blob: ecc8c2c3d8ca8382d5b121c7219d9f73b33baba4 [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
Linus Torvalds1da177e2005-04-16 15:20:36 -07009#include <linux/limits.h>
10#include <linux/ioctl.h>
11
12/*
13 * It's silly to have NR_OPEN bigger than NR_FILE, but you can change
14 * the file limit at runtime and only root can increase the per-process
15 * nr_file rlimit, so it's safe to set up a ridiculously high absolute
16 * upper limit on files-per-process.
17 *
18 * Some programs (notably those using select()) may have to be
19 * recompiled to take full advantage of the new limits..
20 */
21
22/* Fixed constants first: */
23#undef NR_OPEN
24#define NR_OPEN (1024*1024) /* Absolute upper limit on fd num */
25#define INR_OPEN 1024 /* Initial setting for nfile rlimits */
26
27#define BLOCK_SIZE_BITS 10
28#define BLOCK_SIZE (1<<BLOCK_SIZE_BITS)
29
30/* And dynamically-tunable limits and defaults: */
31struct files_stat_struct {
32 int nr_files; /* read only */
33 int nr_free_files; /* read only */
34 int max_files; /* tunable */
35};
36extern struct files_stat_struct files_stat;
Dipankar Sarma529bf6b2006-03-07 21:55:35 -080037extern int get_max_files(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -070038
39struct inodes_stat_t {
40 int nr_inodes;
41 int nr_unused;
42 int dummy[5];
43};
44extern struct inodes_stat_t inodes_stat;
45
46extern int leases_enable, lease_break_time;
47
48#ifdef CONFIG_DNOTIFY
49extern int dir_notify_enable;
50#endif
51
52#define NR_FILE 8192 /* this can well be larger on a larger system */
53
54#define MAY_EXEC 1
55#define MAY_WRITE 2
56#define MAY_READ 4
57#define MAY_APPEND 8
58
59#define FMODE_READ 1
60#define FMODE_WRITE 2
61
62/* Internal kernel extensions */
63#define FMODE_LSEEK 4
64#define FMODE_PREAD 8
65#define FMODE_PWRITE FMODE_PREAD /* These go hand in hand */
66
Oleg Drokinb5005312006-03-25 03:07:01 -080067/* File is being opened for execution. Primary users of this flag are
68 distributed filesystems that can use it to achieve correct ETXTBUSY
69 behavior for cross-node execution/opening_for_writing of files */
70#define FMODE_EXEC 16
71
Linus Torvalds1da177e2005-04-16 15:20:36 -070072#define RW_MASK 1
73#define RWA_MASK 2
74#define READ 0
75#define WRITE 1
76#define READA 2 /* read-ahead - don't block if no resources */
Jan Karaa7662232005-09-06 15:19:10 -070077#define SWRITE 3 /* for ll_rw_block() - wait for buffer lock */
Linus Torvalds1da177e2005-04-16 15:20:36 -070078#define SPECIAL 4 /* For non-blockdevice requests in request queue */
79#define READ_SYNC (READ | (1 << BIO_RW_SYNC))
80#define WRITE_SYNC (WRITE | (1 << BIO_RW_SYNC))
81#define WRITE_BARRIER ((1 << BIO_RW) | (1 << BIO_RW_BARRIER))
82
83#define SEL_IN 1
84#define SEL_OUT 2
85#define SEL_EX 4
86
87/* public flags for file_system_type */
88#define FS_REQUIRES_DEV 1
89#define FS_BINARY_MOUNTDATA 2
90#define FS_REVAL_DOT 16384 /* Check the paths ".", ".." for staleness */
91#define FS_ODD_RENAME 32768 /* Temporary stuff; will go away as soon
92 * as nfs_rename() will be cleaned up
93 */
94/*
95 * These are the fs-independent mount-flags: up to 32 flags are supported
96 */
97#define MS_RDONLY 1 /* Mount read-only */
98#define MS_NOSUID 2 /* Ignore suid and sgid bits */
99#define MS_NODEV 4 /* Disallow access to device special files */
100#define MS_NOEXEC 8 /* Disallow program execution */
101#define MS_SYNCHRONOUS 16 /* Writes are synced at once */
102#define MS_REMOUNT 32 /* Alter flags of a mounted FS */
103#define MS_MANDLOCK 64 /* Allow mandatory locks on an FS */
104#define MS_DIRSYNC 128 /* Directory modifications are synchronous */
105#define MS_NOATIME 1024 /* Do not update access times. */
106#define MS_NODIRATIME 2048 /* Do not update directory access times */
107#define MS_BIND 4096
108#define MS_MOVE 8192
109#define MS_REC 16384
Theodore Ts'o9b04c992006-03-24 03:15:10 -0800110#define MS_VERBOSE 32768 /* War is peace. Verbosity is silence.
111 MS_VERBOSE is deprecated. */
112#define MS_SILENT 32768
Miklos Szeredibf066c72006-01-08 01:03:19 -0800113#define MS_POSIXACL (1<<16) /* VFS does not apply the umask */
Ram Pai9676f0c2005-11-07 17:21:20 -0500114#define MS_UNBINDABLE (1<<17) /* change to unbindable */
Ram Pai07b20882005-11-07 17:19:07 -0500115#define MS_PRIVATE (1<<18) /* change to private */
Ram Paia58b0eb2005-11-07 17:20:48 -0500116#define MS_SLAVE (1<<19) /* change to slave */
Ram Pai03e06e62005-11-07 17:19:33 -0500117#define MS_SHARED (1<<20) /* change to shared */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700118#define MS_ACTIVE (1<<30)
119#define MS_NOUSER (1<<31)
120
121/*
122 * Superblock flags that can be altered by MS_REMOUNT
123 */
Christoph Hellwigfc33a7b2006-01-09 20:52:17 -0800124#define MS_RMT_MASK (MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700125
126/*
127 * Old magic mount flag and mask
128 */
129#define MS_MGC_VAL 0xC0ED0000
130#define MS_MGC_MSK 0xffff0000
131
132/* Inode flags - they have nothing to superblock flags now */
133
134#define S_SYNC 1 /* Writes are synced at once */
135#define S_NOATIME 2 /* Do not update access times */
136#define S_APPEND 4 /* Append-only file */
137#define S_IMMUTABLE 8 /* Immutable file */
138#define S_DEAD 16 /* removed, but still open directory */
139#define S_NOQUOTA 32 /* Inode is not counted to quota */
140#define S_DIRSYNC 64 /* Directory modifications are synchronous */
141#define S_NOCMTIME 128 /* Do not update file c/mtime */
142#define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
143#define S_PRIVATE 512 /* Inode is fs-internal */
144
145/*
146 * Note that nosuid etc flags are inode-specific: setting some file-system
147 * flags just means all the inodes inherit those flags by default. It might be
148 * possible to override it selectively if you really wanted to with some
149 * ioctl() that is not currently implemented.
150 *
151 * Exception: MS_RDONLY is always applied to the entire file system.
152 *
153 * Unfortunately, it is possible to change a filesystems flags with it mounted
154 * with files in use. This means that all of the inodes will not have their
155 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
156 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
157 */
158#define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))
159
160#define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
161#define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
162 ((inode)->i_flags & S_SYNC))
163#define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
164 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
165#define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
166
167#define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
168#define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
169#define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700170#define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
171
172#define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
173#define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
174#define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
175#define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
176
177/* the read-only stuff doesn't really belong here, but any other place is
178 probably as bad and I don't want to create yet another include file. */
179
180#define BLKROSET _IO(0x12,93) /* set device read-only (0 = read-write) */
181#define BLKROGET _IO(0x12,94) /* get read-only status (0 = read_write) */
182#define BLKRRPART _IO(0x12,95) /* re-read partition table */
183#define BLKGETSIZE _IO(0x12,96) /* return device size /512 (long *arg) */
184#define BLKFLSBUF _IO(0x12,97) /* flush buffer cache */
185#define BLKRASET _IO(0x12,98) /* set read ahead for block device */
186#define BLKRAGET _IO(0x12,99) /* get current read ahead setting */
187#define BLKFRASET _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
188#define BLKFRAGET _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
189#define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
190#define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
191#define BLKSSZGET _IO(0x12,104)/* get block device sector size */
192#if 0
193#define BLKPG _IO(0x12,105)/* See blkpg.h */
194
195/* Some people are morons. Do not use sizeof! */
196
197#define BLKELVGET _IOR(0x12,106,size_t)/* elevator get */
198#define BLKELVSET _IOW(0x12,107,size_t)/* elevator set */
199/* This was here just to show that the number is taken -
200 probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
201#endif
202/* A jump here: 108-111 have been used for various private purposes. */
203#define BLKBSZGET _IOR(0x12,112,size_t)
204#define BLKBSZSET _IOW(0x12,113,size_t)
205#define BLKGETSIZE64 _IOR(0x12,114,size_t) /* return device size in bytes (u64 *arg) */
Jens Axboe2056a782006-03-23 20:00:26 +0100206#define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup)
207#define BLKTRACESTART _IO(0x12,116)
208#define BLKTRACESTOP _IO(0x12,117)
209#define BLKTRACETEARDOWN _IO(0x12,118)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700210
211#define BMAP_IOCTL 1 /* obsolete - kept for compatibility */
212#define FIBMAP _IO(0x00,1) /* bmap access */
213#define FIGETBSZ _IO(0x00,2) /* get the block size used for bmap */
214
Andrew Mortone46e4902006-05-22 22:35:24 -0700215#define SYNC_FILE_RANGE_WAIT_BEFORE 1
216#define SYNC_FILE_RANGE_WRITE 2
217#define SYNC_FILE_RANGE_WAIT_AFTER 4
218
Linus Torvalds1da177e2005-04-16 15:20:36 -0700219#ifdef __KERNEL__
220
221#include <linux/linkage.h>
222#include <linux/wait.h>
223#include <linux/types.h>
224#include <linux/kdev_t.h>
225#include <linux/dcache.h>
226#include <linux/stat.h>
227#include <linux/cache.h>
228#include <linux/kobject.h>
229#include <linux/list.h>
230#include <linux/radix-tree.h>
231#include <linux/prio_tree.h>
232#include <linux/init.h>
Jens Axboe22e2c502005-06-27 10:55:12 +0200233#include <linux/sched.h>
Jes Sorensen1b1dcc12006-01-09 15:59:24 -0800234#include <linux/mutex.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700235
236#include <asm/atomic.h>
237#include <asm/semaphore.h>
238#include <asm/byteorder.h>
239
Christoph Hellwiga885c8c2006-01-08 01:02:50 -0800240struct hd_geometry;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700241struct iovec;
242struct nameidata;
Christoph Hellwig92198f72005-06-23 22:00:59 -0700243struct kiocb;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700244struct pipe_inode_info;
245struct poll_table_struct;
246struct kstatfs;
247struct vm_area_struct;
248struct vfsmount;
249
Linus Torvalds1da177e2005-04-16 15:20:36 -0700250extern void __init inode_init(unsigned long);
251extern void __init inode_init_early(void);
252extern void __init mnt_init(unsigned long);
253extern void __init files_init(unsigned long);
254
255struct buffer_head;
256typedef int (get_block_t)(struct inode *inode, sector_t iblock,
257 struct buffer_head *bh_result, int create);
Christoph Hellwig92198f72005-06-23 22:00:59 -0700258typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700259 ssize_t bytes, void *private);
260
261/*
262 * Attribute flags. These should be or-ed together to figure out what
263 * has been changed!
264 */
265#define ATTR_MODE 1
266#define ATTR_UID 2
267#define ATTR_GID 4
268#define ATTR_SIZE 8
269#define ATTR_ATIME 16
270#define ATTR_MTIME 32
271#define ATTR_CTIME 64
272#define ATTR_ATIME_SET 128
273#define ATTR_MTIME_SET 256
274#define ATTR_FORCE 512 /* Not a change, but a change it */
275#define ATTR_ATTR_FLAG 1024
276#define ATTR_KILL_SUID 2048
277#define ATTR_KILL_SGID 4096
Miklos Szeredicc4e69d2005-11-07 00:59:49 -0800278#define ATTR_FILE 8192
Linus Torvalds1da177e2005-04-16 15:20:36 -0700279
280/*
281 * This is the Inode Attributes structure, used for notify_change(). It
282 * uses the above definitions as flags, to know which values have changed.
283 * Also, in this manner, a Filesystem can look at only the values it cares
284 * about. Basically, these are the attributes that the VFS layer can
285 * request to change from the FS layer.
286 *
287 * Derek Atkins <warlord@MIT.EDU> 94-10-20
288 */
289struct iattr {
290 unsigned int ia_valid;
291 umode_t ia_mode;
292 uid_t ia_uid;
293 gid_t ia_gid;
294 loff_t ia_size;
295 struct timespec ia_atime;
296 struct timespec ia_mtime;
297 struct timespec ia_ctime;
Miklos Szeredicc4e69d2005-11-07 00:59:49 -0800298
299 /*
300 * Not an attribute, but an auxilary info for filesystems wanting to
301 * implement an ftruncate() like method. NOTE: filesystem should
302 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
303 */
304 struct file *ia_file;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700305};
306
307/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700308 * Includes for diskquotas.
309 */
310#include <linux/quota.h>
311
Zach Brown994fc28c2005-12-15 14:28:17 -0800312/**
313 * enum positive_aop_returns - aop return codes with specific semantics
314 *
315 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
316 * completed, that the page is still locked, and
317 * should be considered active. The VM uses this hint
318 * to return the page to the active list -- it won't
319 * be a candidate for writeback again in the near
320 * future. Other callers must be careful to unlock
321 * the page if they get this return. Returned by
322 * writepage();
323 *
324 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
325 * unlocked it and the page might have been truncated.
326 * The caller should back up to acquiring a new page and
327 * trying again. The aop will be taking reasonable
328 * precautions not to livelock. If the caller held a page
329 * reference, it should drop it before retrying. Returned
330 * by readpage(), prepare_write(), and commit_write().
331 *
332 * address_space_operation functions return these large constants to indicate
333 * special semantics to the caller. These are much larger than the bytes in a
334 * page to allow for functions that return the number of bytes operated on in a
335 * given page.
336 */
337
338enum positive_aop_returns {
339 AOP_WRITEPAGE_ACTIVATE = 0x80000,
340 AOP_TRUNCATED_PAGE = 0x80001,
341};
342
Linus Torvalds1da177e2005-04-16 15:20:36 -0700343/*
344 * oh the beauties of C type declarations.
345 */
346struct page;
347struct address_space;
348struct writeback_control;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700349
350struct address_space_operations {
351 int (*writepage)(struct page *page, struct writeback_control *wbc);
352 int (*readpage)(struct file *, struct page *);
NeilBrown3978d712006-03-26 01:37:17 -0800353 void (*sync_page)(struct page *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700354
355 /* Write back some dirty pages from this mapping. */
356 int (*writepages)(struct address_space *, struct writeback_control *);
357
Andrew Morton4741c9f2006-03-24 03:18:11 -0800358 /* Set a page dirty. Return true if this dirtied it */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700359 int (*set_page_dirty)(struct page *page);
360
361 int (*readpages)(struct file *filp, struct address_space *mapping,
362 struct list_head *pages, unsigned nr_pages);
363
364 /*
365 * ext3 requires that a successful prepare_write() call be followed
366 * by a commit_write() call - they must be balanced
367 */
368 int (*prepare_write)(struct file *, struct page *, unsigned, unsigned);
369 int (*commit_write)(struct file *, struct page *, unsigned, unsigned);
370 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
371 sector_t (*bmap)(struct address_space *, sector_t);
NeilBrown2ff28e22006-03-26 01:37:18 -0800372 void (*invalidatepage) (struct page *, unsigned long);
Al Viro27496a82005-10-21 03:20:48 -0400373 int (*releasepage) (struct page *, gfp_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700374 ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
375 loff_t offset, unsigned long nr_segs);
Carsten Otteceffc072005-06-23 22:05:25 -0700376 struct page* (*get_xip_page)(struct address_space *, sector_t,
377 int);
Christoph Lametere965f962006-02-01 03:05:41 -0800378 /* migrate the contents of a page to the specified target */
379 int (*migratepage) (struct page *, struct page *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700380};
381
382struct backing_dev_info;
383struct address_space {
384 struct inode *host; /* owner: inode, block_device */
385 struct radix_tree_root page_tree; /* radix tree of all pages */
386 rwlock_t tree_lock; /* and rwlock protecting it */
387 unsigned int i_mmap_writable;/* count VM_SHARED mappings */
388 struct prio_tree_root i_mmap; /* tree of private and shared mappings */
389 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
390 spinlock_t i_mmap_lock; /* protect tree, count, list */
391 unsigned int truncate_count; /* Cover race condition with truncate */
392 unsigned long nrpages; /* number of total pages */
393 pgoff_t writeback_index;/* writeback starts here */
394 struct address_space_operations *a_ops; /* methods */
395 unsigned long flags; /* error bits/gfp mask */
396 struct backing_dev_info *backing_dev_info; /* device readahead, etc */
397 spinlock_t private_lock; /* for use by the address_space */
398 struct list_head private_list; /* ditto */
399 struct address_space *assoc_mapping; /* ditto */
400} __attribute__((aligned(sizeof(long))));
401 /*
402 * On most architectures that alignment is already the case; but
403 * must be enforced here for CRIS, to let the least signficant bit
404 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
405 */
406
407struct block_device {
408 dev_t bd_dev; /* not a kdev_t - it's a search key */
409 struct inode * bd_inode; /* will die */
410 int bd_openers;
Arjan van de Venc039e312006-03-23 03:00:28 -0800411 struct mutex bd_mutex; /* open/close mutex */
412 struct mutex bd_mount_mutex; /* mount mutex */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700413 struct list_head bd_inodes;
414 void * bd_holder;
415 int bd_holders;
Jun'ichi Nomura641dc632006-03-27 01:17:57 -0800416#ifdef CONFIG_SYSFS
417 struct list_head bd_holder_list;
418#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700419 struct block_device * bd_contains;
420 unsigned bd_block_size;
421 struct hd_struct * bd_part;
422 /* number of times partitions within this device have been opened. */
423 unsigned bd_part_count;
424 int bd_invalidated;
425 struct gendisk * bd_disk;
426 struct list_head bd_list;
427 struct backing_dev_info *bd_inode_backing_dev_info;
428 /*
429 * Private data. You must have bd_claim'ed the block_device
430 * to use this. NOTE: bd_claim allows an owner to claim
431 * the same device multiple times, the owner must take special
432 * care to not mess up bd_private for that case.
433 */
434 unsigned long bd_private;
435};
436
437/*
438 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
439 * radix trees
440 */
441#define PAGECACHE_TAG_DIRTY 0
442#define PAGECACHE_TAG_WRITEBACK 1
443
444int mapping_tagged(struct address_space *mapping, int tag);
445
446/*
447 * Might pages of this file be mapped into userspace?
448 */
449static inline int mapping_mapped(struct address_space *mapping)
450{
451 return !prio_tree_empty(&mapping->i_mmap) ||
452 !list_empty(&mapping->i_mmap_nonlinear);
453}
454
455/*
456 * Might pages of this file have been modified in userspace?
457 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
458 * marks vma as VM_SHARED if it is shared, and the file was opened for
459 * writing i.e. vma may be mprotected writable even if now readonly.
460 */
461static inline int mapping_writably_mapped(struct address_space *mapping)
462{
463 return mapping->i_mmap_writable != 0;
464}
465
466/*
467 * Use sequence counter to get consistent i_size on 32-bit processors.
468 */
469#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
470#include <linux/seqlock.h>
471#define __NEED_I_SIZE_ORDERED
472#define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
473#else
474#define i_size_ordered_init(inode) do { } while (0)
475#endif
476
477struct inode {
478 struct hlist_node i_hash;
479 struct list_head i_list;
480 struct list_head i_sb_list;
481 struct list_head i_dentry;
482 unsigned long i_ino;
483 atomic_t i_count;
484 umode_t i_mode;
485 unsigned int i_nlink;
486 uid_t i_uid;
487 gid_t i_gid;
488 dev_t i_rdev;
489 loff_t i_size;
490 struct timespec i_atime;
491 struct timespec i_mtime;
492 struct timespec i_ctime;
493 unsigned int i_blkbits;
494 unsigned long i_blksize;
495 unsigned long i_version;
Takashi Satoa0f62ac2006-03-26 01:37:52 -0800496 blkcnt_t i_blocks;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700497 unsigned short i_bytes;
498 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
Jes Sorensen1b1dcc12006-01-09 15:59:24 -0800499 struct mutex i_mutex;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700500 struct rw_semaphore i_alloc_sem;
501 struct inode_operations *i_op;
Arjan van de Ven99ac48f2006-03-28 01:56:41 -0800502 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700503 struct super_block *i_sb;
504 struct file_lock *i_flock;
505 struct address_space *i_mapping;
506 struct address_space i_data;
507#ifdef CONFIG_QUOTA
508 struct dquot *i_dquot[MAXQUOTAS];
509#endif
510 /* These three should probably be a union */
511 struct list_head i_devices;
512 struct pipe_inode_info *i_pipe;
513 struct block_device *i_bdev;
514 struct cdev *i_cdev;
515 int i_cindex;
516
517 __u32 i_generation;
518
519#ifdef CONFIG_DNOTIFY
520 unsigned long i_dnotify_mask; /* Directory notify events */
521 struct dnotify_struct *i_dnotify; /* for directory notifications */
522#endif
523
Robert Love0eeca282005-07-12 17:06:03 -0400524#ifdef CONFIG_INOTIFY
525 struct list_head inotify_watches; /* watches on this inode */
Ingo Molnard4f9af92006-03-23 03:00:30 -0800526 struct mutex inotify_mutex; /* protects the watches list */
Robert Love0eeca282005-07-12 17:06:03 -0400527#endif
528
Linus Torvalds1da177e2005-04-16 15:20:36 -0700529 unsigned long i_state;
530 unsigned long dirtied_when; /* jiffies of first dirtying */
531
532 unsigned int i_flags;
533
534 atomic_t i_writecount;
535 void *i_security;
536 union {
537 void *generic_ip;
538 } u;
539#ifdef __NEED_I_SIZE_ORDERED
540 seqcount_t i_size_seqcount;
541#endif
542};
543
544/*
545 * NOTE: in a 32bit arch with a preemptable kernel and
546 * an UP compile the i_size_read/write must be atomic
547 * with respect to the local cpu (unlike with preempt disabled),
548 * but they don't need to be atomic with respect to other cpus like in
549 * true SMP (so they need either to either locally disable irq around
550 * the read or for example on x86 they can be still implemented as a
551 * cmpxchg8b without the need of the lock prefix). For SMP compiles
552 * and 64bit archs it makes no difference if preempt is enabled or not.
553 */
554static inline loff_t i_size_read(struct inode *inode)
555{
556#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
557 loff_t i_size;
558 unsigned int seq;
559
560 do {
561 seq = read_seqcount_begin(&inode->i_size_seqcount);
562 i_size = inode->i_size;
563 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
564 return i_size;
565#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
566 loff_t i_size;
567
568 preempt_disable();
569 i_size = inode->i_size;
570 preempt_enable();
571 return i_size;
572#else
573 return inode->i_size;
574#endif
575}
576
577
578static inline void i_size_write(struct inode *inode, loff_t i_size)
579{
580#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
581 write_seqcount_begin(&inode->i_size_seqcount);
582 inode->i_size = i_size;
583 write_seqcount_end(&inode->i_size_seqcount);
584#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
585 preempt_disable();
586 inode->i_size = i_size;
587 preempt_enable();
588#else
589 inode->i_size = i_size;
590#endif
591}
592
593static inline unsigned iminor(struct inode *inode)
594{
595 return MINOR(inode->i_rdev);
596}
597
598static inline unsigned imajor(struct inode *inode)
599{
600 return MAJOR(inode->i_rdev);
601}
602
603extern struct block_device *I_BDEV(struct inode *inode);
604
605struct fown_struct {
606 rwlock_t lock; /* protects pid, uid, euid fields */
607 int pid; /* pid or -pgrp where SIGIO should be sent */
608 uid_t uid, euid; /* uid/euid of process setting the owner */
609 void *security;
610 int signum; /* posix.1b rt signal to be delivered on IO */
611};
612
613/*
614 * Track a single file's readahead state
615 */
616struct file_ra_state {
617 unsigned long start; /* Current window */
618 unsigned long size;
619 unsigned long flags; /* ra flags RA_FLAG_xxx*/
620 unsigned long cache_hit; /* cache hit count*/
621 unsigned long prev_page; /* Cache last read() position */
622 unsigned long ahead_start; /* Ahead window */
623 unsigned long ahead_size;
624 unsigned long ra_pages; /* Maximum readahead window */
625 unsigned long mmap_hit; /* Cache hit stat for mmap accesses */
626 unsigned long mmap_miss; /* Cache miss stat for mmap accesses */
627};
628#define RA_FLAG_MISS 0x01 /* a cache miss occured against this file */
629#define RA_FLAG_INCACHE 0x02 /* file is already in cache */
630
631struct file {
Eric Dumazet2f512012005-10-30 15:02:16 -0800632 /*
633 * fu_list becomes invalid after file_free is called and queued via
634 * fu_rcuhead for RCU freeing
635 */
636 union {
637 struct list_head fu_list;
638 struct rcu_head fu_rcuhead;
639 } f_u;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700640 struct dentry *f_dentry;
641 struct vfsmount *f_vfsmnt;
Arjan van de Ven99ac48f2006-03-28 01:56:41 -0800642 const struct file_operations *f_op;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700643 atomic_t f_count;
644 unsigned int f_flags;
645 mode_t f_mode;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700646 loff_t f_pos;
647 struct fown_struct f_owner;
648 unsigned int f_uid, f_gid;
649 struct file_ra_state f_ra;
650
Linus Torvalds1da177e2005-04-16 15:20:36 -0700651 unsigned long f_version;
652 void *f_security;
653
654 /* needed for tty driver, and maybe others */
655 void *private_data;
656
657#ifdef CONFIG_EPOLL
658 /* Used by fs/eventpoll.c to link all the hooks to this file */
659 struct list_head f_ep_links;
660 spinlock_t f_ep_lock;
661#endif /* #ifdef CONFIG_EPOLL */
662 struct address_space *f_mapping;
663};
664extern spinlock_t files_lock;
665#define file_list_lock() spin_lock(&files_lock);
666#define file_list_unlock() spin_unlock(&files_lock);
667
Nick Piggin095975d2006-01-08 01:02:19 -0800668#define get_file(x) atomic_inc(&(x)->f_count)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700669#define file_count(x) atomic_read(&(x)->f_count)
670
671#define MAX_NON_LFS ((1UL<<31) - 1)
672
673/* Page cache limit. The filesystems should put that into their s_maxbytes
674 limits, otherwise bad things can happen in VM. */
675#if BITS_PER_LONG==32
676#define MAX_LFS_FILESIZE (((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
677#elif BITS_PER_LONG==64
678#define MAX_LFS_FILESIZE 0x7fffffffffffffffUL
679#endif
680
681#define FL_POSIX 1
682#define FL_FLOCK 2
683#define FL_ACCESS 8 /* not trying to lock, just looking */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700684#define FL_LEASE 32 /* lease held on this file */
685#define FL_SLEEP 128 /* A blocking lock */
686
687/*
688 * The POSIX file lock owner is determined by
689 * the "struct files_struct" in the thread group
690 * (or NULL for no owner - BSD locks).
691 *
692 * Lockd stuffs a "host" pointer into this.
693 */
694typedef struct files_struct *fl_owner_t;
695
696struct file_lock_operations {
697 void (*fl_insert)(struct file_lock *); /* lock insertion callback */
698 void (*fl_remove)(struct file_lock *); /* lock removal callback */
699 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
700 void (*fl_release_private)(struct file_lock *);
701};
702
703struct lock_manager_operations {
704 int (*fl_compare_owner)(struct file_lock *, struct file_lock *);
705 void (*fl_notify)(struct file_lock *); /* unblock callback */
706 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
707 void (*fl_release_private)(struct file_lock *);
708 void (*fl_break)(struct file_lock *);
709 int (*fl_mylease)(struct file_lock *, struct file_lock *);
710 int (*fl_change)(struct file_lock **, int);
711};
712
713/* that will die - we need it for nfs_lock_info */
714#include <linux/nfs_fs_i.h>
715
716struct file_lock {
717 struct file_lock *fl_next; /* singly linked list for this inode */
718 struct list_head fl_link; /* doubly linked list of all locks */
719 struct list_head fl_block; /* circular list of blocked processes */
720 fl_owner_t fl_owner;
721 unsigned int fl_pid;
722 wait_queue_head_t fl_wait;
723 struct file *fl_file;
724 unsigned char fl_flags;
725 unsigned char fl_type;
726 loff_t fl_start;
727 loff_t fl_end;
728
729 struct fasync_struct * fl_fasync; /* for lease break notifications */
730 unsigned long fl_break_time; /* for nonblocking lease breaks */
731
732 struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
733 struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
734 union {
735 struct nfs_lock_info nfs_fl;
Trond Myklebust8d0a8a92005-06-22 17:16:32 +0000736 struct nfs4_lock_info nfs4_fl;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700737 } fl_u;
738};
739
740/* The following constant reflects the upper bound of the file/locking space */
741#ifndef OFFSET_MAX
742#define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
743#define OFFSET_MAX INT_LIMIT(loff_t)
744#define OFFT_OFFSET_MAX INT_LIMIT(off_t)
745#endif
746
Linus Torvalds1da177e2005-04-16 15:20:36 -0700747#include <linux/fcntl.h>
748
749extern int fcntl_getlk(struct file *, struct flock __user *);
Peter Staubachc2936212005-07-27 11:45:09 -0700750extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
751 struct flock __user *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700752
753#if BITS_PER_LONG == 32
754extern int fcntl_getlk64(struct file *, struct flock64 __user *);
Peter Staubachc2936212005-07-27 11:45:09 -0700755extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
756 struct flock64 __user *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700757#endif
758
759extern void send_sigio(struct fown_struct *fown, int fd, int band);
760extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
761extern int fcntl_getlease(struct file *filp);
762
Andrew Mortonf79e2ab2006-03-31 02:30:42 -0800763/* fs/sync.c */
Andrew Mortonf79e2ab2006-03-31 02:30:42 -0800764extern int do_sync_file_range(struct file *file, loff_t offset, loff_t endbyte,
Andrew Morton5246d052006-04-10 22:53:57 -0700765 unsigned int flags);
Andrew Mortonf79e2ab2006-03-31 02:30:42 -0800766
Linus Torvalds1da177e2005-04-16 15:20:36 -0700767/* fs/locks.c */
768extern void locks_init_lock(struct file_lock *);
769extern void locks_copy_lock(struct file_lock *, struct file_lock *);
770extern void locks_remove_posix(struct file *, fl_owner_t);
771extern void locks_remove_flock(struct file *);
Andy Adamson8dc7c312006-03-20 13:44:26 -0500772extern int posix_test_lock(struct file *, struct file_lock *, struct file_lock *);
Andy Adamson5842add2006-03-26 01:37:26 -0800773extern int posix_lock_file_conf(struct file *, struct file_lock *, struct file_lock *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700774extern int posix_lock_file(struct file *, struct file_lock *);
775extern int posix_lock_file_wait(struct file *, struct file_lock *);
J. Bruce Fields64a318e2006-01-03 09:55:46 +0100776extern int posix_unblock_lock(struct file *, struct file_lock *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700777extern int posix_locks_deadlock(struct file_lock *, struct file_lock *);
778extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
779extern int __break_lease(struct inode *inode, unsigned int flags);
780extern void lease_get_mtime(struct inode *, struct timespec *time);
781extern int setlease(struct file *, long, struct file_lock **);
782extern int lease_modify(struct file_lock **, int);
783extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
784extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
785extern void steal_locks(fl_owner_t from);
786
787struct fasync_struct {
788 int magic;
789 int fa_fd;
790 struct fasync_struct *fa_next; /* singly linked list */
791 struct file *fa_file;
792};
793
794#define FASYNC_MAGIC 0x4601
795
796/* SMP safe fasync helpers: */
797extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
798/* can be called from interrupts */
799extern void kill_fasync(struct fasync_struct **, int, int);
800/* only for net: no internal synchronization */
801extern void __kill_fasync(struct fasync_struct *, int, int);
802
803extern int f_setown(struct file *filp, unsigned long arg, int force);
804extern void f_delown(struct file *filp);
805extern int send_sigurg(struct fown_struct *fown);
806
807/*
808 * Umount options
809 */
810
811#define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
812#define MNT_DETACH 0x00000002 /* Just detach from the tree */
813#define MNT_EXPIRE 0x00000004 /* Mark for expiry */
814
815extern struct list_head super_blocks;
816extern spinlock_t sb_lock;
817
818#define sb_entry(list) list_entry((list), struct super_block, s_list)
819#define S_BIAS (1<<30)
820struct super_block {
821 struct list_head s_list; /* Keep this first */
822 dev_t s_dev; /* search index; _not_ kdev_t */
823 unsigned long s_blocksize;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700824 unsigned char s_blocksize_bits;
825 unsigned char s_dirt;
826 unsigned long long s_maxbytes; /* Max file size */
827 struct file_system_type *s_type;
828 struct super_operations *s_op;
829 struct dquot_operations *dq_op;
830 struct quotactl_ops *s_qcop;
831 struct export_operations *s_export_op;
832 unsigned long s_flags;
833 unsigned long s_magic;
834 struct dentry *s_root;
835 struct rw_semaphore s_umount;
Ingo Molnar7892f2f42006-01-09 15:59:25 -0800836 struct mutex s_lock;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700837 int s_count;
838 int s_syncing;
839 int s_need_sync_fs;
840 atomic_t s_active;
841 void *s_security;
842 struct xattr_handler **s_xattr;
843
844 struct list_head s_inodes; /* all inodes */
845 struct list_head s_dirty; /* dirty inodes */
846 struct list_head s_io; /* parked for writeback */
847 struct hlist_head s_anon; /* anonymous dentries for (nfs) exporting */
848 struct list_head s_files;
849
850 struct block_device *s_bdev;
851 struct list_head s_instances;
852 struct quota_info s_dquot; /* Diskquota specific options */
853
854 int s_frozen;
855 wait_queue_head_t s_wait_unfrozen;
856
857 char s_id[32]; /* Informational name */
858
859 void *s_fs_info; /* Filesystem private info */
860
861 /*
862 * The next field is for VFS *only*. No filesystems have any business
863 * even looking at it. You had been warned.
864 */
Arjan van de Vena11f3a02006-03-23 03:00:33 -0800865 struct mutex s_vfs_rename_mutex; /* Kludge */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700866
Kalin KOZHUHAROV8ba8e952006-04-01 01:41:22 +0200867 /* Granularity of c/m/atime in ns.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700868 Cannot be worse than a second */
869 u32 s_time_gran;
870};
871
872extern struct timespec current_fs_time(struct super_block *sb);
873
874/*
875 * Snapshotting support.
876 */
877enum {
878 SB_UNFROZEN = 0,
879 SB_FREEZE_WRITE = 1,
880 SB_FREEZE_TRANS = 2,
881};
882
883#define vfs_check_frozen(sb, level) \
884 wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
885
Jens Axboe22e2c502005-06-27 10:55:12 +0200886static inline void get_fs_excl(void)
887{
888 atomic_inc(&current->fs_excl);
889}
890
891static inline void put_fs_excl(void)
892{
893 atomic_dec(&current->fs_excl);
894}
895
896static inline int has_fs_excl(void)
897{
898 return atomic_read(&current->fs_excl);
899}
900
901
Linus Torvalds1da177e2005-04-16 15:20:36 -0700902/*
903 * Superblock locking.
904 */
905static inline void lock_super(struct super_block * sb)
906{
Jens Axboe22e2c502005-06-27 10:55:12 +0200907 get_fs_excl();
Ingo Molnar7892f2f42006-01-09 15:59:25 -0800908 mutex_lock(&sb->s_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700909}
910
911static inline void unlock_super(struct super_block * sb)
912{
Jens Axboe22e2c502005-06-27 10:55:12 +0200913 put_fs_excl();
Ingo Molnar7892f2f42006-01-09 15:59:25 -0800914 mutex_unlock(&sb->s_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700915}
916
917/*
918 * VFS helper functions..
919 */
Christoph Hellwige4543ed2005-11-08 21:35:04 -0800920extern int vfs_permission(struct nameidata *, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700921extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *);
922extern int vfs_mkdir(struct inode *, struct dentry *, int);
923extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t);
924extern int vfs_symlink(struct inode *, struct dentry *, const char *, int);
925extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
926extern int vfs_rmdir(struct inode *, struct dentry *);
927extern int vfs_unlink(struct inode *, struct dentry *);
928extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
929
930/*
931 * VFS dentry helper functions.
932 */
933extern void dentry_unhash(struct dentry *dentry);
934
935/*
Christoph Hellwig8c744fb2005-11-08 21:35:04 -0800936 * VFS file helper functions.
937 */
938extern int file_permission(struct file *, int);
939
940/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700941 * File types
942 *
943 * NOTE! These match bits 12..15 of stat.st_mode
944 * (ie "(i_mode >> 12) & 15").
945 */
946#define DT_UNKNOWN 0
947#define DT_FIFO 1
948#define DT_CHR 2
949#define DT_DIR 4
950#define DT_BLK 6
951#define DT_REG 8
952#define DT_LNK 10
953#define DT_SOCK 12
954#define DT_WHT 14
955
956#define OSYNC_METADATA (1<<0)
957#define OSYNC_DATA (1<<1)
958#define OSYNC_INODE (1<<2)
959int generic_osync_inode(struct inode *, struct address_space *, int);
960
961/*
962 * This is the "filldir" function type, used by readdir() to let
963 * the kernel specify what kind of dirent layout it wants to have.
964 * This allows the kernel to read directories into kernel space or
965 * to have different dirent layouts depending on the binary type.
966 */
967typedef int (*filldir_t)(void *, const char *, int, loff_t, ino_t, unsigned);
968
969struct block_device_operations {
970 int (*open) (struct inode *, struct file *);
971 int (*release) (struct inode *, struct file *);
972 int (*ioctl) (struct inode *, struct file *, unsigned, unsigned long);
Arnd Bergmannbb93e3a2005-06-23 00:10:15 -0700973 long (*unlocked_ioctl) (struct file *, unsigned, unsigned long);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700974 long (*compat_ioctl) (struct file *, unsigned, unsigned long);
Carsten Otte420edbc2005-06-23 22:05:23 -0700975 int (*direct_access) (struct block_device *, sector_t, unsigned long *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700976 int (*media_changed) (struct gendisk *);
977 int (*revalidate_disk) (struct gendisk *);
Christoph Hellwiga885c8c2006-01-08 01:02:50 -0800978 int (*getgeo)(struct block_device *, struct hd_geometry *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700979 struct module *owner;
980};
981
982/*
983 * "descriptor" for what we're up to with a read for sendfile().
984 * This allows us to use the same read code yet
985 * have multiple different users of the data that
986 * we read from a file.
987 *
988 * The simplest case just copies the data to user
989 * mode.
990 */
991typedef struct {
992 size_t written;
993 size_t count;
994 union {
995 char __user * buf;
996 void *data;
997 } arg;
998 int error;
999} read_descriptor_t;
1000
1001typedef int (*read_actor_t)(read_descriptor_t *, struct page *, unsigned long, unsigned long);
1002
1003/* These macros are for out of kernel modules to test that
1004 * the kernel supports the unlocked_ioctl and compat_ioctl
1005 * fields in struct file_operations. */
1006#define HAVE_COMPAT_IOCTL 1
1007#define HAVE_UNLOCKED_IOCTL 1
1008
1009/*
1010 * NOTE:
1011 * read, write, poll, fsync, readv, writev, unlocked_ioctl and compat_ioctl
1012 * can be called without the big kernel lock held in all filesystems.
1013 */
1014struct file_operations {
1015 struct module *owner;
1016 loff_t (*llseek) (struct file *, loff_t, int);
1017 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1018 ssize_t (*aio_read) (struct kiocb *, char __user *, size_t, loff_t);
1019 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1020 ssize_t (*aio_write) (struct kiocb *, const char __user *, size_t, loff_t);
1021 int (*readdir) (struct file *, void *, filldir_t);
1022 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1023 int (*ioctl) (struct inode *, struct file *, unsigned int, unsigned long);
1024 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1025 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1026 int (*mmap) (struct file *, struct vm_area_struct *);
1027 int (*open) (struct inode *, struct file *);
1028 int (*flush) (struct file *);
1029 int (*release) (struct inode *, struct file *);
1030 int (*fsync) (struct file *, struct dentry *, int datasync);
1031 int (*aio_fsync) (struct kiocb *, int datasync);
1032 int (*fasync) (int, struct file *, int);
1033 int (*lock) (struct file *, int, struct file_lock *);
1034 ssize_t (*readv) (struct file *, const struct iovec *, unsigned long, loff_t *);
1035 ssize_t (*writev) (struct file *, const struct iovec *, unsigned long, loff_t *);
1036 ssize_t (*sendfile) (struct file *, loff_t *, size_t, read_actor_t, void *);
1037 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1038 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1039 int (*check_flags)(int);
1040 int (*dir_notify)(struct file *filp, unsigned long arg);
1041 int (*flock) (struct file *, int, struct file_lock *);
Jens Axboecbb7e572006-04-11 14:57:50 +02001042 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1043 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001044};
1045
1046struct inode_operations {
1047 int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
1048 struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
1049 int (*link) (struct dentry *,struct inode *,struct dentry *);
1050 int (*unlink) (struct inode *,struct dentry *);
1051 int (*symlink) (struct inode *,struct dentry *,const char *);
1052 int (*mkdir) (struct inode *,struct dentry *,int);
1053 int (*rmdir) (struct inode *,struct dentry *);
1054 int (*mknod) (struct inode *,struct dentry *,int,dev_t);
1055 int (*rename) (struct inode *, struct dentry *,
1056 struct inode *, struct dentry *);
1057 int (*readlink) (struct dentry *, char __user *,int);
Linus Torvaldscc314ee2005-08-19 18:02:56 -07001058 void * (*follow_link) (struct dentry *, struct nameidata *);
1059 void (*put_link) (struct dentry *, struct nameidata *, void *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001060 void (*truncate) (struct inode *);
1061 int (*permission) (struct inode *, int, struct nameidata *);
1062 int (*setattr) (struct dentry *, struct iattr *);
1063 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1064 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1065 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1066 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1067 int (*removexattr) (struct dentry *, const char *);
Badari Pulavartyf6b3ec22006-01-06 00:10:38 -08001068 void (*truncate_range)(struct inode *, loff_t, loff_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001069};
1070
1071struct seq_file;
1072
1073extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1074extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1075extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1076 unsigned long, loff_t *);
1077extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1078 unsigned long, loff_t *);
1079
1080/*
1081 * NOTE: write_inode, delete_inode, clear_inode, put_inode can be called
1082 * without the big kernel lock held in all filesystems.
1083 */
1084struct super_operations {
1085 struct inode *(*alloc_inode)(struct super_block *sb);
1086 void (*destroy_inode)(struct inode *);
1087
1088 void (*read_inode) (struct inode *);
1089
1090 void (*dirty_inode) (struct inode *);
1091 int (*write_inode) (struct inode *, int);
1092 void (*put_inode) (struct inode *);
1093 void (*drop_inode) (struct inode *);
1094 void (*delete_inode) (struct inode *);
1095 void (*put_super) (struct super_block *);
1096 void (*write_super) (struct super_block *);
1097 int (*sync_fs)(struct super_block *sb, int wait);
1098 void (*write_super_lockfs) (struct super_block *);
1099 void (*unlockfs) (struct super_block *);
1100 int (*statfs) (struct super_block *, struct kstatfs *);
1101 int (*remount_fs) (struct super_block *, int *, char *);
1102 void (*clear_inode) (struct inode *);
1103 void (*umount_begin) (struct super_block *);
1104
1105 int (*show_options)(struct seq_file *, struct vfsmount *);
Chuck Leverb4629fe2006-03-20 13:44:12 -05001106 int (*show_stats)(struct seq_file *, struct vfsmount *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001107
1108 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1109 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1110};
1111
1112/* Inode state bits. Protected by inode_lock. */
1113#define I_DIRTY_SYNC 1 /* Not dirty enough for O_DATASYNC */
1114#define I_DIRTY_DATASYNC 2 /* Data-related inode changes pending */
1115#define I_DIRTY_PAGES 4 /* Data-related inode changes pending */
1116#define __I_LOCK 3
1117#define I_LOCK (1 << __I_LOCK)
1118#define I_FREEING 16
1119#define I_CLEAR 32
1120#define I_NEW 64
Alexander Viro991114c2005-06-23 00:09:01 -07001121#define I_WILL_FREE 128
Linus Torvalds1da177e2005-04-16 15:20:36 -07001122
1123#define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1124
1125extern void __mark_inode_dirty(struct inode *, int);
1126static inline void mark_inode_dirty(struct inode *inode)
1127{
1128 __mark_inode_dirty(inode, I_DIRTY);
1129}
1130
1131static inline void mark_inode_dirty_sync(struct inode *inode)
1132{
1133 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1134}
1135
Alexey Dobriyan66d21732006-03-23 03:00:51 -08001136static inline void inode_inc_link_count(struct inode *inode)
1137{
1138 inode->i_nlink++;
1139 mark_inode_dirty(inode);
1140}
1141
1142static inline void inode_dec_link_count(struct inode *inode)
1143{
1144 inode->i_nlink--;
1145 mark_inode_dirty(inode);
1146}
1147
Christoph Hellwig869243a2006-01-09 20:52:03 -08001148extern void touch_atime(struct vfsmount *mnt, struct dentry *dentry);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001149static inline void file_accessed(struct file *file)
1150{
1151 if (!(file->f_flags & O_NOATIME))
1152 touch_atime(file->f_vfsmnt, file->f_dentry);
1153}
1154
1155int sync_inode(struct inode *inode, struct writeback_control *wbc);
1156
1157/**
Pavel Pisa4dc3b162005-05-01 08:59:25 -07001158 * struct export_operations - for nfsd to communicate with file systems
1159 * @decode_fh: decode a file handle fragment and return a &struct dentry
1160 * @encode_fh: encode a file handle fragment from a dentry
1161 * @get_name: find the name for a given inode in a given directory
1162 * @get_parent: find the parent of a given directory
1163 * @get_dentry: find a dentry for the inode given a file handle sub-fragment
Randy Dunlap8f2709b2005-11-07 01:01:05 -08001164 * @find_exported_dentry:
1165 * set by the exporting module to a standard helper function.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001166 *
1167 * Description:
1168 * The export_operations structure provides a means for nfsd to communicate
1169 * with a particular exported file system - particularly enabling nfsd and
1170 * the filesystem to co-operate when dealing with file handles.
1171 *
Martin Waitz67be2dd2005-05-01 08:59:26 -07001172 * export_operations contains two basic operation for dealing with file
1173 * handles, decode_fh() and encode_fh(), and allows for some other
1174 * operations to be defined which standard helper routines use to get
1175 * specific information from the filesystem.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001176 *
1177 * nfsd encodes information use to determine which filesystem a filehandle
Martin Waitz67be2dd2005-05-01 08:59:26 -07001178 * applies to in the initial part of the file handle. The remainder, termed
1179 * a file handle fragment, is controlled completely by the filesystem. The
1180 * standard helper routines assume that this fragment will contain one or
1181 * two sub-fragments, one which identifies the file, and one which may be
1182 * used to identify the (a) directory containing the file.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001183 *
1184 * In some situations, nfsd needs to get a dentry which is connected into a
Martin Waitz67be2dd2005-05-01 08:59:26 -07001185 * specific part of the file tree. To allow for this, it passes the
1186 * function acceptable() together with a @context which can be used to see
1187 * if the dentry is acceptable. As there can be multiple dentrys for a
1188 * given file, the filesystem should check each one for acceptability before
1189 * looking for the next. As soon as an acceptable one is found, it should
1190 * be returned.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001191 *
1192 * decode_fh:
Martin Waitz67be2dd2005-05-01 08:59:26 -07001193 * @decode_fh is given a &struct super_block (@sb), a file handle fragment
1194 * (@fh, @fh_len) and an acceptability testing function (@acceptable,
1195 * @context). It should return a &struct dentry which refers to the same
1196 * file that the file handle fragment refers to, and which passes the
1197 * acceptability test. If it cannot, it should return a %NULL pointer if
1198 * the file was found but no acceptable &dentries were available, or a
1199 * %ERR_PTR error code indicating why it couldn't be found (e.g. %ENOENT or
1200 * %ENOMEM).
Linus Torvalds1da177e2005-04-16 15:20:36 -07001201 *
1202 * encode_fh:
Martin Waitz67be2dd2005-05-01 08:59:26 -07001203 * @encode_fh should store in the file handle fragment @fh (using at most
1204 * @max_len bytes) information that can be used by @decode_fh to recover the
1205 * file refered to by the &struct dentry @de. If the @connectable flag is
1206 * set, the encode_fh() should store sufficient information so that a good
1207 * attempt can be made to find not only the file but also it's place in the
1208 * filesystem. This typically means storing a reference to de->d_parent in
1209 * the filehandle fragment. encode_fh() should return the number of bytes
1210 * stored or a negative error code such as %-ENOSPC
Linus Torvalds1da177e2005-04-16 15:20:36 -07001211 *
1212 * get_name:
Martin Waitz67be2dd2005-05-01 08:59:26 -07001213 * @get_name should find a name for the given @child in the given @parent
1214 * directory. The name should be stored in the @name (with the
1215 * understanding that it is already pointing to a a %NAME_MAX+1 sized
1216 * buffer. get_name() should return %0 on success, a negative error code
Jes Sorensen1b1dcc12006-01-09 15:59:24 -08001217 * or error. @get_name will be called without @parent->i_mutex held.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001218 *
1219 * get_parent:
Martin Waitz67be2dd2005-05-01 08:59:26 -07001220 * @get_parent should find the parent directory for the given @child which
1221 * is also a directory. In the event that it cannot be found, or storage
1222 * space cannot be allocated, a %ERR_PTR should be returned.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001223 *
1224 * get_dentry:
Martin Waitz67be2dd2005-05-01 08:59:26 -07001225 * Given a &super_block (@sb) and a pointer to a file-system specific inode
1226 * identifier, possibly an inode number, (@inump) get_dentry() should find
1227 * the identified inode and return a dentry for that inode. Any suitable
1228 * dentry can be returned including, if necessary, a new dentry created with
1229 * d_alloc_root. The caller can then find any other extant dentrys by
1230 * following the d_alias links. If a new dentry was created using
1231 * d_alloc_root, DCACHE_NFSD_DISCONNECTED should be set, and the dentry
1232 * should be d_rehash()ed.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001233 *
Martin Waitz67be2dd2005-05-01 08:59:26 -07001234 * If the inode cannot be found, either a %NULL pointer or an %ERR_PTR code
1235 * can be returned. The @inump will be whatever was passed to
1236 * nfsd_find_fh_dentry() in either the @obj or @parent parameters.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001237 *
1238 * Locking rules:
Jes Sorensen1b1dcc12006-01-09 15:59:24 -08001239 * get_parent is called with child->d_inode->i_mutex down
Martin Waitz67be2dd2005-05-01 08:59:26 -07001240 * get_name is not (which is possibly inconsistent)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001241 */
1242
1243struct export_operations {
1244 struct dentry *(*decode_fh)(struct super_block *sb, __u32 *fh, int fh_len, int fh_type,
1245 int (*acceptable)(void *context, struct dentry *de),
1246 void *context);
1247 int (*encode_fh)(struct dentry *de, __u32 *fh, int *max_len,
1248 int connectable);
1249
1250 /* the following are only called from the filesystem itself */
1251 int (*get_name)(struct dentry *parent, char *name,
1252 struct dentry *child);
1253 struct dentry * (*get_parent)(struct dentry *child);
1254 struct dentry * (*get_dentry)(struct super_block *sb, void *inump);
1255
1256 /* This is set by the exporting module to a standard helper */
1257 struct dentry * (*find_exported_dentry)(
1258 struct super_block *sb, void *obj, void *parent,
1259 int (*acceptable)(void *context, struct dentry *de),
1260 void *context);
1261
1262
1263};
1264
1265extern struct dentry *
1266find_exported_dentry(struct super_block *sb, void *obj, void *parent,
1267 int (*acceptable)(void *context, struct dentry *de),
1268 void *context);
1269
1270struct file_system_type {
1271 const char *name;
1272 int fs_flags;
1273 struct super_block *(*get_sb) (struct file_system_type *, int,
1274 const char *, void *);
1275 void (*kill_sb) (struct super_block *);
1276 struct module *owner;
1277 struct file_system_type * next;
1278 struct list_head fs_supers;
1279};
1280
1281struct super_block *get_sb_bdev(struct file_system_type *fs_type,
1282 int flags, const char *dev_name, void *data,
1283 int (*fill_super)(struct super_block *, void *, int));
1284struct super_block *get_sb_single(struct file_system_type *fs_type,
1285 int flags, void *data,
1286 int (*fill_super)(struct super_block *, void *, int));
1287struct super_block *get_sb_nodev(struct file_system_type *fs_type,
1288 int flags, void *data,
1289 int (*fill_super)(struct super_block *, void *, int));
1290void generic_shutdown_super(struct super_block *sb);
1291void kill_block_super(struct super_block *sb);
1292void kill_anon_super(struct super_block *sb);
1293void kill_litter_super(struct super_block *sb);
1294void deactivate_super(struct super_block *sb);
1295int set_anon_super(struct super_block *s, void *data);
1296struct super_block *sget(struct file_system_type *type,
1297 int (*test)(struct super_block *,void *),
1298 int (*set)(struct super_block *,void *),
1299 void *data);
1300struct super_block *get_sb_pseudo(struct file_system_type *, char *,
1301 struct super_operations *ops, unsigned long);
1302int __put_super(struct super_block *sb);
1303int __put_super_and_need_restart(struct super_block *sb);
1304void unnamed_dev_init(void);
1305
1306/* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1307#define fops_get(fops) \
1308 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1309#define fops_put(fops) \
1310 do { if (fops) module_put((fops)->owner); } while(0)
1311
1312extern int register_filesystem(struct file_system_type *);
1313extern int unregister_filesystem(struct file_system_type *);
1314extern struct vfsmount *kern_mount(struct file_system_type *);
1315extern int may_umount_tree(struct vfsmount *);
1316extern int may_umount(struct vfsmount *);
Ram Paia05964f2005-11-07 17:20:17 -05001317extern void umount_tree(struct vfsmount *, int, struct list_head *);
Ram Paib90fa9a2005-11-07 17:19:50 -05001318extern void release_mounts(struct list_head *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001319extern long do_mount(char *, char *, char *, unsigned long, void *);
Ram Paib90fa9a2005-11-07 17:19:50 -05001320extern struct vfsmount *copy_tree(struct vfsmount *, struct dentry *, int);
1321extern void mnt_set_mountpoint(struct vfsmount *, struct dentry *,
1322 struct vfsmount *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001323
1324extern int vfs_statfs(struct super_block *, struct kstatfs *);
1325
Miklos Szeredif87fd4c2006-01-16 22:14:23 -08001326/* /sys/fs */
1327extern struct subsystem fs_subsys;
1328
Linus Torvalds1da177e2005-04-16 15:20:36 -07001329#define FLOCK_VERIFY_READ 1
1330#define FLOCK_VERIFY_WRITE 2
1331
1332extern int locks_mandatory_locked(struct inode *);
1333extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1334
1335/*
1336 * Candidates for mandatory locking have the setgid bit set
1337 * but no group execute bit - an otherwise meaningless combination.
1338 */
1339#define MANDATORY_LOCK(inode) \
1340 (IS_MANDLOCK(inode) && ((inode)->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID)
1341
1342static inline int locks_verify_locked(struct inode *inode)
1343{
1344 if (MANDATORY_LOCK(inode))
1345 return locks_mandatory_locked(inode);
1346 return 0;
1347}
1348
1349extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1350
1351static inline int locks_verify_truncate(struct inode *inode,
1352 struct file *filp,
1353 loff_t size)
1354{
1355 if (inode->i_flock && MANDATORY_LOCK(inode))
1356 return locks_mandatory_area(
1357 FLOCK_VERIFY_WRITE, inode, filp,
1358 size < inode->i_size ? size : inode->i_size,
1359 (size < inode->i_size ? inode->i_size - size
1360 : size - inode->i_size)
1361 );
1362 return 0;
1363}
1364
1365static inline int break_lease(struct inode *inode, unsigned int mode)
1366{
1367 if (inode->i_flock)
1368 return __break_lease(inode, mode);
1369 return 0;
1370}
1371
1372/* fs/open.c */
1373
NeilBrown4a301312006-01-08 01:02:39 -08001374extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
1375 struct file *filp);
Ulrich Drepper5590ff02006-01-18 17:43:53 -08001376extern long do_sys_open(int fdf, const char __user *filename, int flags,
1377 int mode);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001378extern struct file *filp_open(const char *, int, int);
1379extern struct file * dentry_open(struct dentry *, struct vfsmount *, int);
1380extern int filp_close(struct file *, fl_owner_t id);
1381extern char * getname(const char __user *);
1382
1383/* fs/dcache.c */
1384extern void __init vfs_caches_init_early(void);
1385extern void __init vfs_caches_init(unsigned long);
1386
1387#define __getname() kmem_cache_alloc(names_cachep, SLAB_KERNEL)
1388#define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
1389#ifndef CONFIG_AUDITSYSCALL
1390#define putname(name) __putname(name)
1391#else
1392extern void putname(const char *name);
1393#endif
1394
1395extern int register_blkdev(unsigned int, const char *);
1396extern int unregister_blkdev(unsigned int, const char *);
1397extern struct block_device *bdget(dev_t);
1398extern void bd_set_size(struct block_device *, loff_t size);
1399extern void bd_forget(struct inode *inode);
1400extern void bdput(struct block_device *);
1401extern struct block_device *open_by_devnum(dev_t, unsigned);
Arjan van de Ven4b6f5d22006-03-28 01:56:42 -08001402extern const struct file_operations def_blk_fops;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001403extern struct address_space_operations def_blk_aops;
Arjan van de Ven4b6f5d22006-03-28 01:56:42 -08001404extern const struct file_operations def_chr_fops;
1405extern const struct file_operations bad_sock_fops;
1406extern const struct file_operations def_fifo_fops;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001407extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
1408extern int blkdev_ioctl(struct inode *, struct file *, unsigned, unsigned long);
1409extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
1410extern int blkdev_get(struct block_device *, mode_t, unsigned);
1411extern int blkdev_put(struct block_device *);
1412extern int bd_claim(struct block_device *, void *);
1413extern void bd_release(struct block_device *);
Jun'ichi Nomura641dc632006-03-27 01:17:57 -08001414#ifdef CONFIG_SYSFS
1415extern int bd_claim_by_disk(struct block_device *, void *, struct gendisk *);
1416extern void bd_release_from_disk(struct block_device *, struct gendisk *);
1417#else
1418#define bd_claim_by_disk(bdev, holder, disk) bd_claim(bdev, holder)
1419#define bd_release_from_disk(bdev, disk) bd_release(bdev)
1420#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001421
1422/* fs/char_dev.c */
Joe Korty68eef3b2006-03-31 02:30:32 -08001423#define CHRDEV_MAJOR_HASH_SIZE 255
Linus Torvalds1da177e2005-04-16 15:20:36 -07001424extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
1425extern int register_chrdev_region(dev_t, unsigned, const char *);
1426extern int register_chrdev(unsigned int, const char *,
Arjan van de Ven99ac48f2006-03-28 01:56:41 -08001427 const struct file_operations *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001428extern int unregister_chrdev(unsigned int, const char *);
1429extern void unregister_chrdev_region(dev_t, unsigned);
1430extern int chrdev_open(struct inode *, struct file *);
Joe Korty68eef3b2006-03-31 02:30:32 -08001431extern void chrdev_show(struct seq_file *,off_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001432
1433/* fs/block_dev.c */
Joe Korty68eef3b2006-03-31 02:30:32 -08001434#define BLKDEV_MAJOR_HASH_SIZE 255
Linus Torvalds1da177e2005-04-16 15:20:36 -07001435#define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
1436extern const char *__bdevname(dev_t, char *buffer);
1437extern const char *bdevname(struct block_device *bdev, char *buffer);
1438extern struct block_device *lookup_bdev(const char *);
1439extern struct block_device *open_bdev_excl(const char *, int, void *);
1440extern void close_bdev_excl(struct block_device *);
Joe Korty68eef3b2006-03-31 02:30:32 -08001441extern void blkdev_show(struct seq_file *,off_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001442
1443extern void init_special_inode(struct inode *, umode_t, dev_t);
1444
1445/* Invalid inode operations -- fs/bad_inode.c */
1446extern void make_bad_inode(struct inode *);
1447extern int is_bad_inode(struct inode *);
1448
Arjan van de Ven4b6f5d22006-03-28 01:56:42 -08001449extern const struct file_operations read_fifo_fops;
1450extern const struct file_operations write_fifo_fops;
1451extern const struct file_operations rdwr_fifo_fops;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001452
1453extern int fs_may_remount_ro(struct super_block *);
1454
1455/*
1456 * return READ, READA, or WRITE
1457 */
1458#define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
1459
1460/*
1461 * return data direction, READ or WRITE
1462 */
1463#define bio_data_dir(bio) ((bio)->bi_rw & 1)
1464
1465extern int check_disk_change(struct block_device *);
1466extern int invalidate_inodes(struct super_block *);
Christoph Hellwig2ef41632005-05-05 16:15:59 -07001467extern int __invalidate_device(struct block_device *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001468extern int invalidate_partition(struct gendisk *, int);
1469unsigned long invalidate_mapping_pages(struct address_space *mapping,
1470 pgoff_t start, pgoff_t end);
1471unsigned long invalidate_inode_pages(struct address_space *mapping);
1472static inline void invalidate_remote_inode(struct inode *inode)
1473{
1474 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
1475 S_ISLNK(inode->i_mode))
1476 invalidate_inode_pages(inode->i_mapping);
1477}
1478extern int invalidate_inode_pages2(struct address_space *mapping);
1479extern int invalidate_inode_pages2_range(struct address_space *mapping,
1480 pgoff_t start, pgoff_t end);
1481extern int write_inode_now(struct inode *, int);
1482extern int filemap_fdatawrite(struct address_space *);
1483extern int filemap_flush(struct address_space *);
1484extern int filemap_fdatawait(struct address_space *);
1485extern int filemap_write_and_wait(struct address_space *mapping);
1486extern int filemap_write_and_wait_range(struct address_space *mapping,
1487 loff_t lstart, loff_t lend);
Andrew Mortonebcf28e2006-03-24 03:18:04 -08001488extern int wait_on_page_writeback_range(struct address_space *mapping,
1489 pgoff_t start, pgoff_t end);
1490extern int __filemap_fdatawrite_range(struct address_space *mapping,
1491 loff_t start, loff_t end, int sync_mode);
1492
Andrew Morton18e79b42006-03-24 03:18:14 -08001493extern long do_fsync(struct file *file, int datasync);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001494extern void sync_supers(void);
1495extern void sync_filesystems(int wait);
1496extern void emergency_sync(void);
1497extern void emergency_remount(void);
1498extern int do_remount_sb(struct super_block *sb, int flags,
1499 void *data, int force);
1500extern sector_t bmap(struct inode *, sector_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001501extern int notify_change(struct dentry *, struct iattr *);
1502extern int permission(struct inode *, int, struct nameidata *);
1503extern int generic_permission(struct inode *, int,
1504 int (*check_acl)(struct inode *, int));
1505
1506extern int get_write_access(struct inode *);
1507extern int deny_write_access(struct file *);
1508static inline void put_write_access(struct inode * inode)
1509{
1510 atomic_dec(&inode->i_writecount);
1511}
1512static inline void allow_write_access(struct file *file)
1513{
1514 if (file)
1515 atomic_inc(&file->f_dentry->d_inode->i_writecount);
1516}
1517extern int do_pipe(int *);
1518
Ulrich Drepper5590ff02006-01-18 17:43:53 -08001519extern int open_namei(int dfd, const char *, int, int, struct nameidata *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001520extern int may_open(struct nameidata *, int, int);
1521
1522extern int kernel_read(struct file *, unsigned long, char *, unsigned long);
1523extern struct file * open_exec(const char *);
1524
1525/* fs/dcache.c -- generic fs support functions */
1526extern int is_subdir(struct dentry *, struct dentry *);
1527extern ino_t find_inode_number(struct dentry *, struct qstr *);
1528
1529#include <linux/err.h>
1530
1531/* needed for stackable file system support */
1532extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
1533
1534extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
1535
1536extern void inode_init_once(struct inode *);
1537extern void iput(struct inode *);
1538extern struct inode * igrab(struct inode *);
1539extern ino_t iunique(struct super_block *, ino_t);
1540extern int inode_needs_sync(struct inode *inode);
1541extern void generic_delete_inode(struct inode *inode);
Mark Fashehcb2c0232005-07-07 17:56:03 -07001542extern void generic_drop_inode(struct inode *inode);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001543
Anton Altaparmakov88bd5122005-07-13 01:10:44 -07001544extern struct inode *ilookup5_nowait(struct super_block *sb,
1545 unsigned long hashval, int (*test)(struct inode *, void *),
1546 void *data);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001547extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
1548 int (*test)(struct inode *, void *), void *data);
1549extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
1550
1551extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
1552extern struct inode * iget_locked(struct super_block *, unsigned long);
1553extern void unlock_new_inode(struct inode *);
1554
1555static inline struct inode *iget(struct super_block *sb, unsigned long ino)
1556{
1557 struct inode *inode = iget_locked(sb, ino);
1558
1559 if (inode && (inode->i_state & I_NEW)) {
1560 sb->s_op->read_inode(inode);
1561 unlock_new_inode(inode);
1562 }
1563
1564 return inode;
1565}
1566
1567extern void __iget(struct inode * inode);
1568extern void clear_inode(struct inode *);
1569extern void destroy_inode(struct inode *);
1570extern struct inode *new_inode(struct super_block *);
1571extern int remove_suid(struct dentry *);
1572extern void remove_dquot_ref(struct super_block *, int, struct list_head *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001573
1574extern void __insert_inode_hash(struct inode *, unsigned long hashval);
1575extern void remove_inode_hash(struct inode *);
1576static inline void insert_inode_hash(struct inode *inode) {
1577 __insert_inode_hash(inode, inode->i_ino);
1578}
1579
1580extern struct file * get_empty_filp(void);
1581extern void file_move(struct file *f, struct list_head *list);
1582extern void file_kill(struct file *f);
1583struct bio;
1584extern void submit_bio(int, struct bio *);
1585extern int bdev_read_only(struct block_device *);
1586extern int set_blocksize(struct block_device *, int);
1587extern int sb_set_blocksize(struct super_block *, int);
1588extern int sb_min_blocksize(struct super_block *, int);
1589
1590extern int generic_file_mmap(struct file *, struct vm_area_struct *);
1591extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
1592extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
1593extern int file_send_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
1594extern ssize_t generic_file_read(struct file *, char __user *, size_t, loff_t *);
1595int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
1596extern ssize_t generic_file_write(struct file *, const char __user *, size_t, loff_t *);
1597extern ssize_t generic_file_aio_read(struct kiocb *, char __user *, size_t, loff_t);
1598extern ssize_t __generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t *);
1599extern ssize_t generic_file_aio_write(struct kiocb *, const char __user *, size_t, loff_t);
1600extern ssize_t generic_file_aio_write_nolock(struct kiocb *, const struct iovec *,
1601 unsigned long, loff_t *);
1602extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
1603 unsigned long *, loff_t, loff_t *, size_t, size_t);
1604extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
1605 unsigned long, loff_t, loff_t *, size_t, ssize_t);
1606extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
1607extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
1608ssize_t generic_file_write_nolock(struct file *file, const struct iovec *iov,
1609 unsigned long nr_segs, loff_t *ppos);
1610extern ssize_t generic_file_sendfile(struct file *, loff_t *, size_t, read_actor_t, void *);
1611extern void do_generic_mapping_read(struct address_space *mapping,
1612 struct file_ra_state *, struct file *,
1613 loff_t *, read_descriptor_t *, read_actor_t);
Andrew Morton88e6fae2006-04-11 13:59:36 +02001614
1615/* fs/splice.c */
Jens Axboecbb7e572006-04-11 14:57:50 +02001616extern ssize_t generic_file_splice_read(struct file *, loff_t *,
Andrew Morton88e6fae2006-04-11 13:59:36 +02001617 struct pipe_inode_info *, size_t, unsigned int);
1618extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
Jens Axboecbb7e572006-04-11 14:57:50 +02001619 struct file *, loff_t *, size_t, unsigned int);
Andrew Morton88e6fae2006-04-11 13:59:36 +02001620extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
Jens Axboecbb7e572006-04-11 14:57:50 +02001621 struct file *out, loff_t *, size_t len, unsigned int flags);
1622extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
Andrew Morton88e6fae2006-04-11 13:59:36 +02001623 size_t len, unsigned int flags);
1624
Linus Torvalds1da177e2005-04-16 15:20:36 -07001625extern void
1626file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001627extern ssize_t generic_file_readv(struct file *filp, const struct iovec *iov,
1628 unsigned long nr_segs, loff_t *ppos);
1629ssize_t generic_file_writev(struct file *filp, const struct iovec *iov,
1630 unsigned long nr_segs, loff_t *ppos);
1631extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
1632extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
1633extern loff_t remote_llseek(struct file *file, loff_t offset, int origin);
1634extern int generic_file_open(struct inode * inode, struct file * filp);
1635extern int nonseekable_open(struct inode * inode, struct file * filp);
1636
Carsten Otteceffc072005-06-23 22:05:25 -07001637#ifdef CONFIG_FS_XIP
Carsten Otteeb6fe0c2005-06-23 22:05:28 -07001638extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
1639 loff_t *ppos);
Carsten Otteceffc072005-06-23 22:05:25 -07001640extern ssize_t xip_file_sendfile(struct file *in_file, loff_t *ppos,
1641 size_t count, read_actor_t actor,
1642 void *target);
1643extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
Carsten Otteeb6fe0c2005-06-23 22:05:28 -07001644extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
1645 size_t len, loff_t *ppos);
Carsten Otteceffc072005-06-23 22:05:25 -07001646extern int xip_truncate_page(struct address_space *mapping, loff_t from);
Carsten Otte6d791252005-06-23 22:05:26 -07001647#else
1648static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
1649{
1650 return 0;
1651}
Carsten Otteceffc072005-06-23 22:05:25 -07001652#endif
1653
Linus Torvalds1da177e2005-04-16 15:20:36 -07001654static inline void do_generic_file_read(struct file * filp, loff_t *ppos,
1655 read_descriptor_t * desc,
1656 read_actor_t actor)
1657{
1658 do_generic_mapping_read(filp->f_mapping,
1659 &filp->f_ra,
1660 filp,
1661 ppos,
1662 desc,
1663 actor);
1664}
1665
1666ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
1667 struct block_device *bdev, const struct iovec *iov, loff_t offset,
Badari Pulavarty1d8fa7a2006-03-26 01:38:02 -08001668 unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001669 int lock_type);
1670
1671enum {
1672 DIO_LOCKING = 1, /* need locking between buffered and direct access */
1673 DIO_NO_LOCKING, /* bdev; no locking at all between buffered/direct */
1674 DIO_OWN_LOCKING, /* filesystem locks buffered and direct internally */
1675};
1676
1677static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
1678 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
Badari Pulavarty1d8fa7a2006-03-26 01:38:02 -08001679 loff_t offset, unsigned long nr_segs, get_block_t get_block,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001680 dio_iodone_t end_io)
1681{
1682 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
Badari Pulavarty1d8fa7a2006-03-26 01:38:02 -08001683 nr_segs, get_block, end_io, DIO_LOCKING);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001684}
1685
1686static inline ssize_t blockdev_direct_IO_no_locking(int rw, struct kiocb *iocb,
1687 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
Badari Pulavarty1d8fa7a2006-03-26 01:38:02 -08001688 loff_t offset, unsigned long nr_segs, get_block_t get_block,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001689 dio_iodone_t end_io)
1690{
1691 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
Badari Pulavarty1d8fa7a2006-03-26 01:38:02 -08001692 nr_segs, get_block, end_io, DIO_NO_LOCKING);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001693}
1694
1695static inline ssize_t blockdev_direct_IO_own_locking(int rw, struct kiocb *iocb,
1696 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
Badari Pulavarty1d8fa7a2006-03-26 01:38:02 -08001697 loff_t offset, unsigned long nr_segs, get_block_t get_block,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001698 dio_iodone_t end_io)
1699{
1700 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
Badari Pulavarty1d8fa7a2006-03-26 01:38:02 -08001701 nr_segs, get_block, end_io, DIO_OWN_LOCKING);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001702}
1703
Arjan van de Ven4b6f5d22006-03-28 01:56:42 -08001704extern const struct file_operations generic_ro_fops;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001705
1706#define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
1707
1708extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
1709extern int vfs_follow_link(struct nameidata *, const char *);
1710extern int page_readlink(struct dentry *, char __user *, int);
Linus Torvaldscc314ee2005-08-19 18:02:56 -07001711extern void *page_follow_link_light(struct dentry *, struct nameidata *);
1712extern void page_put_link(struct dentry *, struct nameidata *, void *);
Kirill Korotaev0adb25d2006-03-11 03:27:13 -08001713extern int __page_symlink(struct inode *inode, const char *symname, int len,
1714 gfp_t gfp_mask);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001715extern int page_symlink(struct inode *inode, const char *symname, int len);
1716extern struct inode_operations page_symlink_inode_operations;
1717extern int generic_readlink(struct dentry *, char __user *, int);
1718extern void generic_fillattr(struct inode *, struct kstat *);
1719extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1720void inode_add_bytes(struct inode *inode, loff_t bytes);
1721void inode_sub_bytes(struct inode *inode, loff_t bytes);
1722loff_t inode_get_bytes(struct inode *inode);
1723void inode_set_bytes(struct inode *inode, loff_t bytes);
1724
1725extern int vfs_readdir(struct file *, filldir_t, void *);
1726
1727extern int vfs_stat(char __user *, struct kstat *);
1728extern int vfs_lstat(char __user *, struct kstat *);
Ulrich Drepper5590ff02006-01-18 17:43:53 -08001729extern int vfs_stat_fd(int dfd, char __user *, struct kstat *);
1730extern int vfs_lstat_fd(int dfd, char __user *, struct kstat *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001731extern int vfs_fstat(unsigned int, struct kstat *);
1732
1733extern int vfs_ioctl(struct file *, unsigned int, unsigned int, unsigned long);
1734
1735extern struct file_system_type *get_fs_type(const char *name);
1736extern struct super_block *get_super(struct block_device *);
1737extern struct super_block *user_get_super(dev_t);
1738extern void drop_super(struct super_block *sb);
1739
1740extern int dcache_dir_open(struct inode *, struct file *);
1741extern int dcache_dir_close(struct inode *, struct file *);
1742extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
1743extern int dcache_readdir(struct file *, void *, filldir_t);
1744extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1745extern int simple_statfs(struct super_block *, struct kstatfs *);
1746extern int simple_link(struct dentry *, struct inode *, struct dentry *);
1747extern int simple_unlink(struct inode *, struct dentry *);
1748extern int simple_rmdir(struct inode *, struct dentry *);
1749extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1750extern int simple_sync_file(struct file *, struct dentry *, int);
1751extern int simple_empty(struct dentry *);
1752extern int simple_readpage(struct file *file, struct page *page);
1753extern int simple_prepare_write(struct file *file, struct page *page,
1754 unsigned offset, unsigned to);
1755extern int simple_commit_write(struct file *file, struct page *page,
1756 unsigned offset, unsigned to);
1757
1758extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
1759extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
Arjan van de Ven4b6f5d22006-03-28 01:56:42 -08001760extern const struct file_operations simple_dir_operations;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001761extern struct inode_operations simple_dir_inode_operations;
Arjan van de Ven4b6f5d22006-03-28 01:56:42 -08001762struct tree_descr { char *name; const struct file_operations *ops; int mode; };
Linus Torvalds1da177e2005-04-16 15:20:36 -07001763struct dentry *d_alloc_name(struct dentry *, const char *);
1764extern int simple_fill_super(struct super_block *, int, struct tree_descr *);
1765extern int simple_pin_fs(char *name, struct vfsmount **mount, int *count);
1766extern void simple_release_fs(struct vfsmount **mount, int *count);
1767
1768extern ssize_t simple_read_from_buffer(void __user *, size_t, loff_t *, const void *, size_t);
1769
Christoph Lametere965f962006-02-01 03:05:41 -08001770#ifdef CONFIG_MIGRATION
1771extern int buffer_migrate_page(struct page *, struct page *);
1772#else
1773#define buffer_migrate_page NULL
1774#endif
1775
Linus Torvalds1da177e2005-04-16 15:20:36 -07001776extern int inode_change_ok(struct inode *, struct iattr *);
1777extern int __must_check inode_setattr(struct inode *, struct iattr *);
1778
Christoph Hellwig870f4812006-01-09 20:52:01 -08001779extern void file_update_time(struct file *file);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001780
1781static inline ino_t parent_ino(struct dentry *dentry)
1782{
1783 ino_t res;
1784
1785 spin_lock(&dentry->d_lock);
1786 res = dentry->d_parent->d_inode->i_ino;
1787 spin_unlock(&dentry->d_lock);
1788 return res;
1789}
1790
1791/* kernel/fork.c */
1792extern int unshare_files(void);
1793
1794/* Transaction based IO helpers */
1795
1796/*
1797 * An argresp is stored in an allocated page and holds the
1798 * size of the argument or response, along with its content
1799 */
1800struct simple_transaction_argresp {
1801 ssize_t size;
1802 char data[0];
1803};
1804
1805#define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
1806
1807char *simple_transaction_get(struct file *file, const char __user *buf,
1808 size_t size);
1809ssize_t simple_transaction_read(struct file *file, char __user *buf,
1810 size_t size, loff_t *pos);
1811int simple_transaction_release(struct inode *inode, struct file *file);
1812
1813static inline void simple_transaction_set(struct file *file, size_t n)
1814{
1815 struct simple_transaction_argresp *ar = file->private_data;
1816
1817 BUG_ON(n > SIMPLE_TRANSACTION_LIMIT);
1818
1819 /*
1820 * The barrier ensures that ar->size will really remain zero until
1821 * ar->data is ready for reading.
1822 */
1823 smp_mb();
1824 ar->size = n;
1825}
1826
Arnd Bergmannacaefc22005-05-18 14:40:59 +02001827/*
1828 * simple attribute files
1829 *
1830 * These attributes behave similar to those in sysfs:
1831 *
1832 * Writing to an attribute immediately sets a value, an open file can be
1833 * written to multiple times.
1834 *
1835 * Reading from an attribute creates a buffer from the value that might get
1836 * read with multiple read calls. When the attribute has been read
1837 * completely, no further read calls are possible until the file is opened
1838 * again.
1839 *
1840 * All attributes contain a text representation of a numeric value
1841 * that are accessed with the get() and set() functions.
1842 */
1843#define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
1844static int __fops ## _open(struct inode *inode, struct file *file) \
1845{ \
1846 __simple_attr_check_format(__fmt, 0ull); \
1847 return simple_attr_open(inode, file, __get, __set, __fmt); \
1848} \
1849static struct file_operations __fops = { \
1850 .owner = THIS_MODULE, \
1851 .open = __fops ## _open, \
1852 .release = simple_attr_close, \
1853 .read = simple_attr_read, \
1854 .write = simple_attr_write, \
1855};
1856
1857static inline void __attribute__((format(printf, 1, 2)))
1858__simple_attr_check_format(const char *fmt, ...)
1859{
1860 /* don't do anything, just let the compiler check the arguments; */
1861}
1862
1863int simple_attr_open(struct inode *inode, struct file *file,
1864 u64 (*get)(void *), void (*set)(void *, u64),
1865 const char *fmt);
1866int simple_attr_close(struct inode *inode, struct file *file);
1867ssize_t simple_attr_read(struct file *file, char __user *buf,
1868 size_t len, loff_t *ppos);
1869ssize_t simple_attr_write(struct file *file, const char __user *buf,
1870 size_t len, loff_t *ppos);
1871
1872
Linus Torvalds1da177e2005-04-16 15:20:36 -07001873#ifdef CONFIG_SECURITY
1874static inline char *alloc_secdata(void)
1875{
1876 return (char *)get_zeroed_page(GFP_KERNEL);
1877}
1878
1879static inline void free_secdata(void *secdata)
1880{
1881 free_page((unsigned long)secdata);
1882}
1883#else
1884static inline char *alloc_secdata(void)
1885{
1886 return (char *)1;
1887}
1888
1889static inline void free_secdata(void *secdata)
1890{ }
1891#endif /* CONFIG_SECURITY */
1892
1893#endif /* __KERNEL__ */
1894#endif /* _LINUX_FS_H */