700762extern int posix_locks_deadlock(struct file_lock *, struct file_lock *);
763extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
764extern int __break_lease(struct inode *inode, unsigned int flags);
765extern void lease_get_mtime(struct inode *, struct timespec *time);
766extern int setlease(struct file *, long, struct file_lock **);
767extern int lease_modify(struct file_lock **, int);
768extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
769extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
770extern void steal_locks(fl_owner_t from);
771
772struct fasync_struct {
773 int magic;
774 int fa_fd;
775 struct fasync_struct *fa_next; /* singly linked list */
776 struct file *fa_file;
777};
778
779#define FASYNC_MAGIC 0x4601
780
781/* SMP safe fasync helpers: */
782extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
783/* can be called from interrupts */
784extern void kill_fasync(struct fasync_struct **, int, int);
785/* only for net: no internal synchronization */
786extern void __kill_fasync(struct fasync_struct *, int, int);
787
788extern int f_setown(struct file *filp, unsigned long arg, int force);
789extern void f_delown(struct file *filp);
790extern int send_sigurg(struct fown_struct *fown);
791
792/*
793 * Umount options
794 */
795
796#define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
797#define MNT_DETACH 0x00000002 /* Just detach from the tree */
798#define MNT_EXPIRE 0x00000004 /* Mark for expiry */
799
800extern struct list_head super_blocks;
801extern spinlock_t sb_lock;
802
803#define sb_entry(list) list_entry((list), struct super_block, s_list)
804#define S_BIAS (1<<30)
805struct super_block {
806 struct list_head s_list; /* Keep this first */
807 dev_t s_dev; /* search index; _not_ kdev_t */
808 unsigned long s_blocksize;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700809 unsigned char s_blocksize_bits;
810 unsigned char s_dirt;
811 unsigned long long s_maxbytes; /* Max file size */
812 struct file_system_type *s_type;
813 struct super_operations *s_op;
814 struct dquot_operations *dq_op;
815 struct quotactl_ops *s_qcop;
816 struct export_operations *s_export_op;
817 unsigned long s_flags;
818 unsigned long s_magic;
819 struct dentry *s_root;
820 struct rw_semaphore s_umount;
Ingo Molnar7892f2f2006-01-09 15:59:25 -0800821 struct mutex s_lock;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700822 int s_count;
823 int s_syncing;
824 int s_need_sync_fs;
825 atomic_t s_active;
826 void *s_security;
827 struct xattr_handler **s_xattr;
828
829 struct list_head s_inodes; /* all inodes */
830 struct list_head s_dirty; /* dirty inodes */
831 struct list_head s_io; /* parked for writeback */
832 struct hlist_head s_anon; /* anonymous dentries for (nfs) exporting */
833 struct list_head s_files;
834
835 struct block_device *s_bdev;
836 struct list_head s_instances;
837 struct quota_info s_dquot; /* Diskquota specific options */
838
839 int s_frozen;
840 wait_queue_head_t s_wait_unfrozen;
841
842 char s_id[32]; /* Informational name */
843
844 void *s_fs_info; /* Filesystem private info */
845
846 /*
847 * The next field is for VFS *only*. No filesystems have any business
848 * even looking at it. You had been warned.
849 */
850 struct semaphore s_vfs_rename_sem; /* Kludge */
851
852 /* Granuality of c/m/atime in ns.
853 Cannot be worse than a second */
854 u32 s_time_gran;
855};
856
857extern struct timespec current_fs_time(struct super_block *sb);
858
859/*
860 * Snapshotting support.
861 */
862enum {
863 SB_UNFROZEN = 0,
864 SB_FREEZE_WRITE = 1,
865 SB_FREEZE_TRANS = 2,
866};
867
868#define vfs_check_frozen(sb, level) \
869 wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
870
Jens Axboe22e2c502005-06-27 10:55:12 +0200871static inline void get_fs_excl(void)
872{
873 atomic_inc(&current->fs_excl);
874}
875
876static inline void put_fs_excl(void)
877{
878 atomic_dec(&current->fs_excl);
879}
880
881static inline int has_fs_excl(void)
882{
883 return atomic_read(&current->fs_excl);
884}
885
886
Linus Torvalds1da177e2005-04-16 15:20:36 -0700887/*
888 * Superblock locking.
889 */
890static inline void lock_super(struct super_block * sb)
891{
Jens Axboe22e2c502005-06-27 10:55:12 +0200892 get_fs_excl();
Ingo Molnar7892f2f2006-01-09 15:59:25 -0800893 mutex_lock(&sb->s_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700894}
895
896static inline void unlock_super(struct super_block * sb)
897{
Jens Axboe22e2c502005-06-27 10:55:12 +0200898 put_fs_excl();
Ingo Molnar7892f2f2006-01-09 15:59:25 -0800899 mutex_unlock(&sb->s_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700900}
901
902/*
903 * VFS helper functions..
904 */
Christoph Hellwige4543ed2005-11-08 21:35:04 -0800905extern int vfs_permission(struct nameidata *, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700906extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *);
907extern int vfs_mkdir(struct inode *, struct dentry *, int);
908extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t);
909extern int vfs_symlink(struct inode *, struct dentry *, const char *, int);
910extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
911extern int vfs_rmdir(struct inode *, struct dentry *);
912extern int vfs_unlink(struct inode *, struct dentry *);
913extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
914
915/*
916 * VFS dentry helper functions.
917 */
918extern void dentry_unhash(struct dentry *dentry);
919
920/*
Christoph Hellwig8c744fb2005-11-08 21:35:04 -0800921 * VFS file helper functions.
922 */
923extern int file_permission(struct file *, int);
924
925/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700926 * File types
927 *
928 * NOTE! These match bits 12..15 of stat.st_mode
929 * (ie "(i_mode >> 12) & 15").
930 */
931#define DT_UNKNOWN 0
932#define DT_FIFO 1
933#define DT_CHR 2
934#define DT_DIR 4
935#define DT_BLK 6
936#define DT_REG 8
937#define DT_LNK 10
938#define DT_SOCK 12
939#define DT_WHT 14
940
941#define OSYNC_METADATA (1<<0)
942#define OSYNC_DATA (1<<1)
943#define OSYNC_INODE (1<<2)
944int generic_osync_inode(struct inode *, struct address_space *, int);
945
946/*
947 * This is the "filldir" function type, used by readdir() to let
948 * the kernel specify what kind of dirent layout it wants to have.
949 * This allows the kernel to read directories into kernel space or
950 * to have different dirent layouts depending on the binary type.
951 */
952typedef int (*filldir_t)(void *, const char *, int, loff_t, ino_t, unsigned);
953
954struct block_device_operations {
955 int (*open) (struct inode *, struct file *);
956 int (*release) (struct inode *, struct file *);
957 int (*ioctl) (struct inode *, struct file *, unsigned, unsigned long);
Arnd Bergmannbb93e3a2005-06-23 00:10:15 -0700958 long (*unlocked_ioctl) (struct file *, unsigned, unsigned long);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700959 long (*compat_ioctl) (struct file *, unsigned, unsigned long);
Carsten Otte420edbc2005-06-23 22:05:23 -0700960 int (*direct_access) (struct block_device *, sector_t, unsigned long *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700961 int (*media_changed) (struct gendisk *);
962 int (*revalidate_disk) (struct gendisk *);
Christoph Hellwiga885c8c2006-01-08 01:02:50 -0800963 int (*getgeo)(struct block_device *, struct hd_geometry *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700964 struct module *owner;
965};
966
967/*
968 * "descriptor" for what we're up to with a read for sendfile().
969 * This allows us to use the same read code yet
970 * have multiple different users of the data that
971 * we read from a file.
972 *
973 * The simplest case just copies the data to user
974 * mode.
975 */
976typedef struct {
977 size_t written;
978 size_t count;
979 union {
980 char __user * buf;
981 void *data;
982 } arg;
983 int error;
984} read_descriptor_t;
985
986typedef int (*read_actor_t)(read_descriptor_t *, struct page *, unsigned long, unsigned long);
987
988/* These macros are for out of kernel modules to test that
989 * the kernel supports the unlocked_ioctl and compat_ioctl
990 * fields in struct file_operations. */
991#define HAVE_COMPAT_IOCTL 1
992#define HAVE_UNLOCKED_IOCTL 1
993
994/*
995 * NOTE:
996 * read, write, poll, fsync, readv, writev, unlocked_ioctl and compat_ioctl
997 * can be called without the big kernel lock held in all filesystems.
998 */
999struct file_operations {
1000 struct module *owner;
1001 loff_t (*llseek) (struct file *, loff_t, int);
1002 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1003 ssize_t (*aio_read) (struct kiocb *, char __user *, size_t, loff_t);
1004 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1005 ssize_t (*aio_write) (struct kiocb *, const char __user *, size_t, loff_t);
1006 int (*readdir) (struct file *, void *, filldir_t);
1007 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1008 int (*ioctl) (struct inode *, struct file *, unsigned int, unsigned long);
1009 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1010 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1011 int (*mmap) (struct file *, struct vm_area_struct *);
1012 int (*open) (struct inode *, struct file *);
1013 int (*flush) (struct file *);
1014 int (*release) (struct inode *, struct file *);
1015 int (*fsync) (struct file *, struct dentry *, int datasync);
1016 int (*aio_fsync) (struct kiocb *, int datasync);
1017 int (*fasync) (int, struct file *, int);
1018 int (*lock) (struct file *, int, struct file_lock *);
1019 ssize_t (*readv) (struct file *, const struct iovec *, unsigned long, loff_t *);
1020 ssize_t (*writev) (struct file *, const struct iovec *, unsigned long, loff_t *);
1021 ssize_t (*sendfile) (struct file *, loff_t *, size_t, read_actor_t, void *);
1022 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1023 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1024 int (*check_flags)(int);
1025 int (*dir_notify)(struct file *filp, unsigned long arg);
1026 int (*flock) (struct file *, int, struct file_lock *);
1027};
1028
1029struct inode_operations {
1030 int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
1031 struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
1032 int (*link) (struct dentry *,struct inode *,struct dentry *);
1033 int (*unlink) (struct inode *,struct dentry *);
1034 int (*symlink) (struct inode *,struct dentry *,const char *);
1035 int (*mkdir) (struct inode *,struct dentry *,int);
1036 int (*rmdir) (struct inode *,struct dentry *);
1037 int (*mknod) (struct inode *,struct dentry *,int,dev_t);
1038 int (*rename) (struct inode *, struct dentry *,
1039 struct inode *, struct dentry *);
1040 int (*readlink) (struct dentry *, char __user *,int);
Linus Torvaldscc314ee2005-08-19 18:02:56 -07001041 void * (*follow_link) (struct dentry *, struct nameidata *);
1042 void (*put_link) (struct dentry *, struct nameidata *, void *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001043 void (*truncate) (struct inode *);
1044 int (*permission) (struct inode *, int, struct nameidata *);
1045 int (*setattr) (struct dentry *, struct iattr *);
1046 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1047 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1048 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1049 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1050 int (*removexattr) (struct dentry *, const char *);
Badari Pulavartyf6b3ec22006-01-06 00:10:38 -08001051 void (*truncate_range)(struct inode *, loff_t, loff_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001052};
1053
1054struct seq_file;
1055
1056extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1057extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1058extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1059 unsigned long, loff_t *);
1060extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1061 unsigned long, loff_t *);
1062
1063/*
1064 * NOTE: write_inode, delete_inode, clear_inode, put_inode can be called
1065 * without the big kernel lock held in all filesystems.
1066 */
1067struct super_operations {
1068 struct inode *(*alloc_inode)(struct super_block *sb);
1069 void (*destroy_inode)(struct inode *);
1070
1071 void (*read_inode) (struct inode *);
1072
1073 void (*dirty_inode) (struct inode *);
1074 int (*write_inode) (struct inode *, int);
1075 void (*put_inode) (struct inode *);
1076 void (*drop_inode) (struct inode *);
1077 void (*delete_inode) (struct inode *);
1078 void (*put_super) (struct super_block *);
1079 void (*write_super) (struct super_block *);
1080 int (*sync_fs)(struct super_block *sb, int wait);
1081 void (*write_super_lockfs) (struct super_block *);
1082 void (*unlockfs) (struct super_block *);
1083 int (*statfs) (struct super_block *, struct kstatfs *);
1084 int (*remount_fs) (struct super_block *, int *, char *);
1085 void (*clear_inode) (struct inode *);
1086 void (*umount_begin) (struct super_block *);
1087
1088 int (*show_options)(struct seq_file *, struct vfsmount *);
1089
1090 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1091 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1092};
1093
1094/* Inode state bits. Protected by inode_lock. */
1095#define I_DIRTY_SYNC 1 /* Not dirty enough for O_DATASYNC */
1096#define I_DIRTY_DATASYNC 2 /* Data-related inode changes pending */
1097#define I_DIRTY_PAGES 4 /* Data-related inode changes pending */
1098#define __I_LOCK 3
1099#define I_LOCK (1 << __I_LOCK)
1100#define I_FREEING 16
1101#define I_CLEAR 32
1102#define I_NEW 64
Alexander Viro991114c2005-06-23 00:09:01 -07001103#define I_WILL_FREE 128
Linus Torvalds1da177e2005-04-16 15:20:36 -07001104
1105#define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1106
1107extern void __mark_inode_dirty(struct inode *, int);
1108static inline void mark_inode_dirty(struct inode *inode)
1109{
1110 __mark_inode_dirty(inode, I_DIRTY);
1111}
1112
1113static inline void mark_inode_dirty_sync(struct inode *inode)
1114{
1115 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1116}
1117
Christoph Hellwig869243a2006-01-09 20:52:03 -08001118extern void touch_atime(struct vfsmount *mnt, struct dentry *dentry);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001119static inline void file_accessed(struct file *file)
1120{
1121 if (!(file->f_flags & O_NOATIME))
1122 touch_atime(file->f_vfsmnt, file->f_dentry);
1123}
1124
1125int sync_inode(struct inode *inode, struct writeback_control *wbc);
1126
1127/**
Pavel Pisa4dc3b162005-05-01 08:59:25 -07001128 * struct export_operations - for nfsd to communicate with file systems
1129 * @decode_fh: decode a file handle fragment and return a &struct dentry
1130 * @encode_fh: encode a file handle fragment from a dentry
1131 * @get_name: find the name for a given inode in a given directory
1132 * @get_parent: find the parent of a given directory
1133 * @get_dentry: find a dentry for the inode given a file handle sub-fragment
Randy Dunlap8f2709b2005-11-07 01:01:05 -08001134 * @find_exported_dentry:
1135 * set by the exporting module to a standard helper function.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001136 *
1137 * Description:
1138 * The export_operations structure provides a means for nfsd to communicate
1139 * with a particular exported file system - particularly enabling nfsd and
1140 * the filesystem to co-operate when dealing with file handles.
1141 *
Martin Waitz67be2dd2005-05-01 08:59:26 -07001142 * export_operations contains two basic operation for dealing with file
1143 * handles, decode_fh() and encode_fh(), and allows for some other
1144 * operations to be defined which standard helper routines use to get
1145 * specific information from the filesystem.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001146 *
1147 * nfsd encodes information use to determine which filesystem a filehandle
Martin Waitz67be2dd2005-05-01 08:59:26 -07001148 * applies to in the initial part of the file handle. The remainder, termed
1149 * a file handle fragment, is controlled completely by the filesystem. The
1150 * standard helper routines assume that this fragment will contain one or
1151 * two sub-fragments, one which identifies the file, and one which may be
1152 * used to identify the (a) directory containing the file.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001153 *
1154 * In some situations, nfsd needs to get a dentry which is connected into a
Martin Waitz67be2dd2005-05-01 08:59:26 -07001155 * specific part of the file tree. To allow for this, it passes the
1156 * function acceptable() together with a @context which can be used to see
1157 * if the dentry is acceptable. As there can be multiple dentrys for a
1158 * given file, the filesystem should check each one for acceptability before
1159 * looking for the next. As soon as an acceptable one is found, it should
1160 * be returned.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001161 *
1162 * decode_fh:
Martin Waitz67be2dd2005-05-01 08:59:26 -07001163 * @decode_fh is given a &struct super_block (@sb), a file handle fragment
1164 * (@fh, @fh_len) and an acceptability testing function (@acceptable,
1165 * @context). It should return a &struct dentry which refers to the same
1166 * file that the file handle fragment refers to, and which passes the
1167 * acceptability test. If it cannot, it should return a %NULL pointer if
1168 * the file was found but no acceptable &dentries were available, or a
1169 * %ERR_PTR error code indicating why it couldn't be found (e.g. %ENOENT or
1170 * %ENOMEM).
Linus Torvalds1da177e2005-04-16 15:20:36 -07001171 *
1172 * encode_fh:
Martin Waitz67be2dd2005-05-01 08:59:26 -07001173 * @encode_fh should store in the file handle fragment @fh (using at most
1174 * @max_len bytes) information that can be used by @decode_fh to recover the
1175 * file refered to by the &struct dentry @de. If the @connectable flag is
1176 * set, the encode_fh() should store sufficient information so that a good
1177 * attempt can be made to find not only the file but also it's place in the
1178 * filesystem. This typically means storing a reference to de->d_parent in
1179 * the filehandle fragment. encode_fh() should return the number of bytes
1180 * stored or a negative error code such as %-ENOSPC
Linus Torvalds1da177e2005-04-16 15:20:36 -07001181 *
1182 * get_name:
Martin Waitz67be2dd2005-05-01 08:59:26 -07001183 * @get_name should find a name for the given @child in the given @parent
1184 * directory. The name should be stored in the @name (with the
1185 * understanding that it is already pointing to a a %NAME_MAX+1 sized
1186 * buffer. get_name() should return %0 on success, a negative error code
Jes Sorensen1b1dcc12006-01-09 15:59:24 -08001187 * or error. @get_name will be called without @parent->i_mutex held.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001188 *
1189 * get_parent:
Martin Waitz67be2dd2005-05-01 08:59:26 -07001190 * @get_parent should find the parent directory for the given @child which
1191 * is also a directory. In the event that it cannot be found, or storage
1192 * space cannot be allocated, a %ERR_PTR should be returned.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001193 *
1194 * get_dentry:
Martin Waitz67be2dd2005-05-01 08:59:26 -07001195 * Given a &super_block (@sb) and a pointer to a file-system specific inode
1196 * identifier, possibly an inode number, (@inump) get_dentry() should find
1197 * the identified inode and return a dentry for that inode. Any suitable
1198 * dentry can be returned including, if necessary, a new dentry created with
1199 * d_alloc_root. The caller can then find any other extant dentrys by
1200 * following the d_alias links. If a new dentry was created using
1201 * d_alloc_root, DCACHE_NFSD_DISCONNECTED should be set, and the dentry
1202 * should be d_rehash()ed.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001203 *
Martin Waitz67be2dd2005-05-01 08:59:26 -07001204 * If the inode cannot be found, either a %NULL pointer or an %ERR_PTR code
1205 * can be returned. The @inump will be whatever was passed to
1206 * nfsd_find_fh_dentry() in either the @obj or @parent parameters.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001207 *
1208 * Locking rules:
Jes Sorensen1b1dcc12006-01-09 15:59:24 -08001209 * get_parent is called with child->d_inode->i_mutex down
Martin Waitz67be2dd2005-05-01 08:59:26 -07001210 * get_name is not (which is possibly inconsistent)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001211 */
1212
1213struct export_operations {
1214 struct dentry *(*decode_fh)(struct super_block *sb, __u32 *fh, int fh_len, int fh_type,
1215 int (*acceptable)(void *context, struct dentry *de),
1216 void *context);
1217 int (*encode_fh)(struct dentry *de, __u32 *fh, int *max_len,
1218 int connectable);
1219
1220 /* the following are only called from the filesystem itself */
1221 int (*get_name)(struct dentry *parent, char *name,
1222 struct dentry *child);
1223 struct dentry * (*get_parent)(struct dentry *child);
1224 struct dentry * (*get_dentry)(struct super_block *sb, void *inump);
1225
1226 /* This is set by the exporting module to a standard helper */
1227 struct dentry * (*find_exported_dentry)(
1228 struct super_block *sb, void *obj, void *parent,
1229 int (*acceptable)(void *context, struct dentry *de),
1230 void *context);
1231
1232
1233};
1234
1235extern struct dentry *
1236find_exported_dentry(struct super_block *sb, void *obj, void *parent,
1237 int (*acceptable)(void *context, struct dentry *de),
1238 void *context);
1239
1240struct file_system_type {
1241 const char *name;
1242 int fs_flags;
1243 struct super_block *(*get_sb) (struct file_system_type *, int,
1244 const char *, void *);
1245 void (*kill_sb) (struct super_block *);
1246 struct module *owner;
1247 struct file_system_type * next;
1248 struct list_head fs_supers;
1249};
1250
1251struct super_block *get_sb_bdev(struct file_system_type *fs_type,
1252 int flags, const char *dev_name, void *data,
1253 int (*fill_super)(struct super_block *, void *, int));
1254struct super_block *get_sb_single(struct file_system_type *fs_type,
1255 int flags, void *data,
1256 int (*fill_super)(struct super_block *, void *, int));
1257struct super_block *get_sb_nodev(struct file_system_type *fs_type,
1258 int flags, void *data,
1259 int (*fill_super)(struct super_block *, void *, int));
1260void generic_shutdown_super(struct super_block *sb);
1261void kill_block_super(struct super_block *sb);
1262void kill_anon_super(struct super_block *sb);
1263void kill_litter_super(struct super_block *sb);
1264void deactivate_super(struct super_block *sb);
1265int set_anon_super(struct super_block *s, void *data);
1266struct super_block *sget(struct file_system_type *type,
1267 int (*test)(struct super_block *,void *),
1268 int (*set)(struct super_block *,void *),
1269 void *data);
1270struct super_block *get_sb_pseudo(struct file_system_type *, char *,
1271 struct super_operations *ops, unsigned long);
1272int __put_super(struct super_block *sb);
1273int __put_super_and_need_restart(struct super_block *sb);
1274void unnamed_dev_init(void);
1275
1276/* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1277#define fops_get(fops) \
1278 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1279#define fops_put(fops) \
1280 do { if (fops) module_put((fops)->owner); } while(0)
1281
1282extern int register_filesystem(struct file_system_type *);
1283extern int unregister_filesystem(struct file_system_type *);
1284extern struct vfsmount *kern_mount(struct file_system_type *);
1285extern int may_umount_tree(struct vfsmount *);
1286extern int may_umount(struct vfsmount *);
Ram Paia05964f2005-11-07 17:20:17 -05001287extern void umount_tree(struct vfsmount *, int, struct list_head *);
Ram Paib90fa9a2005-11-07 17:19:50 -05001288extern void release_mounts(struct list_head *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001289extern long do_mount(char *, char *, char *, unsigned long, void *);
Ram Paib90fa9a2005-11-07 17:19:50 -05001290extern struct vfsmount *copy_tree(struct vfsmount *, struct dentry *, int);
1291extern void mnt_set_mountpoint(struct vfsmount *, struct dentry *,
1292 struct vfsmount *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001293
1294extern int vfs_statfs(struct super_block *, struct kstatfs *);
1295
Miklos Szeredif87fd4c2006-01-16 22:14:23 -08001296/* /sys/fs */
1297extern struct subsystem fs_subsys;
1298
Linus Torvalds1da177e2005-04-16 15:20:36 -07001299#define FLOCK_VERIFY_READ 1
1300#define FLOCK_VERIFY_WRITE 2
1301
1302extern int locks_mandatory_locked(struct inode *);
1303extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1304
1305/*
1306 * Candidates for mandatory locking have the setgid bit set
1307 * but no group execute bit - an otherwise meaningless combination.
1308 */
1309#define MANDATORY_LOCK(inode) \
1310 (IS_MANDLOCK(inode) && ((inode)->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID)
1311
1312static inline int locks_verify_locked(struct inode *inode)
1313{
1314 if (MANDATORY_LOCK(inode))
1315 return locks_mandatory_locked(inode);
1316 return 0;
1317}
1318
1319extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1320
1321static inline int locks_verify_truncate(struct inode *inode,
1322 struct file *filp,
1323 loff_t size)
1324{
1325 if (inode->i_flock && MANDATORY_LOCK(inode))
1326 return locks_mandatory_area(
1327 FLOCK_VERIFY_WRITE, inode, filp,
1328 size < inode->i_size ? size : inode->i_size,
1329 (size < inode->i_size ? inode->i_size - size
1330 : size - inode->i_size)
1331 );
1332 return 0;
1333}
1334
1335static inline int break_lease(struct inode *inode, unsigned int mode)
1336{
1337 if (inode->i_flock)
1338 return __break_lease(inode, mode);
1339 return 0;
1340}
1341
1342/* fs/open.c */
1343
NeilBrown4a301312006-01-08 01:02:39 -08001344extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
1345 struct file *filp);
Ulrich Drepper5590ff02006-01-18 17:43:53 -08001346extern long do_sys_open(int fdf, const char __user *filename, int flags,
1347 int mode);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001348extern struct file *filp_open(const char *, int, int);
1349extern struct file * dentry_open(struct dentry *, struct vfsmount *, int);
1350extern int filp_close(struct file *, fl_owner_t id);
1351extern char * getname(const char __user *);
1352
1353/* fs/dcache.c */
1354extern void __init vfs_caches_init_early(void);
1355extern void __init vfs_caches_init(unsigned long);
1356
1357#define __getname() kmem_cache_alloc(names_cachep, SLAB_KERNEL)
1358#define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
1359#ifndef CONFIG_AUDITSYSCALL
1360#define putname(name) __putname(name)
1361#else
1362extern void putname(const char *name);
1363#endif
1364
1365extern int register_blkdev(unsigned int, const char *);
1366extern int unregister_blkdev(unsigned int, const char *);
1367extern struct block_device *bdget(dev_t);
1368extern void bd_set_size(struct block_device *, loff_t size);
1369extern void bd_forget(struct inode *inode);
1370extern void bdput(struct block_device *);
1371extern struct block_device *open_by_devnum(dev_t, unsigned);
1372extern struct file_operations def_blk_fops;
1373extern struct address_space_operations def_blk_aops;
1374extern struct file_operations def_chr_fops;
1375extern struct file_operations bad_sock_fops;
1376extern struct file_operations def_fifo_fops;
1377extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
1378extern int blkdev_ioctl(struct inode *, struct file *, unsigned, unsigned long);
1379extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
1380extern int blkdev_get(struct block_device *, mode_t, unsigned);
1381extern int blkdev_put(struct block_device *);
1382extern int bd_claim(struct block_device *, void *);
1383extern void bd_release(struct block_device *);
1384
1385/* fs/char_dev.c */
1386extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
1387extern int register_chrdev_region(dev_t, unsigned, const char *);
1388extern int register_chrdev(unsigned int, const char *,
1389 struct file_operations *);
1390extern int unregister_chrdev(unsigned int, const char *);
1391extern void unregister_chrdev_region(dev_t, unsigned);
1392extern int chrdev_open(struct inode *, struct file *);
Neil Horman7170be52006-01-14 13:20:38 -08001393extern int get_chrdev_list(char *);
1394extern void *acquire_chrdev_list(void);
1395extern int count_chrdev_list(void);
1396extern void *get_next_chrdev(void *);
1397extern int get_chrdev_info(void *, int *, char **);
1398extern void release_chrdev_list(void *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001399
1400/* fs/block_dev.c */
1401#define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
1402extern const char *__bdevname(dev_t, char *buffer);
1403extern const char *bdevname(struct block_device *bdev, char *buffer);
1404extern struct block_device *lookup_bdev(const char *);
1405extern struct block_device *open_bdev_excl(const char *, int, void *);
1406extern void close_bdev_excl(struct block_device *);
Neil Horman7170be52006-01-14 13:20:38 -08001407extern void *acquire_blkdev_list(void);
1408extern int count_blkdev_list(void);
1409extern void *get_next_blkdev(void *);
1410extern int get_blkdev_info(void *, int *, char **);
1411extern void release_blkdev_list(void *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001412
1413extern void init_special_inode(struct inode *, umode_t, dev_t);
1414
1415/* Invalid inode operations -- fs/bad_inode.c */
1416extern void make_bad_inode(struct inode *);
1417extern int is_bad_inode(struct inode *);
1418
1419extern struct file_operations read_fifo_fops;
1420extern struct file_operations write_fifo_fops;
1421extern struct file_operations rdwr_fifo_fops;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001422
1423extern int fs_may_remount_ro(struct super_block *);
1424
1425/*
1426 * return READ, READA, or WRITE
1427 */
1428#define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
1429
1430/*
1431 * return data direction, READ or WRITE
1432 */
1433#define bio_data_dir(bio) ((bio)->bi_rw & 1)
1434
1435extern int check_disk_change(struct block_device *);
1436extern int invalidate_inodes(struct super_block *);
Christoph Hellwig2ef41632005-05-05 16:15:59 -07001437extern int __invalidate_device(struct block_device *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001438extern int invalidate_partition(struct gendisk *, int);
1439unsigned long invalidate_mapping_pages(struct address_space *mapping,
1440 pgoff_t start, pgoff_t end);
1441unsigned long invalidate_inode_pages(struct address_space *mapping);
1442static inline void invalidate_remote_inode(struct inode *inode)
1443{
1444 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
1445 S_ISLNK(inode->i_mode))
1446 invalidate_inode_pages(inode->i_mapping);
1447}
1448extern int invalidate_inode_pages2(struct address_space *mapping);
1449extern int invalidate_inode_pages2_range(struct address_space *mapping,
1450 pgoff_t start, pgoff_t end);
1451extern int write_inode_now(struct inode *, int);
1452extern int filemap_fdatawrite(struct address_space *);
1453extern int filemap_flush(struct address_space *);
1454extern int filemap_fdatawait(struct address_space *);
1455extern int filemap_write_and_wait(struct address_space *mapping);
1456extern int filemap_write_and_wait_range(struct address_space *mapping,
1457 loff_t lstart, loff_t lend);
1458extern void sync_supers(void);
1459extern void sync_filesystems(int wait);
1460extern void emergency_sync(void);
1461extern void emergency_remount(void);
1462extern int do_remount_sb(struct super_block *sb, int flags,
1463 void *data, int force);
1464extern sector_t bmap(struct inode *, sector_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001465extern int notify_change(struct dentry *, struct iattr *);
1466extern int permission(struct inode *, int, struct nameidata *);
1467extern int generic_permission(struct inode *, int,
1468 int (*check_acl)(struct inode *, int));
1469
1470extern int get_write_access(struct inode *);
1471extern int deny_write_access(struct file *);
1472static inline void put_write_access(struct inode * inode)
1473{
1474 atomic_dec(&inode->i_writecount);
1475}
1476static inline void allow_write_access(struct file *file)
1477{
1478 if (file)
1479 atomic_inc(&file->f_dentry->d_inode->i_writecount);
1480}
1481extern int do_pipe(int *);
1482
Ulrich Drepper5590ff02006-01-18 17:43:53 -08001483extern int open_namei(int dfd, const char *, int, int, struct nameidata *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001484extern int may_open(struct nameidata *, int, int);
1485
1486extern int kernel_read(struct file *, unsigned long, char *, unsigned long);
1487extern struct file * open_exec(const char *);
1488
1489/* fs/dcache.c -- generic fs support functions */
1490extern int is_subdir(struct dentry *, struct dentry *);
1491extern ino_t find_inode_number(struct dentry *, struct qstr *);
1492
1493#include <linux/err.h>
1494
1495/* needed for stackable file system support */
1496extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
1497
1498extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
1499
1500extern void inode_init_once(struct inode *);
1501extern void iput(struct inode *);
1502extern struct inode * igrab(struct inode *);
1503extern ino_t iunique(struct super_block *, ino_t);
1504extern int inode_needs_sync(struct inode *inode);
1505extern void generic_delete_inode(struct inode *inode);
Mark Fashehcb2c0232005-07-07 17:56:03 -07001506extern void generic_drop_inode(struct inode *inode);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001507
Anton Altaparmakov88bd5122005-07-13 01:10:44 -07001508extern struct inode *ilookup5_nowait(struct super_block *sb,
1509 unsigned long hashval, int (*test)(struct inode *, void *),
1510 void *data);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001511extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
1512 int (*test)(struct inode *, void *), void *data);
1513extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
1514
1515extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
1516extern struct inode * iget_locked(struct super_block *, unsigned long);
1517extern void unlock_new_inode(struct inode *);
1518
1519static inline struct inode *iget(struct super_block *sb, unsigned long ino)
1520{
1521 struct inode *inode = iget_locked(sb, ino);
1522
1523 if (inode && (inode->i_state & I_NEW)) {
1524 sb->s_op->read_inode(inode);
1525 unlock_new_inode(inode);
1526 }
1527
1528 return inode;
1529}
1530
1531extern void __iget(struct inode * inode);
1532extern void clear_inode(struct inode *);
1533extern void destroy_inode(struct inode *);
1534extern struct inode *new_inode(struct super_block *);
1535extern int remove_suid(struct dentry *);
1536extern void remove_dquot_ref(struct super_block *, int, struct list_head *);
1537extern struct semaphore iprune_sem;
1538
1539extern void __insert_inode_hash(struct inode *, unsigned long hashval);
1540extern void remove_inode_hash(struct inode *);
1541static inline void insert_inode_hash(struct inode *inode) {
1542 __insert_inode_hash(inode, inode->i_ino);
1543}
1544
1545extern struct file * get_empty_filp(void);
1546extern void file_move(struct file *f, struct list_head *list);
1547extern void file_kill(struct file *f);
1548struct bio;
1549extern void submit_bio(int, struct bio *);
1550extern int bdev_read_only(struct block_device *);
1551extern int set_blocksize(struct block_device *, int);
1552extern int sb_set_blocksize(struct super_block *, int);
1553extern int sb_min_blocksize(struct super_block *, int);
1554
1555extern int generic_file_mmap(struct file *, struct vm_area_struct *);
1556extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
1557extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
1558extern int file_send_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
1559extern ssize_t generic_file_read(struct file *, char __user *, size_t, loff_t *);
1560int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
1561extern ssize_t generic_file_write(struct file *, const char __user *, size_t, loff_t *);
1562extern ssize_t generic_file_aio_read(struct kiocb *, char __user *, size_t, loff_t);
1563extern ssize_t __generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t *);
1564extern ssize_t generic_file_aio_write(struct kiocb *, const char __user *, size_t, loff_t);
1565extern ssize_t generic_file_aio_write_nolock(struct kiocb *, const struct iovec *,
1566 unsigned long, loff_t *);
1567extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
1568 unsigned long *, loff_t, loff_t *, size_t, size_t);
1569extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
1570 unsigned long, loff_t, loff_t *, size_t, ssize_t);
1571extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
1572extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
1573ssize_t generic_file_write_nolock(struct file *file, const struct iovec *iov,
1574 unsigned long nr_segs, loff_t *ppos);
1575extern ssize_t generic_file_sendfile(struct file *, loff_t *, size_t, read_actor_t, void *);
1576extern void do_generic_mapping_read(struct address_space *mapping,
1577 struct file_ra_state *, struct file *,
1578 loff_t *, read_descriptor_t *, read_actor_t);
1579extern void
1580file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001581extern ssize_t generic_file_readv(struct file *filp, const struct iovec *iov,
1582 unsigned long nr_segs, loff_t *ppos);
1583ssize_t generic_file_writev(struct file *filp, const struct iovec *iov,
1584 unsigned long nr_segs, loff_t *ppos);
1585extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
1586extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
1587extern loff_t remote_llseek(struct file *file, loff_t offset, int origin);
1588extern int generic_file_open(struct inode * inode, struct file * filp);
1589extern int nonseekable_open(struct inode * inode, struct file * filp);
1590
Carsten Otteceffc072005-06-23 22:05:25 -07001591#ifdef CONFIG_FS_XIP
Carsten Otteeb6fe0c2005-06-23 22:05:28 -07001592extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
1593 loff_t *ppos);
Carsten Otteceffc072005-06-23 22:05:25 -07001594extern ssize_t xip_file_sendfile(struct file *in_file, loff_t *ppos,
1595 size_t count, read_actor_t actor,
1596 void *target);
1597extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
Carsten Otteeb6fe0c2005-06-23 22:05:28 -07001598extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
1599 size_t len, loff_t *ppos);
Carsten Otteceffc072005-06-23 22:05:25 -07001600extern int xip_truncate_page(struct address_space *mapping, loff_t from);
Carsten Otte6d791252005-06-23 22:05:26 -07001601#else
1602static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
1603{
1604 return 0;
1605}
Carsten Otteceffc072005-06-23 22:05:25 -07001606#endif
1607
Linus Torvalds1da177e2005-04-16 15:20:36 -07001608static inline void do_generic_file_read(struct file * filp, loff_t *ppos,
1609 read_descriptor_t * desc,
1610 read_actor_t actor)
1611{
1612 do_generic_mapping_read(filp->f_mapping,
1613 &filp->f_ra,
1614 filp,
1615 ppos,
1616 desc,
1617 actor);
1618}
1619
1620ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
1621 struct block_device *bdev, const struct iovec *iov, loff_t offset,
1622 unsigned long nr_segs, get_blocks_t get_blocks, dio_iodone_t end_io,
1623 int lock_type);
1624
1625enum {
1626 DIO_LOCKING = 1, /* need locking between buffered and direct access */
1627 DIO_NO_LOCKING, /* bdev; no locking at all between buffered/direct */
1628 DIO_OWN_LOCKING, /* filesystem locks buffered and direct internally */
1629};
1630
1631static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
1632 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1633 loff_t offset, unsigned long nr_segs, get_blocks_t get_blocks,
1634 dio_iodone_t end_io)
1635{
1636 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1637 nr_segs, get_blocks, end_io, DIO_LOCKING);
1638}
1639
1640static inline ssize_t blockdev_direct_IO_no_locking(int rw, struct kiocb *iocb,
1641 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1642 loff_t offset, unsigned long nr_segs, get_blocks_t get_blocks,
1643 dio_iodone_t end_io)
1644{
1645 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1646 nr_segs, get_blocks, end_io, DIO_NO_LOCKING);
1647}
1648
1649static inline ssize_t blockdev_direct_IO_own_locking(int rw, struct kiocb *iocb,
1650 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1651 loff_t offset, unsigned long nr_segs, get_blocks_t get_blocks,
1652 dio_iodone_t end_io)
1653{
1654 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1655 nr_segs, get_blocks, end_io, DIO_OWN_LOCKING);
1656}
1657
1658extern struct file_operations generic_ro_fops;
1659
1660#define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
1661
1662extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
1663extern int vfs_follow_link(struct nameidata *, const char *);
1664extern int page_readlink(struct dentry *, char __user *, int);
Linus Torvaldscc314ee2005-08-19 18:02:56 -07001665extern void *page_follow_link_light(struct dentry *, struct nameidata *);
1666extern void page_put_link(struct dentry *, struct nameidata *, void *);
Kirill Korotaev0adb25d2006-03-11 03:27:13 -08001667extern int __page_symlink(struct inode *inode, const char *symname, int len,
1668 gfp_t gfp_mask);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001669extern int page_symlink(struct inode *inode, const char *symname, int len);
1670extern struct inode_operations page_symlink_inode_operations;
1671extern int generic_readlink(struct dentry *, char __user *, int);
1672extern void generic_fillattr(struct inode *, struct kstat *);
1673extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1674void inode_add_bytes(struct inode *inode, loff_t bytes);
1675void inode_sub_bytes(struct inode *inode, loff_t bytes);
1676loff_t inode_get_bytes(struct inode *inode);
1677void inode_set_bytes(struct inode *inode, loff_t bytes);
1678
1679extern int vfs_readdir(struct file *, filldir_t, void *);
1680
1681extern int vfs_stat(char __user *, struct kstat *);
1682extern int vfs_lstat(char __user *, struct kstat *);
Ulrich Drepper5590ff02006-01-18 17:43:53 -08001683extern int vfs_stat_fd(int dfd, char __user *, struct kstat *);
1684extern int vfs_lstat_fd(int dfd, char __user *, struct kstat *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001685extern int vfs_fstat(unsigned int, struct kstat *);
1686
1687extern int vfs_ioctl(struct file *, unsigned int, unsigned int, unsigned long);
1688
1689extern struct file_system_type *get_fs_type(const char *name);
1690extern struct super_block *get_super(struct block_device *);
1691extern struct super_block *user_get_super(dev_t);
1692extern void drop_super(struct super_block *sb);
1693
1694extern int dcache_dir_open(struct inode *, struct file *);
1695extern int dcache_dir_close(struct inode *, struct file *);
1696extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
1697extern int dcache_readdir(struct file *, void *, filldir_t);
1698extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1699extern int simple_statfs(struct super_block *, struct kstatfs *);
1700extern int simple_link(struct dentry *, struct inode *, struct dentry *);
1701extern int simple_unlink(struct inode *, struct dentry *);
1702extern int simple_rmdir(struct inode *, struct dentry *);
1703extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1704extern int simple_sync_file(struct file *, struct dentry *, int);
1705extern int simple_empty(struct dentry *);
1706extern int simple_readpage(struct file *file, struct page *page);
1707extern int simple_prepare_write(struct file *file, struct page *page,
1708 unsigned offset, unsigned to);
1709extern int simple_commit_write(struct file *file, struct page *page,
1710 unsigned offset, unsigned to);
1711
1712extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
1713extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
1714extern struct file_operations simple_dir_operations;
1715extern struct inode_operations simple_dir_inode_operations;
1716struct tree_descr { char *name; struct file_operations *ops; int mode; };
1717struct dentry *d_alloc_name(struct dentry *, const char *);
1718extern int simple_fill_super(struct super_block *, int, struct tree_descr *);
1719extern int simple_pin_fs(char *name, struct vfsmount **mount, int *count);
1720extern void simple_release_fs(struct vfsmount **mount, int *count);
1721
1722extern ssize_t simple_read_from_buffer(void __user *, size_t, loff_t *, const void *, size_t);
1723
Christoph Lametere965f962006-02-01 03:05:41 -08001724#ifdef CONFIG_MIGRATION
1725extern int buffer_migrate_page(struct page *, struct page *);
1726#else
1727#define buffer_migrate_page NULL
1728#endif
1729
Linus Torvalds1da177e2005-04-16 15:20:36 -07001730extern int inode_change_ok(struct inode *, struct iattr *);
1731extern int __must_check inode_setattr(struct inode *, struct iattr *);
1732
Christoph Hellwig870f4812006-01-09 20:52:01 -08001733extern void file_update_time(struct file *file);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001734
1735static inline ino_t parent_ino(struct dentry *dentry)
1736{
1737 ino_t res;
1738
1739 spin_lock(&dentry->d_lock);
1740 res = dentry->d_parent->d_inode->i_ino;
1741 spin_unlock(&dentry->d_lock);
1742 return res;
1743}
1744
1745/* kernel/fork.c */
1746extern int unshare_files(void);
1747
1748/* Transaction based IO helpers */
1749
1750/*
1751 * An argresp is stored in an allocated page and holds the
1752 * size of the argument or response, along with its content
1753 */
1754struct simple_transaction_argresp {
1755 ssize_t size;
1756 char data[0];
1757};
1758
1759#define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
1760
1761char *simple_transaction_get(struct file *file, const char __user *buf,
1762 size_t size);
1763ssize_t simple_transaction_read(struct file *file, char __user *buf,
1764 size_t size, loff_t *pos);
1765int simple_transaction_release(struct inode *inode, struct file *file);
1766
1767static inline void simple_transaction_set(struct file *file, size_t n)
1768{
1769 struct simple_transaction_argresp *ar = file->private_data;
1770
1771 BUG_ON(n > SIMPLE_TRANSACTION_LIMIT);
1772
1773 /*
1774 * The barrier ensures that ar->size will really remain zero until
1775 * ar->data is ready for reading.
1776 */
1777 smp_mb();
1778 ar->size = n;
1779}
1780
Arnd Bergmannacaefc22005-05-18 14:40:59 +02001781/*
1782 * simple attribute files
1783 *
1784 * These attributes behave similar to those in sysfs:
1785 *
1786 * Writing to an attribute immediately sets a value, an open file can be
1787 * written to multiple times.
1788 *
1789 * Reading from an attribute creates a buffer from the value that might get
1790 * read with multiple read calls. When the attribute has been read
1791 * completely, no further read calls are possible until the file is opened
1792 * again.
1793 *
1794 * All attributes contain a text representation of a numeric value
1795 * that are accessed with the get() and set() functions.
1796 */
1797#define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
1798static int __fops ## _open(struct inode *inode, struct file *file) \
1799{ \
1800 __simple_attr_check_format(__fmt, 0ull); \
1801 return simple_attr_open(inode, file, __get, __set, __fmt); \
1802} \
1803static struct file_operations __fops = { \
1804 .owner = THIS_MODULE, \
1805 .open = __fops ## _open, \
1806 .release = simple_attr_close, \
1807 .read = simple_attr_read, \
1808 .write = simple_attr_write, \
1809};
1810
1811static inline void __attribute__((format(printf, 1, 2)))
1812__simple_attr_check_format(const char *fmt, ...)
1813{
1814 /* don't do anything, just let the compiler check the arguments; */
1815}
1816
1817int simple_attr_open(struct inode *inode, struct file *file,
1818 u64 (*get)(void *), void (*set)(void *, u64),
1819 const char *fmt);
1820int simple_attr_close(struct inode *inode, struct file *file);
1821ssize_t simple_attr_read(struct file *file, char __user *buf,
1822 size_t len, loff_t *ppos);
1823ssize_t simple_attr_write(struct file *file, const char __user *buf,
1824 size_t len, loff_t *ppos);
1825
1826
Linus Torvalds1da177e2005-04-16 15:20:36 -07001827#ifdef CONFIG_SECURITY
1828static inline char *alloc_secdata(void)
1829{
1830 return (char *)get_zeroed_page(GFP_KERNEL);
1831}
1832
1833static inline void free_secdata(void *secdata)
1834{
1835 free_page((unsigned long)secdata);
1836}
1837#else
1838static inline char *alloc_secdata(void)
1839{
1840 return (char *)1;
1841}
1842
1843static inline void free_secdata(void *secdata)
1844{ }
1845#endif /* CONFIG_SECURITY */
1846
1847#endif /* __KERNEL__ */
1848#endif /* _LINUX_FS_H */