blob: ef355bc73714104fcb710c2d444188c7be20ecd2 [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
Oleg Drokinb5005312006-03-25 03:07:01 -080068/* File is being opened for execution. Primary users of this flag are
69 distributed filesystems that can use it to achieve correct ETXTBUSY
70 behavior for cross-node execution/opening_for_writing of files */
71#define FMODE_EXEC 16
72
Linus Torvalds1da177e2005-04-16 15:20:36 -070073#define RW_MASK 1
74#define RWA_MASK 2
75#define READ 0
76#define WRITE 1
77#define READA 2 /* read-ahead - don't block if no resources */
Jan Karaa7662232005-09-06 15:19:10 -070078#define SWRITE 3 /* for ll_rw_block() - wait for buffer lock */
Linus Torvalds1da177e2005-04-16 15:20:36 -070079#define SPECIAL 4 /* For non-blockdevice requests in request queue */
80#define READ_SYNC (READ | (1 << BIO_RW_SYNC))
81#define WRITE_SYNC (WRITE | (1 << BIO_RW_SYNC))
82#define WRITE_BARRIER ((1 << BIO_RW) | (1 << BIO_RW_BARRIER))
83
84#define SEL_IN 1
85#define SEL_OUT 2
86#define SEL_EX 4
87
88/* public flags for file_system_type */
89#define FS_REQUIRES_DEV 1
90#define FS_BINARY_MOUNTDATA 2
91#define FS_REVAL_DOT 16384 /* Check the paths ".", ".." for staleness */
92#define FS_ODD_RENAME 32768 /* Temporary stuff; will go away as soon
93 * as nfs_rename() will be cleaned up
94 */
95/*
96 * These are the fs-independent mount-flags: up to 32 flags are supported
97 */
98#define MS_RDONLY 1 /* Mount read-only */
99#define MS_NOSUID 2 /* Ignore suid and sgid bits */
100#define MS_NODEV 4 /* Disallow access to device special files */
101#define MS_NOEXEC 8 /* Disallow program execution */
102#define MS_SYNCHRONOUS 16 /* Writes are synced at once */
103#define MS_REMOUNT 32 /* Alter flags of a mounted FS */
104#define MS_MANDLOCK 64 /* Allow mandatory locks on an FS */
105#define MS_DIRSYNC 128 /* Directory modifications are synchronous */
106#define MS_NOATIME 1024 /* Do not update access times. */
107#define MS_NODIRATIME 2048 /* Do not update directory access times */
108#define MS_BIND 4096
109#define MS_MOVE 8192
110#define MS_REC 16384
Theodore Ts'o9b04c992006-03-24 03:15:10 -0800111#define MS_VERBOSE 32768 /* War is peace. Verbosity is silence.
112 MS_VERBOSE is deprecated. */
113#define MS_SILENT 32768
Miklos Szeredibf066c72006-01-08 01:03:19 -0800114#define MS_POSIXACL (1<<16) /* VFS does not apply the umask */
Ram Pai9676f0c2005-11-07 17:21:20 -0500115#define MS_UNBINDABLE (1<<17) /* change to unbindable */
Ram Pai07b20882005-11-07 17:19:07 -0500116#define MS_PRIVATE (1<<18) /* change to private */
Ram Paia58b0eb2005-11-07 17:20:48 -0500117#define MS_SLAVE (1<<19) /* change to slave */
Ram Pai03e06e62005-11-07 17:19:33 -0500118#define MS_SHARED (1<<20) /* change to shared */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700119#define MS_ACTIVE (1<<30)
120#define MS_NOUSER (1<<31)
121
122/*
123 * Superblock flags that can be altered by MS_REMOUNT
124 */
Christoph Hellwigfc33a7b2006-01-09 20:52:17 -0800125#define MS_RMT_MASK (MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700126
127/*
128 * Old magic mount flag and mask
129 */
130#define MS_MGC_VAL 0xC0ED0000
131#define MS_MGC_MSK 0xffff0000
132
133/* Inode flags - they have nothing to superblock flags now */
134
135#define S_SYNC 1 /* Writes are synced at once */
136#define S_NOATIME 2 /* Do not update access times */
137#define S_APPEND 4 /* Append-only file */
138#define S_IMMUTABLE 8 /* Immutable file */
139#define S_DEAD 16 /* removed, but still open directory */
140#define S_NOQUOTA 32 /* Inode is not counted to quota */
141#define S_DIRSYNC 64 /* Directory modifications are synchronous */
142#define S_NOCMTIME 128 /* Do not update file c/mtime */
143#define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
144#define S_PRIVATE 512 /* Inode is fs-internal */
145
146/*
147 * Note that nosuid etc flags are inode-specific: setting some file-system
148 * flags just means all the inodes inherit those flags by default. It might be
149 * possible to override it selectively if you really wanted to with some
150 * ioctl() that is not currently implemented.
151 *
152 * Exception: MS_RDONLY is always applied to the entire file system.
153 *
154 * Unfortunately, it is possible to change a filesystems flags with it mounted
155 * with files in use. This means that all of the inodes will not have their
156 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
157 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
158 */
159#define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))
160
161#define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
162#define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
163 ((inode)->i_flags & S_SYNC))
164#define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
165 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
166#define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
167
168#define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
169#define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
170#define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700171#define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
172
173#define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
174#define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
175#define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
176#define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
177
178/* the read-only stuff doesn't really belong here, but any other place is
179 probably as bad and I don't want to create yet another include file. */
180
181#define BLKROSET _IO(0x12,93) /* set device read-only (0 = read-write) */
182#define BLKROGET _IO(0x12,94) /* get read-only status (0 = read_write) */
183#define BLKRRPART _IO(0x12,95) /* re-read partition table */
184#define BLKGETSIZE _IO(0x12,96) /* return device size /512 (long *arg) */
185#define BLKFLSBUF _IO(0x12,97) /* flush buffer cache */
186#define BLKRASET _IO(0x12,98) /* set read ahead for block device */
187#define BLKRAGET _IO(0x12,99) /* get current read ahead setting */
188#define BLKFRASET _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
189#define BLKFRAGET _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
190#define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
191#define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
192#define BLKSSZGET _IO(0x12,104)/* get block device sector size */
193#if 0
194#define BLKPG _IO(0x12,105)/* See blkpg.h */
195
196/* Some people are morons. Do not use sizeof! */
197
198#define BLKELVGET _IOR(0x12,106,size_t)/* elevator get */
199#define BLKELVSET _IOW(0x12,107,size_t)/* elevator set */
200/* This was here just to show that the number is taken -
201 probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
202#endif
203/* A jump here: 108-111 have been used for various private purposes. */
204#define BLKBSZGET _IOR(0x12,112,size_t)
205#define BLKBSZSET _IOW(0x12,113,size_t)
206#define BLKGETSIZE64 _IOR(0x12,114,size_t) /* return device size in bytes (u64 *arg) */
Jens Axboe2056a782006-03-23 20:00:26 +0100207#define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup)
208#define BLKTRACESTART _IO(0x12,116)
209#define BLKTRACESTOP _IO(0x12,117)
210#define BLKTRACETEARDOWN _IO(0x12,118)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700211
212#define BMAP_IOCTL 1 /* obsolete - kept for compatibility */
213#define FIBMAP _IO(0x00,1) /* bmap access */
214#define FIGETBSZ _IO(0x00,2) /* get the block size used for bmap */
215
216#ifdef __KERNEL__
217
218#include <linux/linkage.h>
219#include <linux/wait.h>
220#include <linux/types.h>
221#include <linux/kdev_t.h>
222#include <linux/dcache.h>
223#include <linux/stat.h>
224#include <linux/cache.h>
225#include <linux/kobject.h>
226#include <linux/list.h>
227#include <linux/radix-tree.h>
228#include <linux/prio_tree.h>
229#include <linux/init.h>
Jens Axboe22e2c502005-06-27 10:55:12 +0200230#include <linux/sched.h>
Jes Sorensen1b1dcc12006-01-09 15:59:24 -0800231#include <linux/mutex.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700232
233#include <asm/atomic.h>
234#include <asm/semaphore.h>
235#include <asm/byteorder.h>
236
Christoph Hellwiga885c8c2006-01-08 01:02:50 -0800237struct hd_geometry;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700238struct iovec;
239struct nameidata;
Christoph Hellwig92198f72005-06-23 22:00:59 -0700240struct kiocb;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700241struct pipe_inode_info;
242struct poll_table_struct;
243struct kstatfs;
244struct vm_area_struct;
245struct vfsmount;
246
Linus Torvalds1da177e2005-04-16 15:20:36 -0700247extern void __init inode_init(unsigned long);
248extern void __init inode_init_early(void);
249extern void __init mnt_init(unsigned long);
250extern void __init files_init(unsigned long);
251
252struct buffer_head;
253typedef int (get_block_t)(struct inode *inode, sector_t iblock,
254 struct buffer_head *bh_result, int create);
Christoph Hellwig92198f72005-06-23 22:00:59 -0700255typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700256 ssize_t bytes, void *private);
257
258/*
259 * Attribute flags. These should be or-ed together to figure out what
260 * has been changed!
261 */
262#define ATTR_MODE 1
263#define ATTR_UID 2
264#define ATTR_GID 4
265#define ATTR_SIZE 8
266#define ATTR_ATIME 16
267#define ATTR_MTIME 32
268#define ATTR_CTIME 64
269#define ATTR_ATIME_SET 128
270#define ATTR_MTIME_SET 256
271#define ATTR_FORCE 512 /* Not a change, but a change it */
272#define ATTR_ATTR_FLAG 1024
273#define ATTR_KILL_SUID 2048
274#define ATTR_KILL_SGID 4096
Miklos Szeredicc4e69d2005-11-07 00:59:49 -0800275#define ATTR_FILE 8192
Linus Torvalds1da177e2005-04-16 15:20:36 -0700276
277/*
278 * This is the Inode Attributes structure, used for notify_change(). It
279 * uses the above definitions as flags, to know which values have changed.
280 * Also, in this manner, a Filesystem can look at only the values it cares
281 * about. Basically, these are the attributes that the VFS layer can
282 * request to change from the FS layer.
283 *
284 * Derek Atkins <warlord@MIT.EDU> 94-10-20
285 */
286struct iattr {
287 unsigned int ia_valid;
288 umode_t ia_mode;
289 uid_t ia_uid;
290 gid_t ia_gid;
291 loff_t ia_size;
292 struct timespec ia_atime;
293 struct timespec ia_mtime;
294 struct timespec ia_ctime;
Miklos Szeredicc4e69d2005-11-07 00:59:49 -0800295
296 /*
297 * Not an attribute, but an auxilary info for filesystems wanting to
298 * implement an ftruncate() like method. NOTE: filesystem should
299 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
300 */
301 struct file *ia_file;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700302};
303
304/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700305 * Includes for diskquotas.
306 */
307#include <linux/quota.h>
308
Zach Brown994fc28c2005-12-15 14:28:17 -0800309/**
310 * enum positive_aop_returns - aop return codes with specific semantics
311 *
312 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
313 * completed, that the page is still locked, and
314 * should be considered active. The VM uses this hint
315 * to return the page to the active list -- it won't
316 * be a candidate for writeback again in the near
317 * future. Other callers must be careful to unlock
318 * the page if they get this return. Returned by
319 * writepage();
320 *
321 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
322 * unlocked it and the page might have been truncated.
323 * The caller should back up to acquiring a new page and
324 * trying again. The aop will be taking reasonable
325 * precautions not to livelock. If the caller held a page
326 * reference, it should drop it before retrying. Returned
327 * by readpage(), prepare_write(), and commit_write().
328 *
329 * address_space_operation functions return these large constants to indicate
330 * special semantics to the caller. These are much larger than the bytes in a
331 * page to allow for functions that return the number of bytes operated on in a
332 * given page.
333 */
334
335enum positive_aop_returns {
336 AOP_WRITEPAGE_ACTIVATE = 0x80000,
337 AOP_TRUNCATED_PAGE = 0x80001,
338};
339
Linus Torvalds1da177e2005-04-16 15:20:36 -0700340/*
341 * oh the beauties of C type declarations.
342 */
343struct page;
344struct address_space;
345struct writeback_control;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700346
347struct address_space_operations {
348 int (*writepage)(struct page *page, struct writeback_control *wbc);
349 int (*readpage)(struct file *, struct page *);
NeilBrown3978d712006-03-26 01:37:17 -0800350 void (*sync_page)(struct page *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700351
352 /* Write back some dirty pages from this mapping. */
353 int (*writepages)(struct address_space *, struct writeback_control *);
354
Andrew Morton4741c9f2006-03-24 03:18:11 -0800355 /* Set a page dirty. Return true if this dirtied it */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700356 int (*set_page_dirty)(struct page *page);
357
358 int (*readpages)(struct file *filp, struct address_space *mapping,
359 struct list_head *pages, unsigned nr_pages);
360
361 /*
362 * ext3 requires that a successful prepare_write() call be followed
363 * by a commit_write() call - they must be balanced
364 */
365 int (*prepare_write)(struct file *, struct page *, unsigned, unsigned);
366 int (*commit_write)(struct file *, struct page *, unsigned, unsigned);
367 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
368 sector_t (*bmap)(struct address_space *, sector_t);
NeilBrown2ff28e22006-03-26 01:37:18 -0800369 void (*invalidatepage) (struct page *, unsigned long);
Al Viro27496a82005-10-21 03:20:48 -0400370 int (*releasepage) (struct page *, gfp_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700371 ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
372 loff_t offset, unsigned long nr_segs);
Carsten Otteceffc072005-06-23 22:05:25 -0700373 struct page* (*get_xip_page)(struct address_space *, sector_t,
374 int);
Christoph Lametere965f962006-02-01 03:05:41 -0800375 /* migrate the contents of a page to the specified target */
376 int (*migratepage) (struct page *, struct page *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700377};
378
379struct backing_dev_info;
380struct address_space {
381 struct inode *host; /* owner: inode, block_device */
382 struct radix_tree_root page_tree; /* radix tree of all pages */
383 rwlock_t tree_lock; /* and rwlock protecting it */
384 unsigned int i_mmap_writable;/* count VM_SHARED mappings */
385 struct prio_tree_root i_mmap; /* tree of private and shared mappings */
386 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
387 spinlock_t i_mmap_lock; /* protect tree, count, list */
388 unsigned int truncate_count; /* Cover race condition with truncate */
389 unsigned long nrpages; /* number of total pages */
390 pgoff_t writeback_index;/* writeback starts here */
391 struct address_space_operations *a_ops; /* methods */
392 unsigned long flags; /* error bits/gfp mask */
393 struct backing_dev_info *backing_dev_info; /* device readahead, etc */
394 spinlock_t private_lock; /* for use by the address_space */
395 struct list_head private_list; /* ditto */
396 struct address_space *assoc_mapping; /* ditto */
397} __attribute__((aligned(sizeof(long))));
398 /*
399 * On most architectures that alignment is already the case; but
400 * must be enforced here for CRIS, to let the least signficant bit
401 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
402 */
403
404struct block_device {
405 dev_t bd_dev; /* not a kdev_t - it's a search key */
406 struct inode * bd_inode; /* will die */
407 int bd_openers;
Arjan van de Venc039e312006-03-23 03:00:28 -0800408 struct mutex bd_mutex; /* open/close mutex */
409 struct mutex bd_mount_mutex; /* mount mutex */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700410 struct list_head bd_inodes;
411 void * bd_holder;
412 int bd_holders;
Jun'ichi Nomura641dc632006-03-27 01:17:57 -0800413#ifdef CONFIG_SYSFS
414 struct list_head bd_holder_list;
415#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700416 struct block_device * bd_contains;
417 unsigned bd_block_size;
418 struct hd_struct * bd_part;
419 /* number of times partitions within this device have been opened. */
420 unsigned bd_part_count;
421 int bd_invalidated;
422 struct gendisk * bd_disk;
423 struct list_head bd_list;
424 struct backing_dev_info *bd_inode_backing_dev_info;
425 /*
426 * Private data. You must have bd_claim'ed the block_device
427 * to use this. NOTE: bd_claim allows an owner to claim
428 * the same device multiple times, the owner must take special
429 * care to not mess up bd_private for that case.
430 */
431 unsigned long bd_private;
432};
433
434/*
435 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
436 * radix trees
437 */
438#define PAGECACHE_TAG_DIRTY 0
439#define PAGECACHE_TAG_WRITEBACK 1
440
441int mapping_tagged(struct address_space *mapping, int tag);
442
443/*
444 * Might pages of this file be mapped into userspace?
445 */
446static inline int mapping_mapped(struct address_space *mapping)
447{
448 return !prio_tree_empty(&mapping->i_mmap) ||
449 !list_empty(&mapping->i_mmap_nonlinear);
450}
451
452/*
453 * Might pages of this file have been modified in userspace?
454 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
455 * marks vma as VM_SHARED if it is shared, and the file was opened for
456 * writing i.e. vma may be mprotected writable even if now readonly.
457 */
458static inline int mapping_writably_mapped(struct address_space *mapping)
459{
460 return mapping->i_mmap_writable != 0;
461}
462
463/*
464 * Use sequence counter to get consistent i_size on 32-bit processors.
465 */
466#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
467#include <linux/seqlock.h>
468#define __NEED_I_SIZE_ORDERED
469#define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
470#else
471#define i_size_ordered_init(inode) do { } while (0)
472#endif
473
474struct inode {
475 struct hlist_node i_hash;
476 struct list_head i_list;
477 struct list_head i_sb_list;
478 struct list_head i_dentry;
479 unsigned long i_ino;
480 atomic_t i_count;
481 umode_t i_mode;
482 unsigned int i_nlink;
483 uid_t i_uid;
484 gid_t i_gid;
485 dev_t i_rdev;
486 loff_t i_size;
487 struct timespec i_atime;
488 struct timespec i_mtime;
489 struct timespec i_ctime;
490 unsigned int i_blkbits;
491 unsigned long i_blksize;
492 unsigned long i_version;
Takashi Satoa0f62ac2006-03-26 01:37:52 -0800493 blkcnt_t i_blocks;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700494 unsigned short i_bytes;
495 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
Jes Sorensen1b1dcc12006-01-09 15:59:24 -0800496 struct mutex i_mutex;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700497 struct rw_semaphore i_alloc_sem;
498 struct inode_operations *i_op;
Arjan van de Ven99ac48f2006-03-28 01:56:41 -0800499 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700500 struct super_block *i_sb;
501 struct file_lock *i_flock;
502 struct address_space *i_mapping;
503 struct address_space i_data;
504#ifdef CONFIG_QUOTA
505 struct dquot *i_dquot[MAXQUOTAS];
506#endif
507 /* These three should probably be a union */
508 struct list_head i_devices;
509 struct pipe_inode_info *i_pipe;
510 struct block_device *i_bdev;
511 struct cdev *i_cdev;
512 int i_cindex;
513
514 __u32 i_generation;
515
516#ifdef CONFIG_DNOTIFY
517 unsigned long i_dnotify_mask; /* Directory notify events */
518 struct dnotify_struct *i_dnotify; /* for directory notifications */
519#endif
520
Robert Love0eeca282005-07-12 17:06:03 -0400521#ifdef CONFIG_INOTIFY
522 struct list_head inotify_watches; /* watches on this inode */
Ingo Molnard4f9af92006-03-23 03:00:30 -0800523 struct mutex inotify_mutex; /* protects the watches list */
Robert Love0eeca282005-07-12 17:06:03 -0400524#endif
525
Linus Torvalds1da177e2005-04-16 15:20:36 -0700526 unsigned long i_state;
527 unsigned long dirtied_when; /* jiffies of first dirtying */
528
529 unsigned int i_flags;
530
531 atomic_t i_writecount;
532 void *i_security;
533 union {
534 void *generic_ip;
535 } u;
536#ifdef __NEED_I_SIZE_ORDERED
537 seqcount_t i_size_seqcount;
538#endif
539};
540
541/*
542 * NOTE: in a 32bit arch with a preemptable kernel and
543 * an UP compile the i_size_read/write must be atomic
544 * with respect to the local cpu (unlike with preempt disabled),
545 * but they don't need to be atomic with respect to other cpus like in
546 * true SMP (so they need either to either locally disable irq around
547 * the read or for example on x86 they can be still implemented as a
548 * cmpxchg8b without the need of the lock prefix). For SMP compiles
549 * and 64bit archs it makes no difference if preempt is enabled or not.
550 */
551static inline loff_t i_size_read(struct inode *inode)
552{
553#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
554 loff_t i_size;
555 unsigned int seq;
556
557 do {
558 seq = read_seqcount_begin(&inode->i_size_seqcount);
559 i_size = inode->i_size;
560 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
561 return i_size;
562#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
563 loff_t i_size;
564
565 preempt_disable();
566 i_size = inode->i_size;
567 preempt_enable();
568 return i_size;
569#else
570 return inode->i_size;
571#endif
572}
573
574
575static inline void i_size_write(struct inode *inode, loff_t i_size)
576{
577#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
578 write_seqcount_begin(&inode->i_size_seqcount);
579 inode->i_size = i_size;
580 write_seqcount_end(&inode->i_size_seqcount);
581#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
582 preempt_disable();
583 inode->i_size = i_size;
584 preempt_enable();
585#else
586 inode->i_size = i_size;
587#endif
588}
589
590static inline unsigned iminor(struct inode *inode)
591{
592 return MINOR(inode->i_rdev);
593}
594
595static inline unsigned imajor(struct inode *inode)
596{
597 return MAJOR(inode->i_rdev);
598}
599
600extern struct block_device *I_BDEV(struct inode *inode);
601
602struct fown_struct {
603 rwlock_t lock; /* protects pid, uid, euid fields */
604 int pid; /* pid or -pgrp where SIGIO should be sent */
605 uid_t uid, euid; /* uid/euid of process setting the owner */
606 void *security;
607 int signum; /* posix.1b rt signal to be delivered on IO */
608};
609
610/*
611 * Track a single file's readahead state
612 */
613struct file_ra_state {
614 unsigned long start; /* Current window */
615 unsigned long size;
616 unsigned long flags; /* ra flags RA_FLAG_xxx*/
617 unsigned long cache_hit; /* cache hit count*/
618 unsigned long prev_page; /* Cache last read() position */
619 unsigned long ahead_start; /* Ahead window */
620 unsigned long ahead_size;
621 unsigned long ra_pages; /* Maximum readahead window */
622 unsigned long mmap_hit; /* Cache hit stat for mmap accesses */
623 unsigned long mmap_miss; /* Cache miss stat for mmap accesses */
624};
625#define RA_FLAG_MISS 0x01 /* a cache miss occured against this file */
626#define RA_FLAG_INCACHE 0x02 /* file is already in cache */
627
628struct file {
Eric Dumazet2f512012005-10-30 15:02:16 -0800629 /*
630 * fu_list becomes invalid after file_free is called and queued via
631 * fu_rcuhead for RCU freeing
632 */
633 union {
634 struct list_head fu_list;
635 struct rcu_head fu_rcuhead;
636 } f_u;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700637 struct dentry *f_dentry;
638 struct vfsmount *f_vfsmnt;
Arjan van de Ven99ac48f2006-03-28 01:56:41 -0800639 const struct file_operations *f_op;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700640 atomic_t f_count;
641 unsigned int f_flags;
642 mode_t f_mode;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700643 loff_t f_pos;
644 struct fown_struct f_owner;
645 unsigned int f_uid, f_gid;
646 struct file_ra_state f_ra;
647
Linus Torvalds1da177e2005-04-16 15:20:36 -0700648 unsigned long f_version;
649 void *f_security;
650
651 /* needed for tty driver, and maybe others */
652 void *private_data;
653
654#ifdef CONFIG_EPOLL
655 /* Used by fs/eventpoll.c to link all the hooks to this file */
656 struct list_head f_ep_links;
657 spinlock_t f_ep_lock;
658#endif /* #ifdef CONFIG_EPOLL */
659 struct address_space *f_mapping;
660};
661extern spinlock_t files_lock;
662#define file_list_lock() spin_lock(&files_lock);
663#define file_list_unlock() spin_unlock(&files_lock);
664
Nick Piggin095975d2006-01-08 01:02:19 -0800665#define get_file(x) atomic_inc(&(x)->f_count)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700666#define file_count(x) atomic_read(&(x)->f_count)
667
668#define MAX_NON_LFS ((1UL<<31) - 1)
669
670/* Page cache limit. The filesystems should put that into their s_maxbytes
671 limits, otherwise bad things can happen in VM. */
672#if BITS_PER_LONG==32
673#define MAX_LFS_FILESIZE (((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
674#elif BITS_PER_LONG==64
675#define MAX_LFS_FILESIZE 0x7fffffffffffffffUL
676#endif
677
678#define FL_POSIX 1
679#define FL_FLOCK 2
680#define FL_ACCESS 8 /* not trying to lock, just looking */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700681#define FL_LEASE 32 /* lease held on this file */
682#define FL_SLEEP 128 /* A blocking lock */
683
684/*
685 * The POSIX file lock owner is determined by
686 * the "struct files_struct" in the thread group
687 * (or NULL for no owner - BSD locks).
688 *
689 * Lockd stuffs a "host" pointer into this.
690 */
691typedef struct files_struct *fl_owner_t;
692
693struct file_lock_operations {
694 void (*fl_insert)(struct file_lock *); /* lock insertion callback */
695 void (*fl_remove)(struct file_lock *); /* lock removal callback */
696 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
697 void (*fl_release_private)(struct file_lock *);
698};
699
700struct lock_manager_operations {
701 int (*fl_compare_owner)(struct file_lock *, struct file_lock *);
702 void (*fl_notify)(struct file_lock *); /* unblock callback */
703 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
704 void (*fl_release_private)(struct file_lock *);
705 void (*fl_break)(struct file_lock *);
706 int (*fl_mylease)(struct file_lock *, struct file_lock *);
707 int (*fl_change)(struct file_lock **, int);
708};
709
710/* that will die - we need it for nfs_lock_info */
711#include <linux/nfs_fs_i.h>
712
713struct file_lock {
714 struct file_lock *fl_next; /* singly linked list for this inode */
715 struct list_head fl_link; /* doubly linked list of all locks */
716 struct list_head fl_block; /* circular list of blocked processes */
717 fl_owner_t fl_owner;
718 unsigned int fl_pid;
719 wait_queue_head_t fl_wait;
720 struct file *fl_file;
721 unsigned char fl_flags;
722 unsigned char fl_type;
723 loff_t fl_start;
724 loff_t fl_end;
725
726 struct fasync_struct * fl_fasync; /* for lease break notifications */
727 unsigned long fl_break_time; /* for nonblocking lease breaks */
728
729 struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
730 struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
731 union {
732 struct nfs_lock_info nfs_fl;
Trond Myklebust8d0a8a92005-06-22 17:16:32 +0000733 struct nfs4_lock_info nfs4_fl;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700734 } fl_u;
735};
736
737/* The following constant reflects the upper bound of the file/locking space */
738#ifndef OFFSET_MAX
739#define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
740#define OFFSET_MAX INT_LIMIT(loff_t)
741#define OFFT_OFFSET_MAX INT_LIMIT(off_t)
742#endif
743
Linus Torvalds1da177e2005-04-16 15:20:36 -0700744#include <linux/fcntl.h>
745
746extern int fcntl_getlk(struct file *, struct flock __user *);
Peter Staubachc2936212005-07-27 11:45:09 -0700747extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
748 struct flock __user *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700749
750#if BITS_PER_LONG == 32
751extern int fcntl_getlk64(struct file *, struct flock64 __user *);
Peter Staubachc2936212005-07-27 11:45:09 -0700752extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
753 struct flock64 __user *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700754#endif
755
756extern void send_sigio(struct fown_struct *fown, int fd, int band);
757extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
758extern int fcntl_getlease(struct file *filp);
759
760/* fs/locks.c */
761extern void locks_init_lock(struct file_lock *);
762extern void locks_copy_lock(struct file_lock *, struct file_lock *);
763extern void locks_remove_posix(struct file *, fl_owner_t);
764extern void locks_remove_flock(struct file *);
Andy Adamson8dc7c312006-03-20 13:44:26 -0500765extern int posix_test_lock(struct file *, struct file_lock *, struct file_lock *);
Andy Adamson5842add2006-03-26 01:37:26 -0800766extern int posix_lock_file_conf(struct file *, struct file_lock *, struct file_lock *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700767extern int posix_lock_file(struct file *, struct file_lock *);
768extern int posix_lock_file_wait(struct file *, struct file_lock *);
J. Bruce Fields64a318e2006-01-03 09:55:46 +0100769extern int posix_unblock_lock(struct file *, struct file_lock *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700770extern int posix_locks_deadlock(struct file_lock *, struct file_lock *);
771extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
772extern int __break_lease(struct inode *inode, unsigned int flags);
773extern void lease_get_mtime(struct inode *, struct timespec *time);
774extern int setlease(struct file *, long, struct file_lock **);
775extern int lease_modify(struct file_lock **, int);
776extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
777extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
778extern void steal_locks(fl_owner_t from);
779
780struct fasync_struct {
781 int magic;
782 int fa_fd;
783 struct fasync_struct *fa_next; /* singly linked list */
784 struct file *fa_file;
785};
786
787#define FASYNC_MAGIC 0x4601
788
789/* SMP safe fasync helpers: */
790extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
791/* can be called from interrupts */
792extern void kill_fasync(struct fasync_struct **, int, int);
793/* only for net: no internal synchronization */
794extern void __kill_fasync(struct fasync_struct *, int, int);
795
796extern int f_setown(struct file *filp, unsigned long arg, int force);
797extern void f_delown(struct file *filp);
798extern int send_sigurg(struct fown_struct *fown);
799
800/*
801 * Umount options
802 */
803
804#define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
805#define MNT_DETACH 0x00000002 /* Just detach from the tree */
806#define MNT_EXPIRE 0x00000004 /* Mark for expiry */
807
808extern struct list_head super_blocks;
809extern spinlock_t sb_lock;
810
811#define sb_entry(list) list_entry((list), struct super_block, s_list)
812#define S_BIAS (1<<30)
813struct super_block {
814 struct list_head s_list; /* Keep this first */
815 dev_t s_dev; /* search index; _not_ kdev_t */
816 unsigned long s_blocksize;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700817 unsigned char s_blocksize_bits;
818 unsigned char s_dirt;
819 unsigned long long s_maxbytes; /* Max file size */
820 struct file_system_type *s_type;
821 struct super_operations *s_op;
822 struct dquot_operations *dq_op;
823 struct quotactl_ops *s_qcop;
824 struct export_operations *s_export_op;
825 unsigned long s_flags;
826 unsigned long s_magic;
827 struct dentry *s_root;
828 struct rw_semaphore s_umount;
Ingo Molnar7892f2f42006-01-09 15:59:25 -0800829 struct mutex s_lock;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700830 int s_count;
831 int s_syncing;
832 int s_need_sync_fs;
833 atomic_t s_active;
834 void *s_security;
835 struct xattr_handler **s_xattr;
836
837 struct list_head s_inodes; /* all inodes */
838 struct list_head s_dirty; /* dirty inodes */
839 struct list_head s_io; /* parked for writeback */
840 struct hlist_head s_anon; /* anonymous dentries for (nfs) exporting */
841 struct list_head s_files;
842
843 struct block_device *s_bdev;
844 struct list_head s_instances;
845 struct quota_info s_dquot; /* Diskquota specific options */
846
847 int s_frozen;
848 wait_queue_head_t s_wait_unfrozen;
849
850 char s_id[32]; /* Informational name */
851
852 void *s_fs_info; /* Filesystem private info */
853
854 /*
855 * The next field is for VFS *only*. No filesystems have any business
856 * even looking at it. You had been warned.
857 */
Arjan van de Vena11f3a02006-03-23 03:00:33 -0800858 struct mutex s_vfs_rename_mutex; /* Kludge */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700859
860 /* Granuality of c/m/atime in ns.
861 Cannot be worse than a second */
862 u32 s_time_gran;
863};
864
865extern struct timespec current_fs_time(struct super_block *sb);
866
867/*
868 * Snapshotting support.
869 */
870enum {
871 SB_UNFROZEN = 0,
872 SB_FREEZE_WRITE = 1,
873 SB_FREEZE_TRANS = 2,
874};
875
876#define vfs_check_frozen(sb, level) \
877 wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
878
Jens Axboe22e2c502005-06-27 10:55:12 +0200879static inline void get_fs_excl(void)
880{
881 atomic_inc(&current->fs_excl);
882}
883
884static inline void put_fs_excl(void)
885{
886 atomic_dec(&current->fs_excl);
887}
888
889static inline int has_fs_excl(void)
890{
891 return atomic_read(&current->fs_excl);
892}
893
894
Linus Torvalds1da177e2005-04-16 15:20:36 -0700895/*
896 * Superblock locking.
897 */
898static inline void lock_super(struct super_block * sb)
899{
Jens Axboe22e2c502005-06-27 10:55:12 +0200900 get_fs_excl();
Ingo Molnar7892f2f42006-01-09 15:59:25 -0800901 mutex_lock(&sb->s_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700902}
903
904static inline void unlock_super(struct super_block * sb)
905{
Jens Axboe22e2c502005-06-27 10:55:12 +0200906 put_fs_excl();
Ingo Molnar7892f2f42006-01-09 15:59:25 -0800907 mutex_unlock(&sb->s_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700908}
909
910/*
911 * VFS helper functions..
912 */
Christoph Hellwige4543ed2005-11-08 21:35:04 -0800913extern int vfs_permission(struct nameidata *, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700914extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *);
915extern int vfs_mkdir(struct inode *, struct dentry *, int);
916extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t);
917extern int vfs_symlink(struct inode *, struct dentry *, const char *, int);
918extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
919extern int vfs_rmdir(struct inode *, struct dentry *);
920extern int vfs_unlink(struct inode *, struct dentry *);
921extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
922
923/*
924 * VFS dentry helper functions.
925 */
926extern void dentry_unhash(struct dentry *dentry);
927
928/*
Christoph Hellwig8c744fb2005-11-08 21:35:04 -0800929 * VFS file helper functions.
930 */
931extern int file_permission(struct file *, int);
932
933/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700934 * File types
935 *
936 * NOTE! These match bits 12..15 of stat.st_mode
937 * (ie "(i_mode >> 12) & 15").
938 */
939#define DT_UNKNOWN 0
940#define DT_FIFO 1
941#define DT_CHR 2
942#define DT_DIR 4
943#define DT_BLK 6
944#define DT_REG 8
945#define DT_LNK 10
946#define DT_SOCK 12
947#define DT_WHT 14
948
949#define OSYNC_METADATA (1<<0)
950#define OSYNC_DATA (1<<1)
951#define OSYNC_INODE (1<<2)
952int generic_osync_inode(struct inode *, struct address_space *, int);
953
954/*
955 * This is the "filldir" function type, used by readdir() to let
956 * the kernel specify what kind of dirent layout it wants to have.
957 * This allows the kernel to read directories into kernel space or
958 * to have different dirent layouts depending on the binary type.
959 */
960typedef int (*filldir_t)(void *, const char *, int, loff_t, ino_t, unsigned);
961
962struct block_device_operations {
963 int (*open) (struct inode *, struct file *);
964 int (*release) (struct inode *, struct file *);
965 int (*ioctl) (struct inode *, struct file *, unsigned, unsigned long);
Arnd Bergmannbb93e3a2005-06-23 00:10:15 -0700966 long (*unlocked_ioctl) (struct file *, unsigned, unsigned long);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700967 long (*compat_ioctl) (struct file *, unsigned, unsigned long);
Carsten Otte420edbc2005-06-23 22:05:23 -0700968 int (*direct_access) (struct block_device *, sector_t, unsigned long *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700969 int (*media_changed) (struct gendisk *);
970 int (*revalidate_disk) (struct gendisk *);
Christoph Hellwiga885c8c2006-01-08 01:02:50 -0800971 int (*getgeo)(struct block_device *, struct hd_geometry *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700972 struct module *owner;
973};
974
975/*
976 * "descriptor" for what we're up to with a read for sendfile().
977 * This allows us to use the same read code yet
978 * have multiple different users of the data that
979 * we read from a file.
980 *
981 * The simplest case just copies the data to user
982 * mode.
983 */
984typedef struct {
985 size_t written;
986 size_t count;
987 union {
988 char __user * buf;
989 void *data;
990 } arg;
991 int error;
992} read_descriptor_t;
993
994typedef int (*read_actor_t)(read_descriptor_t *, struct page *, unsigned long, unsigned long);
995
996/* These macros are for out of kernel modules to test that
997 * the kernel supports the unlocked_ioctl and compat_ioctl
998 * fields in struct file_operations. */
999#define HAVE_COMPAT_IOCTL 1
1000#define HAVE_UNLOCKED_IOCTL 1
1001
1002/*
1003 * NOTE:
1004 * read, write, poll, fsync, readv, writev, unlocked_ioctl and compat_ioctl
1005 * can be called without the big kernel lock held in all filesystems.
1006 */
1007struct file_operations {
1008 struct module *owner;
1009 loff_t (*llseek) (struct file *, loff_t, int);
1010 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1011 ssize_t (*aio_read) (struct kiocb *, char __user *, size_t, loff_t);
1012 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1013 ssize_t (*aio_write) (struct kiocb *, const char __user *, size_t, loff_t);
1014 int (*readdir) (struct file *, void *, filldir_t);
1015 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1016 int (*ioctl) (struct inode *, struct file *, unsigned int, unsigned long);
1017 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1018 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1019 int (*mmap) (struct file *, struct vm_area_struct *);
1020 int (*open) (struct inode *, struct file *);
1021 int (*flush) (struct file *);
1022 int (*release) (struct inode *, struct file *);
1023 int (*fsync) (struct file *, struct dentry *, int datasync);
1024 int (*aio_fsync) (struct kiocb *, int datasync);
1025 int (*fasync) (int, struct file *, int);
1026 int (*lock) (struct file *, int, struct file_lock *);
1027 ssize_t (*readv) (struct file *, const struct iovec *, unsigned long, loff_t *);
1028 ssize_t (*writev) (struct file *, const struct iovec *, unsigned long, loff_t *);
1029 ssize_t (*sendfile) (struct file *, loff_t *, size_t, read_actor_t, void *);
1030 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1031 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1032 int (*check_flags)(int);
1033 int (*dir_notify)(struct file *filp, unsigned long arg);
1034 int (*flock) (struct file *, int, struct file_lock *);
1035};
1036
1037struct inode_operations {
1038 int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
1039 struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
1040 int (*link) (struct dentry *,struct inode *,struct dentry *);
1041 int (*unlink) (struct inode *,struct dentry *);
1042 int (*symlink) (struct inode *,struct dentry *,const char *);
1043 int (*mkdir) (struct inode *,struct dentry *,int);
1044 int (*rmdir) (struct inode *,struct dentry *);
1045 int (*mknod) (struct inode *,struct dentry *,int,dev_t);
1046 int (*rename) (struct inode *, struct dentry *,
1047 struct inode *, struct dentry *);
1048 int (*readlink) (struct dentry *, char __user *,int);
Linus Torvaldscc314ee2005-08-19 18:02:56 -07001049 void * (*follow_link) (struct dentry *, struct nameidata *);
1050 void (*put_link) (struct dentry *, struct nameidata *, void *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001051 void (*truncate) (struct inode *);
1052 int (*permission) (struct inode *, int, struct nameidata *);
1053 int (*setattr) (struct dentry *, struct iattr *);
1054 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1055 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1056 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1057 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1058 int (*removexattr) (struct dentry *, const char *);
Badari Pulavartyf6b3ec22006-01-06 00:10:38 -08001059 void (*truncate_range)(struct inode *, loff_t, loff_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001060};
1061
1062struct seq_file;
1063
1064extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1065extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1066extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1067 unsigned long, loff_t *);
1068extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1069 unsigned long, loff_t *);
1070
1071/*
1072 * NOTE: write_inode, delete_inode, clear_inode, put_inode can be called
1073 * without the big kernel lock held in all filesystems.
1074 */
1075struct super_operations {
1076 struct inode *(*alloc_inode)(struct super_block *sb);
1077 void (*destroy_inode)(struct inode *);
1078
1079 void (*read_inode) (struct inode *);
1080
1081 void (*dirty_inode) (struct inode *);
1082 int (*write_inode) (struct inode *, int);
1083 void (*put_inode) (struct inode *);
1084 void (*drop_inode) (struct inode *);
1085 void (*delete_inode) (struct inode *);
1086 void (*put_super) (struct super_block *);
1087 void (*write_super) (struct super_block *);
1088 int (*sync_fs)(struct super_block *sb, int wait);
1089 void (*write_super_lockfs) (struct super_block *);
1090 void (*unlockfs) (struct super_block *);
1091 int (*statfs) (struct super_block *, struct kstatfs *);
1092 int (*remount_fs) (struct super_block *, int *, char *);
1093 void (*clear_inode) (struct inode *);
1094 void (*umount_begin) (struct super_block *);
1095
1096 int (*show_options)(struct seq_file *, struct vfsmount *);
Chuck Leverb4629fe2006-03-20 13:44:12 -05001097 int (*show_stats)(struct seq_file *, struct vfsmount *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001098
1099 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1100 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1101};
1102
1103/* Inode state bits. Protected by inode_lock. */
1104#define I_DIRTY_SYNC 1 /* Not dirty enough for O_DATASYNC */
1105#define I_DIRTY_DATASYNC 2 /* Data-related inode changes pending */
1106#define I_DIRTY_PAGES 4 /* Data-related inode changes pending */
1107#define __I_LOCK 3
1108#define I_LOCK (1 << __I_LOCK)
1109#define I_FREEING 16
1110#define I_CLEAR 32
1111#define I_NEW 64
Alexander Viro991114c2005-06-23 00:09:01 -07001112#define I_WILL_FREE 128
Linus Torvalds1da177e2005-04-16 15:20:36 -07001113
1114#define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1115
1116extern void __mark_inode_dirty(struct inode *, int);
1117static inline void mark_inode_dirty(struct inode *inode)
1118{
1119 __mark_inode_dirty(inode, I_DIRTY);
1120}
1121
1122static inline void mark_inode_dirty_sync(struct inode *inode)
1123{
1124 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1125}
1126
Alexey Dobriyan66d21732006-03-23 03:00:51 -08001127static inline void inode_inc_link_count(struct inode *inode)
1128{
1129 inode->i_nlink++;
1130 mark_inode_dirty(inode);
1131}
1132
1133static inline void inode_dec_link_count(struct inode *inode)
1134{
1135 inode->i_nlink--;
1136 mark_inode_dirty(inode);
1137}
1138
Christoph Hellwig869243a2006-01-09 20:52:03 -08001139extern void touch_atime(struct vfsmount *mnt, struct dentry *dentry);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001140static inline void file_accessed(struct file *file)
1141{
1142 if (!(file->f_flags & O_NOATIME))
1143 touch_atime(file->f_vfsmnt, file->f_dentry);
1144}
1145
1146int sync_inode(struct inode *inode, struct writeback_control *wbc);
1147
1148/**
Pavel Pisa4dc3b162005-05-01 08:59:25 -07001149 * struct export_operations - for nfsd to communicate with file systems
1150 * @decode_fh: decode a file handle fragment and return a &struct dentry
1151 * @encode_fh: encode a file handle fragment from a dentry
1152 * @get_name: find the name for a given inode in a given directory
1153 * @get_parent: find the parent of a given directory
1154 * @get_dentry: find a dentry for the inode given a file handle sub-fragment
Randy Dunlap8f2709b2005-11-07 01:01:05 -08001155 * @find_exported_dentry:
1156 * set by the exporting module to a standard helper function.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001157 *
1158 * Description:
1159 * The export_operations structure provides a means for nfsd to communicate
1160 * with a particular exported file system - particularly enabling nfsd and
1161 * the filesystem to co-operate when dealing with file handles.
1162 *
Martin Waitz67be2dd2005-05-01 08:59:26 -07001163 * export_operations contains two basic operation for dealing with file
1164 * handles, decode_fh() and encode_fh(), and allows for some other
1165 * operations to be defined which standard helper routines use to get
1166 * specific information from the filesystem.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001167 *
1168 * nfsd encodes information use to determine which filesystem a filehandle
Martin Waitz67be2dd2005-05-01 08:59:26 -07001169 * applies to in the initial part of the file handle. The remainder, termed
1170 * a file handle fragment, is controlled completely by the filesystem. The
1171 * standard helper routines assume that this fragment will contain one or
1172 * two sub-fragments, one which identifies the file, and one which may be
1173 * used to identify the (a) directory containing the file.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001174 *
1175 * In some situations, nfsd needs to get a dentry which is connected into a
Martin Waitz67be2dd2005-05-01 08:59:26 -07001176 * specific part of the file tree. To allow for this, it passes the
1177 * function acceptable() together with a @context which can be used to see
1178 * if the dentry is acceptable. As there can be multiple dentrys for a
1179 * given file, the filesystem should check each one for acceptability before
1180 * looking for the next. As soon as an acceptable one is found, it should
1181 * be returned.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001182 *
1183 * decode_fh:
Martin Waitz67be2dd2005-05-01 08:59:26 -07001184 * @decode_fh is given a &struct super_block (@sb), a file handle fragment
1185 * (@fh, @fh_len) and an acceptability testing function (@acceptable,
1186 * @context). It should return a &struct dentry which refers to the same
1187 * file that the file handle fragment refers to, and which passes the
1188 * acceptability test. If it cannot, it should return a %NULL pointer if
1189 * the file was found but no acceptable &dentries were available, or a
1190 * %ERR_PTR error code indicating why it couldn't be found (e.g. %ENOENT or
1191 * %ENOMEM).
Linus Torvalds1da177e2005-04-16 15:20:36 -07001192 *
1193 * encode_fh:
Martin Waitz67be2dd2005-05-01 08:59:26 -07001194 * @encode_fh should store in the file handle fragment @fh (using at most
1195 * @max_len bytes) information that can be used by @decode_fh to recover the
1196 * file refered to by the &struct dentry @de. If the @connectable flag is
1197 * set, the encode_fh() should store sufficient information so that a good
1198 * attempt can be made to find not only the file but also it's place in the
1199 * filesystem. This typically means storing a reference to de->d_parent in
1200 * the filehandle fragment. encode_fh() should return the number of bytes
1201 * stored or a negative error code such as %-ENOSPC
Linus Torvalds1da177e2005-04-16 15:20:36 -07001202 *
1203 * get_name:
Martin Waitz67be2dd2005-05-01 08:59:26 -07001204 * @get_name should find a name for the given @child in the given @parent
1205 * directory. The name should be stored in the @name (with the
1206 * understanding that it is already pointing to a a %NAME_MAX+1 sized
1207 * buffer. get_name() should return %0 on success, a negative error code
Jes Sorensen1b1dcc12006-01-09 15:59:24 -08001208 * or error. @get_name will be called without @parent->i_mutex held.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001209 *
1210 * get_parent:
Martin Waitz67be2dd2005-05-01 08:59:26 -07001211 * @get_parent should find the parent directory for the given @child which
1212 * is also a directory. In the event that it cannot be found, or storage
1213 * space cannot be allocated, a %ERR_PTR should be returned.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001214 *
1215 * get_dentry:
Martin Waitz67be2dd2005-05-01 08:59:26 -07001216 * Given a &super_block (@sb) and a pointer to a file-system specific inode
1217 * identifier, possibly an inode number, (@inump) get_dentry() should find
1218 * the identified inode and return a dentry for that inode. Any suitable
1219 * dentry can be returned including, if necessary, a new dentry created with
1220 * d_alloc_root. The caller can then find any other extant dentrys by
1221 * following the d_alias links. If a new dentry was created using
1222 * d_alloc_root, DCACHE_NFSD_DISCONNECTED should be set, and the dentry
1223 * should be d_rehash()ed.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001224 *
Martin Waitz67be2dd2005-05-01 08:59:26 -07001225 * If the inode cannot be found, either a %NULL pointer or an %ERR_PTR code
1226 * can be returned. The @inump will be whatever was passed to
1227 * nfsd_find_fh_dentry() in either the @obj or @parent parameters.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001228 *
1229 * Locking rules:
Jes Sorensen1b1dcc12006-01-09 15:59:24 -08001230 * get_parent is called with child->d_inode->i_mutex down
Martin Waitz67be2dd2005-05-01 08:59:26 -07001231 * get_name is not (which is possibly inconsistent)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001232 */
1233
1234struct export_operations {
1235 struct dentry *(*decode_fh)(struct super_block *sb, __u32 *fh, int fh_len, int fh_type,
1236 int (*acceptable)(void *context, struct dentry *de),
1237 void *context);
1238 int (*encode_fh)(struct dentry *de, __u32 *fh, int *max_len,
1239 int connectable);
1240
1241 /* the following are only called from the filesystem itself */
1242 int (*get_name)(struct dentry *parent, char *name,
1243 struct dentry *child);
1244 struct dentry * (*get_parent)(struct dentry *child);
1245 struct dentry * (*get_dentry)(struct super_block *sb, void *inump);
1246
1247 /* This is set by the exporting module to a standard helper */
1248 struct dentry * (*find_exported_dentry)(
1249 struct super_block *sb, void *obj, void *parent,
1250 int (*acceptable)(void *context, struct dentry *de),
1251 void *context);
1252
1253
1254};
1255
1256extern struct dentry *
1257find_exported_dentry(struct super_block *sb, void *obj, void *parent,
1258 int (*acceptable)(void *context, struct dentry *de),
1259 void *context);
1260
1261struct file_system_type {
1262 const char *name;
1263 int fs_flags;
1264 struct super_block *(*get_sb) (struct file_system_type *, int,
1265 const char *, void *);
1266 void (*kill_sb) (struct super_block *);
1267 struct module *owner;
1268 struct file_system_type * next;
1269 struct list_head fs_supers;
1270};
1271
1272struct super_block *get_sb_bdev(struct file_system_type *fs_type,
1273 int flags, const char *dev_name, void *data,
1274 int (*fill_super)(struct super_block *, void *, int));
1275struct super_block *get_sb_single(struct file_system_type *fs_type,
1276 int flags, void *data,
1277 int (*fill_super)(struct super_block *, void *, int));
1278struct super_block *get_sb_nodev(struct file_system_type *fs_type,
1279 int flags, void *data,
1280 int (*fill_super)(struct super_block *, void *, int));
1281void generic_shutdown_super(struct super_block *sb);
1282void kill_block_super(struct super_block *sb);
1283void kill_anon_super(struct super_block *sb);
1284void kill_litter_super(struct super_block *sb);
1285void deactivate_super(struct super_block *sb);
1286int set_anon_super(struct super_block *s, void *data);
1287struct super_block *sget(struct file_system_type *type,
1288 int (*test)(struct super_block *,void *),
1289 int (*set)(struct super_block *,void *),
1290 void *data);
1291struct super_block *get_sb_pseudo(struct file_system_type *, char *,
1292 struct super_operations *ops, unsigned long);
1293int __put_super(struct super_block *sb);
1294int __put_super_and_need_restart(struct super_block *sb);
1295void unnamed_dev_init(void);
1296
1297/* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1298#define fops_get(fops) \
1299 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1300#define fops_put(fops) \
1301 do { if (fops) module_put((fops)->owner); } while(0)
1302
1303extern int register_filesystem(struct file_system_type *);
1304extern int unregister_filesystem(struct file_system_type *);
1305extern struct vfsmount *kern_mount(struct file_system_type *);
1306extern int may_umount_tree(struct vfsmount *);
1307extern int may_umount(struct vfsmount *);
Ram Paia05964f2005-11-07 17:20:17 -05001308extern void umount_tree(struct vfsmount *, int, struct list_head *);
Ram Paib90fa9a2005-11-07 17:19:50 -05001309extern void release_mounts(struct list_head *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001310extern long do_mount(char *, char *, char *, unsigned long, void *);
Ram Paib90fa9a2005-11-07 17:19:50 -05001311extern struct vfsmount *copy_tree(struct vfsmount *, struct dentry *, int);
1312extern void mnt_set_mountpoint(struct vfsmount *, struct dentry *,
1313 struct vfsmount *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001314
1315extern int vfs_statfs(struct super_block *, struct kstatfs *);
1316
Miklos Szeredif87fd4c2006-01-16 22:14:23 -08001317/* /sys/fs */
1318extern struct subsystem fs_subsys;
1319
Linus Torvalds1da177e2005-04-16 15:20:36 -07001320#define FLOCK_VERIFY_READ 1
1321#define FLOCK_VERIFY_WRITE 2
1322
1323extern int locks_mandatory_locked(struct inode *);
1324extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1325
1326/*
1327 * Candidates for mandatory locking have the setgid bit set
1328 * but no group execute bit - an otherwise meaningless combination.
1329 */
1330#define MANDATORY_LOCK(inode) \
1331 (IS_MANDLOCK(inode) && ((inode)->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID)
1332
1333static inline int locks_verify_locked(struct inode *inode)
1334{
1335 if (MANDATORY_LOCK(inode))
1336 return locks_mandatory_locked(inode);
1337 return 0;
1338}
1339
1340extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1341
1342static inline int locks_verify_truncate(struct inode *inode,
1343 struct file *filp,
1344 loff_t size)
1345{
1346 if (inode->i_flock && MANDATORY_LOCK(inode))
1347 return locks_mandatory_area(
1348 FLOCK_VERIFY_WRITE, inode, filp,
1349 size < inode->i_size ? size : inode->i_size,
1350 (size < inode->i_size ? inode->i_size - size
1351 : size - inode->i_size)
1352 );
1353 return 0;
1354}
1355
1356static inline int break_lease(struct inode *inode, unsigned int mode)
1357{
1358 if (inode->i_flock)
1359 return __break_lease(inode, mode);
1360 return 0;
1361}
1362
1363/* fs/open.c */
1364
NeilBrown4a301312006-01-08 01:02:39 -08001365extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
1366 struct file *filp);
Ulrich Drepper5590ff02006-01-18 17:43:53 -08001367extern long do_sys_open(int fdf, const char __user *filename, int flags,
1368 int mode);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001369extern struct file *filp_open(const char *, int, int);
1370extern struct file * dentry_open(struct dentry *, struct vfsmount *, int);
1371extern int filp_close(struct file *, fl_owner_t id);
1372extern char * getname(const char __user *);
1373
1374/* fs/dcache.c */
1375extern void __init vfs_caches_init_early(void);
1376extern void __init vfs_caches_init(unsigned long);
1377
1378#define __getname() kmem_cache_alloc(names_cachep, SLAB_KERNEL)
1379#define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
1380#ifndef CONFIG_AUDITSYSCALL
1381#define putname(name) __putname(name)
1382#else
1383extern void putname(const char *name);
1384#endif
1385
1386extern int register_blkdev(unsigned int, const char *);
1387extern int unregister_blkdev(unsigned int, const char *);
1388extern struct block_device *bdget(dev_t);
1389extern void bd_set_size(struct block_device *, loff_t size);
1390extern void bd_forget(struct inode *inode);
1391extern void bdput(struct block_device *);
1392extern struct block_device *open_by_devnum(dev_t, unsigned);
1393extern struct file_operations def_blk_fops;
1394extern struct address_space_operations def_blk_aops;
1395extern struct file_operations def_chr_fops;
1396extern struct file_operations bad_sock_fops;
1397extern struct file_operations def_fifo_fops;
1398extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
1399extern int blkdev_ioctl(struct inode *, struct file *, unsigned, unsigned long);
1400extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
1401extern int blkdev_get(struct block_device *, mode_t, unsigned);
1402extern int blkdev_put(struct block_device *);
1403extern int bd_claim(struct block_device *, void *);
1404extern void bd_release(struct block_device *);
Jun'ichi Nomura641dc632006-03-27 01:17:57 -08001405#ifdef CONFIG_SYSFS
1406extern int bd_claim_by_disk(struct block_device *, void *, struct gendisk *);
1407extern void bd_release_from_disk(struct block_device *, struct gendisk *);
1408#else
1409#define bd_claim_by_disk(bdev, holder, disk) bd_claim(bdev, holder)
1410#define bd_release_from_disk(bdev, disk) bd_release(bdev)
1411#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001412
1413/* fs/char_dev.c */
1414extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
1415extern int register_chrdev_region(dev_t, unsigned, const char *);
1416extern int register_chrdev(unsigned int, const char *,
Arjan van de Ven99ac48f2006-03-28 01:56:41 -08001417 const struct file_operations *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001418extern int unregister_chrdev(unsigned int, const char *);
1419extern void unregister_chrdev_region(dev_t, unsigned);
1420extern int chrdev_open(struct inode *, struct file *);
Neil Horman7170be52006-01-14 13:20:38 -08001421extern int get_chrdev_list(char *);
1422extern void *acquire_chrdev_list(void);
1423extern int count_chrdev_list(void);
1424extern void *get_next_chrdev(void *);
1425extern int get_chrdev_info(void *, int *, char **);
1426extern void release_chrdev_list(void *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001427
1428/* fs/block_dev.c */
1429#define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
1430extern const char *__bdevname(dev_t, char *buffer);
1431extern const char *bdevname(struct block_device *bdev, char *buffer);
1432extern struct block_device *lookup_bdev(const char *);
1433extern struct block_device *open_bdev_excl(const char *, int, void *);
1434extern void close_bdev_excl(struct block_device *);
Neil Horman7170be52006-01-14 13:20:38 -08001435extern void *acquire_blkdev_list(void);
1436extern int count_blkdev_list(void);
1437extern void *get_next_blkdev(void *);
1438extern int get_blkdev_info(void *, int *, char **);
1439extern void release_blkdev_list(void *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001440
1441extern void init_special_inode(struct inode *, umode_t, dev_t);
1442
1443/* Invalid inode operations -- fs/bad_inode.c */
1444extern void make_bad_inode(struct inode *);
1445extern int is_bad_inode(struct inode *);
1446
1447extern struct file_operations read_fifo_fops;
1448extern struct file_operations write_fifo_fops;
1449extern struct file_operations rdwr_fifo_fops;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001450
1451extern int fs_may_remount_ro(struct super_block *);
1452
1453/*
1454 * return READ, READA, or WRITE
1455 */
1456#define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
1457
1458/*
1459 * return data direction, READ or WRITE
1460 */
1461#define bio_data_dir(bio) ((bio)->bi_rw & 1)
1462
1463extern int check_disk_change(struct block_device *);
1464extern int invalidate_inodes(struct super_block *);
Christoph Hellwig2ef41632005-05-05 16:15:59 -07001465extern int __invalidate_device(struct block_device *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001466extern int invalidate_partition(struct gendisk *, int);
1467unsigned long invalidate_mapping_pages(struct address_space *mapping,
1468 pgoff_t start, pgoff_t end);
1469unsigned long invalidate_inode_pages(struct address_space *mapping);
1470static inline void invalidate_remote_inode(struct inode *inode)
1471{
1472 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
1473 S_ISLNK(inode->i_mode))
1474 invalidate_inode_pages(inode->i_mapping);
1475}
1476extern int invalidate_inode_pages2(struct address_space *mapping);
1477extern int invalidate_inode_pages2_range(struct address_space *mapping,
1478 pgoff_t start, pgoff_t end);
1479extern int write_inode_now(struct inode *, int);
1480extern int filemap_fdatawrite(struct address_space *);
1481extern int filemap_flush(struct address_space *);
1482extern int filemap_fdatawait(struct address_space *);
1483extern int filemap_write_and_wait(struct address_space *mapping);
1484extern int filemap_write_and_wait_range(struct address_space *mapping,
1485 loff_t lstart, loff_t lend);
Andrew Mortonebcf28e2006-03-24 03:18:04 -08001486extern int wait_on_page_writeback_range(struct address_space *mapping,
1487 pgoff_t start, pgoff_t end);
1488extern int __filemap_fdatawrite_range(struct address_space *mapping,
1489 loff_t start, loff_t end, int sync_mode);
1490
Andrew Morton18e79b42006-03-24 03:18:14 -08001491extern long do_fsync(struct file *file, int datasync);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001492extern void sync_supers(void);
1493extern void sync_filesystems(int wait);
1494extern void emergency_sync(void);
1495extern void emergency_remount(void);
1496extern int do_remount_sb(struct super_block *sb, int flags,
1497 void *data, int force);
1498extern sector_t bmap(struct inode *, sector_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001499extern int notify_change(struct dentry *, struct iattr *);
1500extern int permission(struct inode *, int, struct nameidata *);
1501extern int generic_permission(struct inode *, int,
1502 int (*check_acl)(struct inode *, int));
1503
1504extern int get_write_access(struct inode *);
1505extern int deny_write_access(struct file *);
1506static inline void put_write_access(struct inode * inode)
1507{
1508 atomic_dec(&inode->i_writecount);
1509}
1510static inline void allow_write_access(struct file *file)
1511{
1512 if (file)
1513 atomic_inc(&file->f_dentry->d_inode->i_writecount);
1514}
1515extern int do_pipe(int *);
1516
Ulrich Drepper5590ff02006-01-18 17:43:53 -08001517extern int open_namei(int dfd, const char *, int, int, struct nameidata *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001518extern int may_open(struct nameidata *, int, int);
1519
1520extern int kernel_read(struct file *, unsigned long, char *, unsigned long);
1521extern struct file * open_exec(const char *);
1522
1523/* fs/dcache.c -- generic fs support functions */
1524extern int is_subdir(struct dentry *, struct dentry *);
1525extern ino_t find_inode_number(struct dentry *, struct qstr *);
1526
1527#include <linux/err.h>
1528
1529/* needed for stackable file system support */
1530extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
1531
1532extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
1533
1534extern void inode_init_once(struct inode *);
1535extern void iput(struct inode *);
1536extern struct inode * igrab(struct inode *);
1537extern ino_t iunique(struct super_block *, ino_t);
1538extern int inode_needs_sync(struct inode *inode);
1539extern void generic_delete_inode(struct inode *inode);
Mark Fashehcb2c0232005-07-07 17:56:03 -07001540extern void generic_drop_inode(struct inode *inode);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001541
Anton Altaparmakov88bd5122005-07-13 01:10:44 -07001542extern struct inode *ilookup5_nowait(struct super_block *sb,
1543 unsigned long hashval, int (*test)(struct inode *, void *),
1544 void *data);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001545extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
1546 int (*test)(struct inode *, void *), void *data);
1547extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
1548
1549extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
1550extern struct inode * iget_locked(struct super_block *, unsigned long);
1551extern void unlock_new_inode(struct inode *);
1552
1553static inline struct inode *iget(struct super_block *sb, unsigned long ino)
1554{
1555 struct inode *inode = iget_locked(sb, ino);
1556
1557 if (inode && (inode->i_state & I_NEW)) {
1558 sb->s_op->read_inode(inode);
1559 unlock_new_inode(inode);
1560 }
1561
1562 return inode;
1563}
1564
1565extern void __iget(struct inode * inode);
1566extern void clear_inode(struct inode *);
1567extern void destroy_inode(struct inode *);
1568extern struct inode *new_inode(struct super_block *);
1569extern int remove_suid(struct dentry *);
1570extern void remove_dquot_ref(struct super_block *, int, struct list_head *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001571
1572extern void __insert_inode_hash(struct inode *, unsigned long hashval);
1573extern void remove_inode_hash(struct inode *);
1574static inline void insert_inode_hash(struct inode *inode) {
1575 __insert_inode_hash(inode, inode->i_ino);
1576}
1577
1578extern struct file * get_empty_filp(void);
1579extern void file_move(struct file *f, struct list_head *list);
1580extern void file_kill(struct file *f);
1581struct bio;
1582extern void submit_bio(int, struct bio *);
1583extern int bdev_read_only(struct block_device *);
1584extern int set_blocksize(struct block_device *, int);
1585extern int sb_set_blocksize(struct super_block *, int);
1586extern int sb_min_blocksize(struct super_block *, int);
1587
1588extern int generic_file_mmap(struct file *, struct vm_area_struct *);
1589extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
1590extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
1591extern int file_send_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
1592extern ssize_t generic_file_read(struct file *, char __user *, size_t, loff_t *);
1593int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
1594extern ssize_t generic_file_write(struct file *, const char __user *, size_t, loff_t *);
1595extern ssize_t generic_file_aio_read(struct kiocb *, char __user *, size_t, loff_t);
1596extern ssize_t __generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t *);
1597extern ssize_t generic_file_aio_write(struct kiocb *, const char __user *, size_t, loff_t);
1598extern ssize_t generic_file_aio_write_nolock(struct kiocb *, const struct iovec *,
1599 unsigned long, loff_t *);
1600extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
1601 unsigned long *, loff_t, loff_t *, size_t, size_t);
1602extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
1603 unsigned long, loff_t, loff_t *, size_t, ssize_t);
1604extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
1605extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
1606ssize_t generic_file_write_nolock(struct file *file, const struct iovec *iov,
1607 unsigned long nr_segs, loff_t *ppos);
1608extern ssize_t generic_file_sendfile(struct file *, loff_t *, size_t, read_actor_t, void *);
1609extern void do_generic_mapping_read(struct address_space *mapping,
1610 struct file_ra_state *, struct file *,
1611 loff_t *, read_descriptor_t *, read_actor_t);
1612extern void
1613file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001614extern ssize_t generic_file_readv(struct file *filp, const struct iovec *iov,
1615 unsigned long nr_segs, loff_t *ppos);
1616ssize_t generic_file_writev(struct file *filp, const struct iovec *iov,
1617 unsigned long nr_segs, loff_t *ppos);
1618extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
1619extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
1620extern loff_t remote_llseek(struct file *file, loff_t offset, int origin);
1621extern int generic_file_open(struct inode * inode, struct file * filp);
1622extern int nonseekable_open(struct inode * inode, struct file * filp);
1623
Carsten Otteceffc072005-06-23 22:05:25 -07001624#ifdef CONFIG_FS_XIP
Carsten Otteeb6fe0c2005-06-23 22:05:28 -07001625extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
1626 loff_t *ppos);
Carsten Otteceffc072005-06-23 22:05:25 -07001627extern ssize_t xip_file_sendfile(struct file *in_file, loff_t *ppos,
1628 size_t count, read_actor_t actor,
1629 void *target);
1630extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
Carsten Otteeb6fe0c2005-06-23 22:05:28 -07001631extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
1632 size_t len, loff_t *ppos);
Carsten Otteceffc072005-06-23 22:05:25 -07001633extern int xip_truncate_page(struct address_space *mapping, loff_t from);
Carsten Otte6d791252005-06-23 22:05:26 -07001634#else
1635static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
1636{
1637 return 0;
1638}
Carsten Otteceffc072005-06-23 22:05:25 -07001639#endif
1640
Linus Torvalds1da177e2005-04-16 15:20:36 -07001641static inline void do_generic_file_read(struct file * filp, loff_t *ppos,
1642 read_descriptor_t * desc,
1643 read_actor_t actor)
1644{
1645 do_generic_mapping_read(filp->f_mapping,
1646 &filp->f_ra,
1647 filp,
1648 ppos,
1649 desc,
1650 actor);
1651}
1652
1653ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
1654 struct block_device *bdev, const struct iovec *iov, loff_t offset,
Badari Pulavarty1d8fa7a2006-03-26 01:38:02 -08001655 unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001656 int lock_type);
1657
1658enum {
1659 DIO_LOCKING = 1, /* need locking between buffered and direct access */
1660 DIO_NO_LOCKING, /* bdev; no locking at all between buffered/direct */
1661 DIO_OWN_LOCKING, /* filesystem locks buffered and direct internally */
1662};
1663
1664static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
1665 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
Badari Pulavarty1d8fa7a2006-03-26 01:38:02 -08001666 loff_t offset, unsigned long nr_segs, get_block_t get_block,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001667 dio_iodone_t end_io)
1668{
1669 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
Badari Pulavarty1d8fa7a2006-03-26 01:38:02 -08001670 nr_segs, get_block, end_io, DIO_LOCKING);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001671}
1672
1673static inline ssize_t blockdev_direct_IO_no_locking(int rw, struct kiocb *iocb,
1674 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
Badari Pulavarty1d8fa7a2006-03-26 01:38:02 -08001675 loff_t offset, unsigned long nr_segs, get_block_t get_block,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001676 dio_iodone_t end_io)
1677{
1678 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
Badari Pulavarty1d8fa7a2006-03-26 01:38:02 -08001679 nr_segs, get_block, end_io, DIO_NO_LOCKING);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001680}
1681
1682static inline ssize_t blockdev_direct_IO_own_locking(int rw, struct kiocb *iocb,
1683 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
Badari Pulavarty1d8fa7a2006-03-26 01:38:02 -08001684 loff_t offset, unsigned long nr_segs, get_block_t get_block,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001685 dio_iodone_t end_io)
1686{
1687 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
Badari Pulavarty1d8fa7a2006-03-26 01:38:02 -08001688 nr_segs, get_block, end_io, DIO_OWN_LOCKING);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001689}
1690
1691extern struct file_operations generic_ro_fops;
1692
1693#define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
1694
1695extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
1696extern int vfs_follow_link(struct nameidata *, const char *);
1697extern int page_readlink(struct dentry *, char __user *, int);
Linus Torvaldscc314ee2005-08-19 18:02:56 -07001698extern void *page_follow_link_light(struct dentry *, struct nameidata *);
1699extern void page_put_link(struct dentry *, struct nameidata *, void *);
Kirill Korotaev0adb25d2006-03-11 03:27:13 -08001700extern int __page_symlink(struct inode *inode, const char *symname, int len,
1701 gfp_t gfp_mask);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001702extern int page_symlink(struct inode *inode, const char *symname, int len);
1703extern struct inode_operations page_symlink_inode_operations;
1704extern int generic_readlink(struct dentry *, char __user *, int);
1705extern void generic_fillattr(struct inode *, struct kstat *);
1706extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1707void inode_add_bytes(struct inode *inode, loff_t bytes);
1708void inode_sub_bytes(struct inode *inode, loff_t bytes);
1709loff_t inode_get_bytes(struct inode *inode);
1710void inode_set_bytes(struct inode *inode, loff_t bytes);
1711
1712extern int vfs_readdir(struct file *, filldir_t, void *);
1713
1714extern int vfs_stat(char __user *, struct kstat *);
1715extern int vfs_lstat(char __user *, struct kstat *);
Ulrich Drepper5590ff02006-01-18 17:43:53 -08001716extern int vfs_stat_fd(int dfd, char __user *, struct kstat *);
1717extern int vfs_lstat_fd(int dfd, char __user *, struct kstat *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001718extern int vfs_fstat(unsigned int, struct kstat *);
1719
1720extern int vfs_ioctl(struct file *, unsigned int, unsigned int, unsigned long);
1721
1722extern struct file_system_type *get_fs_type(const char *name);
1723extern struct super_block *get_super(struct block_device *);
1724extern struct super_block *user_get_super(dev_t);
1725extern void drop_super(struct super_block *sb);
1726
1727extern int dcache_dir_open(struct inode *, struct file *);
1728extern int dcache_dir_close(struct inode *, struct file *);
1729extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
1730extern int dcache_readdir(struct file *, void *, filldir_t);
1731extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1732extern int simple_statfs(struct super_block *, struct kstatfs *);
1733extern int simple_link(struct dentry *, struct inode *, struct dentry *);
1734extern int simple_unlink(struct inode *, struct dentry *);
1735extern int simple_rmdir(struct inode *, struct dentry *);
1736extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1737extern int simple_sync_file(struct file *, struct dentry *, int);
1738extern int simple_empty(struct dentry *);
1739extern int simple_readpage(struct file *file, struct page *page);
1740extern int simple_prepare_write(struct file *file, struct page *page,
1741 unsigned offset, unsigned to);
1742extern int simple_commit_write(struct file *file, struct page *page,
1743 unsigned offset, unsigned to);
1744
1745extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
1746extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
1747extern struct file_operations simple_dir_operations;
1748extern struct inode_operations simple_dir_inode_operations;
1749struct tree_descr { char *name; struct file_operations *ops; int mode; };
1750struct dentry *d_alloc_name(struct dentry *, const char *);
1751extern int simple_fill_super(struct super_block *, int, struct tree_descr *);
1752extern int simple_pin_fs(char *name, struct vfsmount **mount, int *count);
1753extern void simple_release_fs(struct vfsmount **mount, int *count);
1754
1755extern ssize_t simple_read_from_buffer(void __user *, size_t, loff_t *, const void *, size_t);
1756
Christoph Lametere965f962006-02-01 03:05:41 -08001757#ifdef CONFIG_MIGRATION
1758extern int buffer_migrate_page(struct page *, struct page *);
1759#else
1760#define buffer_migrate_page NULL
1761#endif
1762
Linus Torvalds1da177e2005-04-16 15:20:36 -07001763extern int inode_change_ok(struct inode *, struct iattr *);
1764extern int __must_check inode_setattr(struct inode *, struct iattr *);
1765
Christoph Hellwig870f4812006-01-09 20:52:01 -08001766extern void file_update_time(struct file *file);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001767
1768static inline ino_t parent_ino(struct dentry *dentry)
1769{
1770 ino_t res;
1771
1772 spin_lock(&dentry->d_lock);
1773 res = dentry->d_parent->d_inode->i_ino;
1774 spin_unlock(&dentry->d_lock);
1775 return res;
1776}
1777
1778/* kernel/fork.c */
1779extern int unshare_files(void);
1780
1781/* Transaction based IO helpers */
1782
1783/*
1784 * An argresp is stored in an allocated page and holds the
1785 * size of the argument or response, along with its content
1786 */
1787struct simple_transaction_argresp {
1788 ssize_t size;
1789 char data[0];
1790};
1791
1792#define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
1793
1794char *simple_transaction_get(struct file *file, const char __user *buf,
1795 size_t size);
1796ssize_t simple_transaction_read(struct file *file, char __user *buf,
1797 size_t size, loff_t *pos);
1798int simple_transaction_release(struct inode *inode, struct file *file);
1799
1800static inline void simple_transaction_set(struct file *file, size_t n)
1801{
1802 struct simple_transaction_argresp *ar = file->private_data;
1803
1804 BUG_ON(n > SIMPLE_TRANSACTION_LIMIT);
1805
1806 /*
1807 * The barrier ensures that ar->size will really remain zero until
1808 * ar->data is ready for reading.
1809 */
1810 smp_mb();
1811 ar->size = n;
1812}
1813
Arnd Bergmannacaefc22005-05-18 14:40:59 +02001814/*
1815 * simple attribute files
1816 *
1817 * These attributes behave similar to those in sysfs:
1818 *
1819 * Writing to an attribute immediately sets a value, an open file can be
1820 * written to multiple times.
1821 *
1822 * Reading from an attribute creates a buffer from the value that might get
1823 * read with multiple read calls. When the attribute has been read
1824 * completely, no further read calls are possible until the file is opened
1825 * again.
1826 *
1827 * All attributes contain a text representation of a numeric value
1828 * that are accessed with the get() and set() functions.
1829 */
1830#define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
1831static int __fops ## _open(struct inode *inode, struct file *file) \
1832{ \
1833 __simple_attr_check_format(__fmt, 0ull); \
1834 return simple_attr_open(inode, file, __get, __set, __fmt); \
1835} \
1836static struct file_operations __fops = { \
1837 .owner = THIS_MODULE, \
1838 .open = __fops ## _open, \
1839 .release = simple_attr_close, \
1840 .read = simple_attr_read, \
1841 .write = simple_attr_write, \
1842};
1843
1844static inline void __attribute__((format(printf, 1, 2)))
1845__simple_attr_check_format(const char *fmt, ...)
1846{
1847 /* don't do anything, just let the compiler check the arguments; */
1848}
1849
1850int simple_attr_open(struct inode *inode, struct file *file,
1851 u64 (*get)(void *), void (*set)(void *, u64),
1852 const char *fmt);
1853int simple_attr_close(struct inode *inode, struct file *file);
1854ssize_t simple_attr_read(struct file *file, char __user *buf,
1855 size_t len, loff_t *ppos);
1856ssize_t simple_attr_write(struct file *file, const char __user *buf,
1857 size_t len, loff_t *ppos);
1858
1859
Linus Torvalds1da177e2005-04-16 15:20:36 -07001860#ifdef CONFIG_SECURITY
1861static inline char *alloc_secdata(void)
1862{
1863 return (char *)get_zeroed_page(GFP_KERNEL);
1864}
1865
1866static inline void free_secdata(void *secdata)
1867{
1868 free_page((unsigned long)secdata);
1869}
1870#else
1871static inline char *alloc_secdata(void)
1872{
1873 return (char *)1;
1874}
1875
1876static inline void free_secdata(void *secdata)
1877{ }
1878#endif /* CONFIG_SECURITY */
1879
1880#endif /* __KERNEL__ */
1881#endif /* _LINUX_FS_H */