blob: 7f140480c6a83f1cd6ec47d866239b6182d19afa [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;
38
39struct inodes_stat_t {
40 int nr_inodes;
41 int nr_unused;
42 int dummy[5];
43};
44extern struct inodes_stat_t inodes_stat;
45
46extern int leases_enable, lease_break_time;
47
48#ifdef CONFIG_DNOTIFY
49extern int dir_notify_enable;
50#endif
51
52#define NR_FILE 8192 /* this can well be larger on a larger system */
53
54#define MAY_EXEC 1
55#define MAY_WRITE 2
56#define MAY_READ 4
57#define MAY_APPEND 8
58
59#define FMODE_READ 1
60#define FMODE_WRITE 2
61
62/* Internal kernel extensions */
63#define FMODE_LSEEK 4
64#define FMODE_PREAD 8
65#define FMODE_PWRITE FMODE_PREAD /* These go hand in hand */
66
67#define RW_MASK 1
68#define RWA_MASK 2
69#define READ 0
70#define WRITE 1
71#define READA 2 /* read-ahead - don't block if no resources */
Jan Karaa7662232005-09-06 15:19:10 -070072#define SWRITE 3 /* for ll_rw_block() - wait for buffer lock */
Linus Torvalds1da177e2005-04-16 15:20:36 -070073#define SPECIAL 4 /* For non-blockdevice requests in request queue */
74#define READ_SYNC (READ | (1 << BIO_RW_SYNC))
75#define WRITE_SYNC (WRITE | (1 << BIO_RW_SYNC))
76#define WRITE_BARRIER ((1 << BIO_RW) | (1 << BIO_RW_BARRIER))
77
78#define SEL_IN 1
79#define SEL_OUT 2
80#define SEL_EX 4
81
82/* public flags for file_system_type */
83#define FS_REQUIRES_DEV 1
84#define FS_BINARY_MOUNTDATA 2
85#define FS_REVAL_DOT 16384 /* Check the paths ".", ".." for staleness */
86#define FS_ODD_RENAME 32768 /* Temporary stuff; will go away as soon
87 * as nfs_rename() will be cleaned up
88 */
89/*
90 * These are the fs-independent mount-flags: up to 32 flags are supported
91 */
92#define MS_RDONLY 1 /* Mount read-only */
93#define MS_NOSUID 2 /* Ignore suid and sgid bits */
94#define MS_NODEV 4 /* Disallow access to device special files */
95#define MS_NOEXEC 8 /* Disallow program execution */
96#define MS_SYNCHRONOUS 16 /* Writes are synced at once */
97#define MS_REMOUNT 32 /* Alter flags of a mounted FS */
98#define MS_MANDLOCK 64 /* Allow mandatory locks on an FS */
99#define MS_DIRSYNC 128 /* Directory modifications are synchronous */
100#define MS_NOATIME 1024 /* Do not update access times. */
101#define MS_NODIRATIME 2048 /* Do not update directory access times */
102#define MS_BIND 4096
103#define MS_MOVE 8192
104#define MS_REC 16384
105#define MS_VERBOSE 32768
Ram Pai9676f0c2005-11-07 17:21:20 -0500106#define MS_UNBINDABLE (1<<17) /* change to unbindable */
Ram Pai07b20882005-11-07 17:19:07 -0500107#define MS_PRIVATE (1<<18) /* change to private */
Ram Paia58b0eb2005-11-07 17:20:48 -0500108#define MS_SLAVE (1<<19) /* change to slave */
Ram Pai03e06e62005-11-07 17:19:33 -0500109#define MS_SHARED (1<<20) /* change to shared */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700110#define MS_POSIXACL (1<<16) /* VFS does not apply the umask */
111#define MS_ACTIVE (1<<30)
112#define MS_NOUSER (1<<31)
113
114/*
115 * Superblock flags that can be altered by MS_REMOUNT
116 */
117#define MS_RMT_MASK (MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK|MS_NOATIME|\
118 MS_NODIRATIME)
119
120/*
121 * Old magic mount flag and mask
122 */
123#define MS_MGC_VAL 0xC0ED0000
124#define MS_MGC_MSK 0xffff0000
125
126/* Inode flags - they have nothing to superblock flags now */
127
128#define S_SYNC 1 /* Writes are synced at once */
129#define S_NOATIME 2 /* Do not update access times */
130#define S_APPEND 4 /* Append-only file */
131#define S_IMMUTABLE 8 /* Immutable file */
132#define S_DEAD 16 /* removed, but still open directory */
133#define S_NOQUOTA 32 /* Inode is not counted to quota */
134#define S_DIRSYNC 64 /* Directory modifications are synchronous */
135#define S_NOCMTIME 128 /* Do not update file c/mtime */
136#define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
137#define S_PRIVATE 512 /* Inode is fs-internal */
138
139/*
140 * Note that nosuid etc flags are inode-specific: setting some file-system
141 * flags just means all the inodes inherit those flags by default. It might be
142 * possible to override it selectively if you really wanted to with some
143 * ioctl() that is not currently implemented.
144 *
145 * Exception: MS_RDONLY is always applied to the entire file system.
146 *
147 * Unfortunately, it is possible to change a filesystems flags with it mounted
148 * with files in use. This means that all of the inodes will not have their
149 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
150 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
151 */
152#define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))
153
154#define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
155#define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
156 ((inode)->i_flags & S_SYNC))
157#define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
158 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
159#define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
160
161#define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
162#define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
163#define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
164#define IS_NOATIME(inode) (__IS_FLG(inode, MS_NOATIME) || ((inode)->i_flags & S_NOATIME))
165#define IS_NODIRATIME(inode) __IS_FLG(inode, MS_NODIRATIME)
166#define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
167
168#define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
169#define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
170#define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
171#define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
172
173/* the read-only stuff doesn't really belong here, but any other place is
174 probably as bad and I don't want to create yet another include file. */
175
176#define BLKROSET _IO(0x12,93) /* set device read-only (0 = read-write) */
177#define BLKROGET _IO(0x12,94) /* get read-only status (0 = read_write) */
178#define BLKRRPART _IO(0x12,95) /* re-read partition table */
179#define BLKGETSIZE _IO(0x12,96) /* return device size /512 (long *arg) */
180#define BLKFLSBUF _IO(0x12,97) /* flush buffer cache */
181#define BLKRASET _IO(0x12,98) /* set read ahead for block device */
182#define BLKRAGET _IO(0x12,99) /* get current read ahead setting */
183#define BLKFRASET _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
184#define BLKFRAGET _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
185#define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
186#define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
187#define BLKSSZGET _IO(0x12,104)/* get block device sector size */
188#if 0
189#define BLKPG _IO(0x12,105)/* See blkpg.h */
190
191/* Some people are morons. Do not use sizeof! */
192
193#define BLKELVGET _IOR(0x12,106,size_t)/* elevator get */
194#define BLKELVSET _IOW(0x12,107,size_t)/* elevator set */
195/* This was here just to show that the number is taken -
196 probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
197#endif
198/* A jump here: 108-111 have been used for various private purposes. */
199#define BLKBSZGET _IOR(0x12,112,size_t)
200#define BLKBSZSET _IOW(0x12,113,size_t)
201#define BLKGETSIZE64 _IOR(0x12,114,size_t) /* return device size in bytes (u64 *arg) */
202
203#define BMAP_IOCTL 1 /* obsolete - kept for compatibility */
204#define FIBMAP _IO(0x00,1) /* bmap access */
205#define FIGETBSZ _IO(0x00,2) /* get the block size used for bmap */
206
207#ifdef __KERNEL__
208
209#include <linux/linkage.h>
210#include <linux/wait.h>
211#include <linux/types.h>
212#include <linux/kdev_t.h>
213#include <linux/dcache.h>
214#include <linux/stat.h>
215#include <linux/cache.h>
216#include <linux/kobject.h>
217#include <linux/list.h>
218#include <linux/radix-tree.h>
219#include <linux/prio_tree.h>
220#include <linux/init.h>
Jens Axboe22e2c502005-06-27 10:55:12 +0200221#include <linux/sched.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700222
223#include <asm/atomic.h>
224#include <asm/semaphore.h>
225#include <asm/byteorder.h>
226
Christoph Hellwiga885c8c2006-01-08 01:02:50 -0800227struct hd_geometry;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700228struct iovec;
229struct nameidata;
Christoph Hellwig92198f72005-06-23 22:00:59 -0700230struct kiocb;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700231struct pipe_inode_info;
232struct poll_table_struct;
233struct kstatfs;
234struct vm_area_struct;
235struct vfsmount;
236
237/* Used to be a macro which just called the function, now just a function */
238extern void update_atime (struct inode *);
239
240extern void __init inode_init(unsigned long);
241extern void __init inode_init_early(void);
242extern void __init mnt_init(unsigned long);
243extern void __init files_init(unsigned long);
244
245struct buffer_head;
246typedef int (get_block_t)(struct inode *inode, sector_t iblock,
247 struct buffer_head *bh_result, int create);
248typedef int (get_blocks_t)(struct inode *inode, sector_t iblock,
249 unsigned long max_blocks,
250 struct buffer_head *bh_result, int create);
Christoph Hellwig92198f72005-06-23 22:00:59 -0700251typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700252 ssize_t bytes, void *private);
253
254/*
255 * Attribute flags. These should be or-ed together to figure out what
256 * has been changed!
257 */
258#define ATTR_MODE 1
259#define ATTR_UID 2
260#define ATTR_GID 4
261#define ATTR_SIZE 8
262#define ATTR_ATIME 16
263#define ATTR_MTIME 32
264#define ATTR_CTIME 64
265#define ATTR_ATIME_SET 128
266#define ATTR_MTIME_SET 256
267#define ATTR_FORCE 512 /* Not a change, but a change it */
268#define ATTR_ATTR_FLAG 1024
269#define ATTR_KILL_SUID 2048
270#define ATTR_KILL_SGID 4096
Miklos Szeredicc4e69d2005-11-07 00:59:49 -0800271#define ATTR_FILE 8192
Linus Torvalds1da177e2005-04-16 15:20:36 -0700272
273/*
274 * This is the Inode Attributes structure, used for notify_change(). It
275 * uses the above definitions as flags, to know which values have changed.
276 * Also, in this manner, a Filesystem can look at only the values it cares
277 * about. Basically, these are the attributes that the VFS layer can
278 * request to change from the FS layer.
279 *
280 * Derek Atkins <warlord@MIT.EDU> 94-10-20
281 */
282struct iattr {
283 unsigned int ia_valid;
284 umode_t ia_mode;
285 uid_t ia_uid;
286 gid_t ia_gid;
287 loff_t ia_size;
288 struct timespec ia_atime;
289 struct timespec ia_mtime;
290 struct timespec ia_ctime;
Miklos Szeredicc4e69d2005-11-07 00:59:49 -0800291
292 /*
293 * Not an attribute, but an auxilary info for filesystems wanting to
294 * implement an ftruncate() like method. NOTE: filesystem should
295 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
296 */
297 struct file *ia_file;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700298};
299
300/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700301 * Includes for diskquotas.
302 */
303#include <linux/quota.h>
304
Zach Brown994fc28c2005-12-15 14:28:17 -0800305/**
306 * enum positive_aop_returns - aop return codes with specific semantics
307 *
308 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
309 * completed, that the page is still locked, and
310 * should be considered active. The VM uses this hint
311 * to return the page to the active list -- it won't
312 * be a candidate for writeback again in the near
313 * future. Other callers must be careful to unlock
314 * the page if they get this return. Returned by
315 * writepage();
316 *
317 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
318 * unlocked it and the page might have been truncated.
319 * The caller should back up to acquiring a new page and
320 * trying again. The aop will be taking reasonable
321 * precautions not to livelock. If the caller held a page
322 * reference, it should drop it before retrying. Returned
323 * by readpage(), prepare_write(), and commit_write().
324 *
325 * address_space_operation functions return these large constants to indicate
326 * special semantics to the caller. These are much larger than the bytes in a
327 * page to allow for functions that return the number of bytes operated on in a
328 * given page.
329 */
330
331enum positive_aop_returns {
332 AOP_WRITEPAGE_ACTIVATE = 0x80000,
333 AOP_TRUNCATED_PAGE = 0x80001,
334};
335
Linus Torvalds1da177e2005-04-16 15:20:36 -0700336/*
337 * oh the beauties of C type declarations.
338 */
339struct page;
340struct address_space;
341struct writeback_control;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700342
343struct address_space_operations {
344 int (*writepage)(struct page *page, struct writeback_control *wbc);
345 int (*readpage)(struct file *, struct page *);
346 int (*sync_page)(struct page *);
347
348 /* Write back some dirty pages from this mapping. */
349 int (*writepages)(struct address_space *, struct writeback_control *);
350
351 /* Set a page dirty */
352 int (*set_page_dirty)(struct page *page);
353
354 int (*readpages)(struct file *filp, struct address_space *mapping,
355 struct list_head *pages, unsigned nr_pages);
356
357 /*
358 * ext3 requires that a successful prepare_write() call be followed
359 * by a commit_write() call - they must be balanced
360 */
361 int (*prepare_write)(struct file *, struct page *, unsigned, unsigned);
362 int (*commit_write)(struct file *, struct page *, unsigned, unsigned);
363 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
364 sector_t (*bmap)(struct address_space *, sector_t);
365 int (*invalidatepage) (struct page *, unsigned long);
Al Viro27496a82005-10-21 03:20:48 -0400366 int (*releasepage) (struct page *, gfp_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700367 ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
368 loff_t offset, unsigned long nr_segs);
Carsten Otteceffc072005-06-23 22:05:25 -0700369 struct page* (*get_xip_page)(struct address_space *, sector_t,
370 int);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700371};
372
373struct backing_dev_info;
374struct address_space {
375 struct inode *host; /* owner: inode, block_device */
376 struct radix_tree_root page_tree; /* radix tree of all pages */
377 rwlock_t tree_lock; /* and rwlock protecting it */
378 unsigned int i_mmap_writable;/* count VM_SHARED mappings */
379 struct prio_tree_root i_mmap; /* tree of private and shared mappings */
380 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
381 spinlock_t i_mmap_lock; /* protect tree, count, list */
382 unsigned int truncate_count; /* Cover race condition with truncate */
383 unsigned long nrpages; /* number of total pages */
384 pgoff_t writeback_index;/* writeback starts here */
385 struct address_space_operations *a_ops; /* methods */
386 unsigned long flags; /* error bits/gfp mask */
387 struct backing_dev_info *backing_dev_info; /* device readahead, etc */
388 spinlock_t private_lock; /* for use by the address_space */
389 struct list_head private_list; /* ditto */
390 struct address_space *assoc_mapping; /* ditto */
391} __attribute__((aligned(sizeof(long))));
392 /*
393 * On most architectures that alignment is already the case; but
394 * must be enforced here for CRIS, to let the least signficant bit
395 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
396 */
397
398struct block_device {
399 dev_t bd_dev; /* not a kdev_t - it's a search key */
400 struct inode * bd_inode; /* will die */
401 int bd_openers;
402 struct semaphore bd_sem; /* open/close mutex */
403 struct semaphore bd_mount_sem; /* mount mutex */
404 struct list_head bd_inodes;
405 void * bd_holder;
406 int bd_holders;
407 struct block_device * bd_contains;
408 unsigned bd_block_size;
409 struct hd_struct * bd_part;
410 /* number of times partitions within this device have been opened. */
411 unsigned bd_part_count;
412 int bd_invalidated;
413 struct gendisk * bd_disk;
414 struct list_head bd_list;
415 struct backing_dev_info *bd_inode_backing_dev_info;
416 /*
417 * Private data. You must have bd_claim'ed the block_device
418 * to use this. NOTE: bd_claim allows an owner to claim
419 * the same device multiple times, the owner must take special
420 * care to not mess up bd_private for that case.
421 */
422 unsigned long bd_private;
423};
424
425/*
426 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
427 * radix trees
428 */
429#define PAGECACHE_TAG_DIRTY 0
430#define PAGECACHE_TAG_WRITEBACK 1
431
432int mapping_tagged(struct address_space *mapping, int tag);
433
434/*
435 * Might pages of this file be mapped into userspace?
436 */
437static inline int mapping_mapped(struct address_space *mapping)
438{
439 return !prio_tree_empty(&mapping->i_mmap) ||
440 !list_empty(&mapping->i_mmap_nonlinear);
441}
442
443/*
444 * Might pages of this file have been modified in userspace?
445 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
446 * marks vma as VM_SHARED if it is shared, and the file was opened for
447 * writing i.e. vma may be mprotected writable even if now readonly.
448 */
449static inline int mapping_writably_mapped(struct address_space *mapping)
450{
451 return mapping->i_mmap_writable != 0;
452}
453
454/*
455 * Use sequence counter to get consistent i_size on 32-bit processors.
456 */
457#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
458#include <linux/seqlock.h>
459#define __NEED_I_SIZE_ORDERED
460#define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
461#else
462#define i_size_ordered_init(inode) do { } while (0)
463#endif
464
465struct inode {
466 struct hlist_node i_hash;
467 struct list_head i_list;
468 struct list_head i_sb_list;
469 struct list_head i_dentry;
470 unsigned long i_ino;
471 atomic_t i_count;
472 umode_t i_mode;
473 unsigned int i_nlink;
474 uid_t i_uid;
475 gid_t i_gid;
476 dev_t i_rdev;
477 loff_t i_size;
478 struct timespec i_atime;
479 struct timespec i_mtime;
480 struct timespec i_ctime;
481 unsigned int i_blkbits;
482 unsigned long i_blksize;
483 unsigned long i_version;
484 unsigned long i_blocks;
485 unsigned short i_bytes;
486 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
487 struct semaphore i_sem;
488 struct rw_semaphore i_alloc_sem;
489 struct inode_operations *i_op;
490 struct file_operations *i_fop; /* former ->i_op->default_file_ops */
491 struct super_block *i_sb;
492 struct file_lock *i_flock;
493 struct address_space *i_mapping;
494 struct address_space i_data;
495#ifdef CONFIG_QUOTA
496 struct dquot *i_dquot[MAXQUOTAS];
497#endif
498 /* These three should probably be a union */
499 struct list_head i_devices;
500 struct pipe_inode_info *i_pipe;
501 struct block_device *i_bdev;
502 struct cdev *i_cdev;
503 int i_cindex;
504
505 __u32 i_generation;
506
507#ifdef CONFIG_DNOTIFY
508 unsigned long i_dnotify_mask; /* Directory notify events */
509 struct dnotify_struct *i_dnotify; /* for directory notifications */
510#endif
511
Robert Love0eeca282005-07-12 17:06:03 -0400512#ifdef CONFIG_INOTIFY
513 struct list_head inotify_watches; /* watches on this inode */
514 struct semaphore inotify_sem; /* protects the watches list */
515#endif
516
Linus Torvalds1da177e2005-04-16 15:20:36 -0700517 unsigned long i_state;
518 unsigned long dirtied_when; /* jiffies of first dirtying */
519
520 unsigned int i_flags;
521
522 atomic_t i_writecount;
523 void *i_security;
524 union {
525 void *generic_ip;
526 } u;
527#ifdef __NEED_I_SIZE_ORDERED
528 seqcount_t i_size_seqcount;
529#endif
530};
531
532/*
533 * NOTE: in a 32bit arch with a preemptable kernel and
534 * an UP compile the i_size_read/write must be atomic
535 * with respect to the local cpu (unlike with preempt disabled),
536 * but they don't need to be atomic with respect to other cpus like in
537 * true SMP (so they need either to either locally disable irq around
538 * the read or for example on x86 they can be still implemented as a
539 * cmpxchg8b without the need of the lock prefix). For SMP compiles
540 * and 64bit archs it makes no difference if preempt is enabled or not.
541 */
542static inline loff_t i_size_read(struct inode *inode)
543{
544#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
545 loff_t i_size;
546 unsigned int seq;
547
548 do {
549 seq = read_seqcount_begin(&inode->i_size_seqcount);
550 i_size = inode->i_size;
551 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
552 return i_size;
553#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
554 loff_t i_size;
555
556 preempt_disable();
557 i_size = inode->i_size;
558 preempt_enable();
559 return i_size;
560#else
561 return inode->i_size;
562#endif
563}
564
565
566static inline void i_size_write(struct inode *inode, loff_t i_size)
567{
568#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
569 write_seqcount_begin(&inode->i_size_seqcount);
570 inode->i_size = i_size;
571 write_seqcount_end(&inode->i_size_seqcount);
572#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
573 preempt_disable();
574 inode->i_size = i_size;
575 preempt_enable();
576#else
577 inode->i_size = i_size;
578#endif
579}
580
581static inline unsigned iminor(struct inode *inode)
582{
583 return MINOR(inode->i_rdev);
584}
585
586static inline unsigned imajor(struct inode *inode)
587{
588 return MAJOR(inode->i_rdev);
589}
590
591extern struct block_device *I_BDEV(struct inode *inode);
592
593struct fown_struct {
594 rwlock_t lock; /* protects pid, uid, euid fields */
595 int pid; /* pid or -pgrp where SIGIO should be sent */
596 uid_t uid, euid; /* uid/euid of process setting the owner */
597 void *security;
598 int signum; /* posix.1b rt signal to be delivered on IO */
599};
600
601/*
602 * Track a single file's readahead state
603 */
604struct file_ra_state {
605 unsigned long start; /* Current window */
606 unsigned long size;
607 unsigned long flags; /* ra flags RA_FLAG_xxx*/
608 unsigned long cache_hit; /* cache hit count*/
609 unsigned long prev_page; /* Cache last read() position */
610 unsigned long ahead_start; /* Ahead window */
611 unsigned long ahead_size;
612 unsigned long ra_pages; /* Maximum readahead window */
613 unsigned long mmap_hit; /* Cache hit stat for mmap accesses */
614 unsigned long mmap_miss; /* Cache miss stat for mmap accesses */
615};
616#define RA_FLAG_MISS 0x01 /* a cache miss occured against this file */
617#define RA_FLAG_INCACHE 0x02 /* file is already in cache */
618
619struct file {
Eric Dumazet2f512012005-10-30 15:02:16 -0800620 /*
621 * fu_list becomes invalid after file_free is called and queued via
622 * fu_rcuhead for RCU freeing
623 */
624 union {
625 struct list_head fu_list;
626 struct rcu_head fu_rcuhead;
627 } f_u;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700628 struct dentry *f_dentry;
629 struct vfsmount *f_vfsmnt;
630 struct file_operations *f_op;
631 atomic_t f_count;
632 unsigned int f_flags;
633 mode_t f_mode;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700634 loff_t f_pos;
635 struct fown_struct f_owner;
636 unsigned int f_uid, f_gid;
637 struct file_ra_state f_ra;
638
Linus Torvalds1da177e2005-04-16 15:20:36 -0700639 unsigned long f_version;
640 void *f_security;
641
642 /* needed for tty driver, and maybe others */
643 void *private_data;
644
645#ifdef CONFIG_EPOLL
646 /* Used by fs/eventpoll.c to link all the hooks to this file */
647 struct list_head f_ep_links;
648 spinlock_t f_ep_lock;
649#endif /* #ifdef CONFIG_EPOLL */
650 struct address_space *f_mapping;
651};
652extern spinlock_t files_lock;
653#define file_list_lock() spin_lock(&files_lock);
654#define file_list_unlock() spin_unlock(&files_lock);
655
Nick Piggin095975d2006-01-08 01:02:19 -0800656#define get_file(x) atomic_inc(&(x)->f_count)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700657#define file_count(x) atomic_read(&(x)->f_count)
658
659#define MAX_NON_LFS ((1UL<<31) - 1)
660
661/* Page cache limit. The filesystems should put that into their s_maxbytes
662 limits, otherwise bad things can happen in VM. */
663#if BITS_PER_LONG==32
664#define MAX_LFS_FILESIZE (((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
665#elif BITS_PER_LONG==64
666#define MAX_LFS_FILESIZE 0x7fffffffffffffffUL
667#endif
668
669#define FL_POSIX 1
670#define FL_FLOCK 2
671#define FL_ACCESS 8 /* not trying to lock, just looking */
672#define FL_LOCKD 16 /* lock held by rpc.lockd */
673#define FL_LEASE 32 /* lease held on this file */
674#define FL_SLEEP 128 /* A blocking lock */
675
676/*
677 * The POSIX file lock owner is determined by
678 * the "struct files_struct" in the thread group
679 * (or NULL for no owner - BSD locks).
680 *
681 * Lockd stuffs a "host" pointer into this.
682 */
683typedef struct files_struct *fl_owner_t;
684
685struct file_lock_operations {
686 void (*fl_insert)(struct file_lock *); /* lock insertion callback */
687 void (*fl_remove)(struct file_lock *); /* lock removal callback */
688 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
689 void (*fl_release_private)(struct file_lock *);
690};
691
692struct lock_manager_operations {
693 int (*fl_compare_owner)(struct file_lock *, struct file_lock *);
694 void (*fl_notify)(struct file_lock *); /* unblock callback */
695 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
696 void (*fl_release_private)(struct file_lock *);
697 void (*fl_break)(struct file_lock *);
698 int (*fl_mylease)(struct file_lock *, struct file_lock *);
699 int (*fl_change)(struct file_lock **, int);
700};
701
702/* that will die - we need it for nfs_lock_info */
703#include <linux/nfs_fs_i.h>
704
705struct file_lock {
706 struct file_lock *fl_next; /* singly linked list for this inode */
707 struct list_head fl_link; /* doubly linked list of all locks */
708 struct list_head fl_block; /* circular list of blocked processes */
709 fl_owner_t fl_owner;
710 unsigned int fl_pid;
711 wait_queue_head_t fl_wait;
712 struct file *fl_file;
713 unsigned char fl_flags;
714 unsigned char fl_type;
715 loff_t fl_start;
716 loff_t fl_end;
717
718 struct fasync_struct * fl_fasync; /* for lease break notifications */
719 unsigned long fl_break_time; /* for nonblocking lease breaks */
720
721 struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
722 struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
723 union {
724 struct nfs_lock_info nfs_fl;
Trond Myklebust8d0a8a92005-06-22 17:16:32 +0000725 struct nfs4_lock_info nfs4_fl;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700726 } fl_u;
727};
728
729/* The following constant reflects the upper bound of the file/locking space */
730#ifndef OFFSET_MAX
731#define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
732#define OFFSET_MAX INT_LIMIT(loff_t)
733#define OFFT_OFFSET_MAX INT_LIMIT(off_t)
734#endif
735
736extern struct list_head file_lock_list;
737
738#include <linux/fcntl.h>
739
740extern int fcntl_getlk(struct file *, struct flock __user *);
Peter Staubachc2936212005-07-27 11:45:09 -0700741extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
742 struct flock __user *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700743
744#if BITS_PER_LONG == 32
745extern int fcntl_getlk64(struct file *, struct flock64 __user *);
Peter Staubachc2936212005-07-27 11:45:09 -0700746extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
747 struct flock64 __user *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700748#endif
749
750extern void send_sigio(struct fown_struct *fown, int fd, int band);
751extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
752extern int fcntl_getlease(struct file *filp);
753
754/* fs/locks.c */
755extern void locks_init_lock(struct file_lock *);
756extern void locks_copy_lock(struct file_lock *, struct file_lock *);
757extern void locks_remove_posix(struct file *, fl_owner_t);
758extern void locks_remove_flock(struct file *);
759extern struct file_lock *posix_test_lock(struct file *, struct file_lock *);
760extern int posix_lock_file(struct file *, struct file_lock *);
761extern int posix_lock_file_wait(struct file *, struct file_lock *);
762extern void posix_block_lock(struct file_lock *, struct file_lock *);
J. Bruce Fields64a318e2006-01-03 09:55:46 +0100763extern int posix_unblock_lock(struct file *, struct file_lock *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700764extern int posix_locks_deadlock(struct file_lock *, struct file_lock *);
765extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
766extern int __break_lease(struct inode *inode, unsigned int flags);
767extern void lease_get_mtime(struct inode *, struct timespec *time);
768extern int setlease(struct file *, long, struct file_lock **);
769extern int lease_modify(struct file_lock **, int);
770extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
771extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
772extern void steal_locks(fl_owner_t from);
773
774struct fasync_struct {
775 int magic;
776 int fa_fd;
777 struct fasync_struct *fa_next; /* singly linked list */
778 struct file *fa_file;
779};
780
781#define FASYNC_MAGIC 0x4601
782
783/* SMP safe fasync helpers: */
784extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
785/* can be called from interrupts */
786extern void kill_fasync(struct fasync_struct **, int, int);
787/* only for net: no internal synchronization */
788extern void __kill_fasync(struct fasync_struct *, int, int);
789
790extern int f_setown(struct file *filp, unsigned long arg, int force);
791extern void f_delown(struct file *filp);
792extern int send_sigurg(struct fown_struct *fown);
793
794/*
795 * Umount options
796 */
797
798#define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
799#define MNT_DETACH 0x00000002 /* Just detach from the tree */
800#define MNT_EXPIRE 0x00000004 /* Mark for expiry */
801
802extern struct list_head super_blocks;
803extern spinlock_t sb_lock;
804
805#define sb_entry(list) list_entry((list), struct super_block, s_list)
806#define S_BIAS (1<<30)
807struct super_block {
808 struct list_head s_list; /* Keep this first */
809 dev_t s_dev; /* search index; _not_ kdev_t */
810 unsigned long s_blocksize;
811 unsigned long s_old_blocksize;
812 unsigned char s_blocksize_bits;
813 unsigned char s_dirt;
814 unsigned long long s_maxbytes; /* Max file size */
815 struct file_system_type *s_type;
816 struct super_operations *s_op;
817 struct dquot_operations *dq_op;
818 struct quotactl_ops *s_qcop;
819 struct export_operations *s_export_op;
820 unsigned long s_flags;
821 unsigned long s_magic;
822 struct dentry *s_root;
823 struct rw_semaphore s_umount;
824 struct semaphore s_lock;
825 int s_count;
826 int s_syncing;
827 int s_need_sync_fs;
828 atomic_t s_active;
829 void *s_security;
830 struct xattr_handler **s_xattr;
831
832 struct list_head s_inodes; /* all inodes */
833 struct list_head s_dirty; /* dirty inodes */
834 struct list_head s_io; /* parked for writeback */
835 struct hlist_head s_anon; /* anonymous dentries for (nfs) exporting */
836 struct list_head s_files;
837
838 struct block_device *s_bdev;
839 struct list_head s_instances;
840 struct quota_info s_dquot; /* Diskquota specific options */
841
842 int s_frozen;
843 wait_queue_head_t s_wait_unfrozen;
844
845 char s_id[32]; /* Informational name */
846
847 void *s_fs_info; /* Filesystem private info */
848
849 /*
850 * The next field is for VFS *only*. No filesystems have any business
851 * even looking at it. You had been warned.
852 */
853 struct semaphore s_vfs_rename_sem; /* Kludge */
854
855 /* Granuality of c/m/atime in ns.
856 Cannot be worse than a second */
857 u32 s_time_gran;
858};
859
860extern struct timespec current_fs_time(struct super_block *sb);
861
862/*
863 * Snapshotting support.
864 */
865enum {
866 SB_UNFROZEN = 0,
867 SB_FREEZE_WRITE = 1,
868 SB_FREEZE_TRANS = 2,
869};
870
871#define vfs_check_frozen(sb, level) \
872 wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
873
Jens Axboe22e2c502005-06-27 10:55:12 +0200874static inline void get_fs_excl(void)
875{
876 atomic_inc(&current->fs_excl);
877}
878
879static inline void put_fs_excl(void)
880{
881 atomic_dec(&current->fs_excl);
882}
883
884static inline int has_fs_excl(void)
885{
886 return atomic_read(&current->fs_excl);
887}
888
889
Linus Torvalds1da177e2005-04-16 15:20:36 -0700890/*
891 * Superblock locking.
892 */
893static inline void lock_super(struct super_block * sb)
894{
Jens Axboe22e2c502005-06-27 10:55:12 +0200895 get_fs_excl();
Linus Torvalds1da177e2005-04-16 15:20:36 -0700896 down(&sb->s_lock);
897}
898
899static inline void unlock_super(struct super_block * sb)
900{
Jens Axboe22e2c502005-06-27 10:55:12 +0200901 put_fs_excl();
Linus Torvalds1da177e2005-04-16 15:20:36 -0700902 up(&sb->s_lock);
903}
904
905/*
906 * VFS helper functions..
907 */
Christoph Hellwige4543ed2005-11-08 21:35:04 -0800908extern int vfs_permission(struct nameidata *, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700909extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *);
910extern int vfs_mkdir(struct inode *, struct dentry *, int);
911extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t);
912extern int vfs_symlink(struct inode *, struct dentry *, const char *, int);
913extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
914extern int vfs_rmdir(struct inode *, struct dentry *);
915extern int vfs_unlink(struct inode *, struct dentry *);
916extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
917
918/*
919 * VFS dentry helper functions.
920 */
921extern void dentry_unhash(struct dentry *dentry);
922
923/*
Christoph Hellwig8c744fb2005-11-08 21:35:04 -0800924 * VFS file helper functions.
925 */
926extern int file_permission(struct file *, int);
927
928/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700929 * File types
930 *
931 * NOTE! These match bits 12..15 of stat.st_mode
932 * (ie "(i_mode >> 12) & 15").
933 */
934#define DT_UNKNOWN 0
935#define DT_FIFO 1
936#define DT_CHR 2
937#define DT_DIR 4
938#define DT_BLK 6
939#define DT_REG 8
940#define DT_LNK 10
941#define DT_SOCK 12
942#define DT_WHT 14
943
944#define OSYNC_METADATA (1<<0)
945#define OSYNC_DATA (1<<1)
946#define OSYNC_INODE (1<<2)
947int generic_osync_inode(struct inode *, struct address_space *, int);
948
949/*
950 * This is the "filldir" function type, used by readdir() to let
951 * the kernel specify what kind of dirent layout it wants to have.
952 * This allows the kernel to read directories into kernel space or
953 * to have different dirent layouts depending on the binary type.
954 */
955typedef int (*filldir_t)(void *, const char *, int, loff_t, ino_t, unsigned);
956
957struct block_device_operations {
958 int (*open) (struct inode *, struct file *);
959 int (*release) (struct inode *, struct file *);
960 int (*ioctl) (struct inode *, struct file *, unsigned, unsigned long);
Arnd Bergmannbb93e3a2005-06-23 00:10:15 -0700961 long (*unlocked_ioctl) (struct file *, unsigned, unsigned long);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700962 long (*compat_ioctl) (struct file *, unsigned, unsigned long);
Carsten Otte420edbc2005-06-23 22:05:23 -0700963 int (*direct_access) (struct block_device *, sector_t, unsigned long *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700964 int (*media_changed) (struct gendisk *);
965 int (*revalidate_disk) (struct gendisk *);
Christoph Hellwiga885c8c2006-01-08 01:02:50 -0800966 int (*getgeo)(struct block_device *, struct hd_geometry *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700967 struct module *owner;
968};
969
970/*
971 * "descriptor" for what we're up to with a read for sendfile().
972 * This allows us to use the same read code yet
973 * have multiple different users of the data that
974 * we read from a file.
975 *
976 * The simplest case just copies the data to user
977 * mode.
978 */
979typedef struct {
980 size_t written;
981 size_t count;
982 union {
983 char __user * buf;
984 void *data;
985 } arg;
986 int error;
987} read_descriptor_t;
988
989typedef int (*read_actor_t)(read_descriptor_t *, struct page *, unsigned long, unsigned long);
990
991/* These macros are for out of kernel modules to test that
992 * the kernel supports the unlocked_ioctl and compat_ioctl
993 * fields in struct file_operations. */
994#define HAVE_COMPAT_IOCTL 1
995#define HAVE_UNLOCKED_IOCTL 1
996
997/*
998 * NOTE:
999 * read, write, poll, fsync, readv, writev, unlocked_ioctl and compat_ioctl
1000 * can be called without the big kernel lock held in all filesystems.
1001 */
1002struct file_operations {
1003 struct module *owner;
1004 loff_t (*llseek) (struct file *, loff_t, int);
1005 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1006 ssize_t (*aio_read) (struct kiocb *, char __user *, size_t, loff_t);
1007 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1008 ssize_t (*aio_write) (struct kiocb *, const char __user *, size_t, loff_t);
1009 int (*readdir) (struct file *, void *, filldir_t);
1010 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1011 int (*ioctl) (struct inode *, struct file *, unsigned int, unsigned long);
1012 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1013 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1014 int (*mmap) (struct file *, struct vm_area_struct *);
1015 int (*open) (struct inode *, struct file *);
1016 int (*flush) (struct file *);
1017 int (*release) (struct inode *, struct file *);
1018 int (*fsync) (struct file *, struct dentry *, int datasync);
1019 int (*aio_fsync) (struct kiocb *, int datasync);
1020 int (*fasync) (int, struct file *, int);
1021 int (*lock) (struct file *, int, struct file_lock *);
1022 ssize_t (*readv) (struct file *, const struct iovec *, unsigned long, loff_t *);
1023 ssize_t (*writev) (struct file *, const struct iovec *, unsigned long, loff_t *);
1024 ssize_t (*sendfile) (struct file *, loff_t *, size_t, read_actor_t, void *);
1025 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1026 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1027 int (*check_flags)(int);
1028 int (*dir_notify)(struct file *filp, unsigned long arg);
1029 int (*flock) (struct file *, int, struct file_lock *);
1030};
1031
1032struct inode_operations {
1033 int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
1034 struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
1035 int (*link) (struct dentry *,struct inode *,struct dentry *);
1036 int (*unlink) (struct inode *,struct dentry *);
1037 int (*symlink) (struct inode *,struct dentry *,const char *);
1038 int (*mkdir) (struct inode *,struct dentry *,int);
1039 int (*rmdir) (struct inode *,struct dentry *);
1040 int (*mknod) (struct inode *,struct dentry *,int,dev_t);
1041 int (*rename) (struct inode *, struct dentry *,
1042 struct inode *, struct dentry *);
1043 int (*readlink) (struct dentry *, char __user *,int);
Linus Torvaldscc314ee2005-08-19 18:02:56 -07001044 void * (*follow_link) (struct dentry *, struct nameidata *);
1045 void (*put_link) (struct dentry *, struct nameidata *, void *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001046 void (*truncate) (struct inode *);
1047 int (*permission) (struct inode *, int, struct nameidata *);
1048 int (*setattr) (struct dentry *, struct iattr *);
1049 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1050 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1051 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1052 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1053 int (*removexattr) (struct dentry *, const char *);
Badari Pulavartyf6b3ec22006-01-06 00:10:38 -08001054 void (*truncate_range)(struct inode *, loff_t, loff_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001055};
1056
1057struct seq_file;
1058
1059extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1060extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1061extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1062 unsigned long, loff_t *);
1063extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1064 unsigned long, loff_t *);
1065
1066/*
1067 * NOTE: write_inode, delete_inode, clear_inode, put_inode can be called
1068 * without the big kernel lock held in all filesystems.
1069 */
1070struct super_operations {
1071 struct inode *(*alloc_inode)(struct super_block *sb);
1072 void (*destroy_inode)(struct inode *);
1073
1074 void (*read_inode) (struct inode *);
1075
1076 void (*dirty_inode) (struct inode *);
1077 int (*write_inode) (struct inode *, int);
1078 void (*put_inode) (struct inode *);
1079 void (*drop_inode) (struct inode *);
1080 void (*delete_inode) (struct inode *);
1081 void (*put_super) (struct super_block *);
1082 void (*write_super) (struct super_block *);
1083 int (*sync_fs)(struct super_block *sb, int wait);
1084 void (*write_super_lockfs) (struct super_block *);
1085 void (*unlockfs) (struct super_block *);
1086 int (*statfs) (struct super_block *, struct kstatfs *);
1087 int (*remount_fs) (struct super_block *, int *, char *);
1088 void (*clear_inode) (struct inode *);
1089 void (*umount_begin) (struct super_block *);
1090
1091 int (*show_options)(struct seq_file *, struct vfsmount *);
1092
1093 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1094 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1095};
1096
1097/* Inode state bits. Protected by inode_lock. */
1098#define I_DIRTY_SYNC 1 /* Not dirty enough for O_DATASYNC */
1099#define I_DIRTY_DATASYNC 2 /* Data-related inode changes pending */
1100#define I_DIRTY_PAGES 4 /* Data-related inode changes pending */
1101#define __I_LOCK 3
1102#define I_LOCK (1 << __I_LOCK)
1103#define I_FREEING 16
1104#define I_CLEAR 32
1105#define I_NEW 64
Alexander Viro991114c2005-06-23 00:09:01 -07001106#define I_WILL_FREE 128
Linus Torvalds1da177e2005-04-16 15:20:36 -07001107
1108#define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1109
1110extern void __mark_inode_dirty(struct inode *, int);
1111static inline void mark_inode_dirty(struct inode *inode)
1112{
1113 __mark_inode_dirty(inode, I_DIRTY);
1114}
1115
1116static inline void mark_inode_dirty_sync(struct inode *inode)
1117{
1118 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1119}
1120
1121static inline void touch_atime(struct vfsmount *mnt, struct dentry *dentry)
1122{
1123 /* per-mountpoint checks will go here */
1124 update_atime(dentry->d_inode);
1125}
1126
1127static inline void file_accessed(struct file *file)
1128{
1129 if (!(file->f_flags & O_NOATIME))
1130 touch_atime(file->f_vfsmnt, file->f_dentry);
1131}
1132
1133int sync_inode(struct inode *inode, struct writeback_control *wbc);
1134
1135/**
Pavel Pisa4dc3b162005-05-01 08:59:25 -07001136 * struct export_operations - for nfsd to communicate with file systems
1137 * @decode_fh: decode a file handle fragment and return a &struct dentry
1138 * @encode_fh: encode a file handle fragment from a dentry
1139 * @get_name: find the name for a given inode in a given directory
1140 * @get_parent: find the parent of a given directory
1141 * @get_dentry: find a dentry for the inode given a file handle sub-fragment
Randy Dunlap8f2709b2005-11-07 01:01:05 -08001142 * @find_exported_dentry:
1143 * set by the exporting module to a standard helper function.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001144 *
1145 * Description:
1146 * The export_operations structure provides a means for nfsd to communicate
1147 * with a particular exported file system - particularly enabling nfsd and
1148 * the filesystem to co-operate when dealing with file handles.
1149 *
Martin Waitz67be2dd2005-05-01 08:59:26 -07001150 * export_operations contains two basic operation for dealing with file
1151 * handles, decode_fh() and encode_fh(), and allows for some other
1152 * operations to be defined which standard helper routines use to get
1153 * specific information from the filesystem.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001154 *
1155 * nfsd encodes information use to determine which filesystem a filehandle
Martin Waitz67be2dd2005-05-01 08:59:26 -07001156 * applies to in the initial part of the file handle. The remainder, termed
1157 * a file handle fragment, is controlled completely by the filesystem. The
1158 * standard helper routines assume that this fragment will contain one or
1159 * two sub-fragments, one which identifies the file, and one which may be
1160 * used to identify the (a) directory containing the file.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001161 *
1162 * In some situations, nfsd needs to get a dentry which is connected into a
Martin Waitz67be2dd2005-05-01 08:59:26 -07001163 * specific part of the file tree. To allow for this, it passes the
1164 * function acceptable() together with a @context which can be used to see
1165 * if the dentry is acceptable. As there can be multiple dentrys for a
1166 * given file, the filesystem should check each one for acceptability before
1167 * looking for the next. As soon as an acceptable one is found, it should
1168 * be returned.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001169 *
1170 * decode_fh:
Martin Waitz67be2dd2005-05-01 08:59:26 -07001171 * @decode_fh is given a &struct super_block (@sb), a file handle fragment
1172 * (@fh, @fh_len) and an acceptability testing function (@acceptable,
1173 * @context). It should return a &struct dentry which refers to the same
1174 * file that the file handle fragment refers to, and which passes the
1175 * acceptability test. If it cannot, it should return a %NULL pointer if
1176 * the file was found but no acceptable &dentries were available, or a
1177 * %ERR_PTR error code indicating why it couldn't be found (e.g. %ENOENT or
1178 * %ENOMEM).
Linus Torvalds1da177e2005-04-16 15:20:36 -07001179 *
1180 * encode_fh:
Martin Waitz67be2dd2005-05-01 08:59:26 -07001181 * @encode_fh should store in the file handle fragment @fh (using at most
1182 * @max_len bytes) information that can be used by @decode_fh to recover the
1183 * file refered to by the &struct dentry @de. If the @connectable flag is
1184 * set, the encode_fh() should store sufficient information so that a good
1185 * attempt can be made to find not only the file but also it's place in the
1186 * filesystem. This typically means storing a reference to de->d_parent in
1187 * the filehandle fragment. encode_fh() should return the number of bytes
1188 * stored or a negative error code such as %-ENOSPC
Linus Torvalds1da177e2005-04-16 15:20:36 -07001189 *
1190 * get_name:
Martin Waitz67be2dd2005-05-01 08:59:26 -07001191 * @get_name should find a name for the given @child in the given @parent
1192 * directory. The name should be stored in the @name (with the
1193 * understanding that it is already pointing to a a %NAME_MAX+1 sized
1194 * buffer. get_name() should return %0 on success, a negative error code
1195 * or error. @get_name will be called without @parent->i_sem held.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001196 *
1197 * get_parent:
Martin Waitz67be2dd2005-05-01 08:59:26 -07001198 * @get_parent should find the parent directory for the given @child which
1199 * is also a directory. In the event that it cannot be found, or storage
1200 * space cannot be allocated, a %ERR_PTR should be returned.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001201 *
1202 * get_dentry:
Martin Waitz67be2dd2005-05-01 08:59:26 -07001203 * Given a &super_block (@sb) and a pointer to a file-system specific inode
1204 * identifier, possibly an inode number, (@inump) get_dentry() should find
1205 * the identified inode and return a dentry for that inode. Any suitable
1206 * dentry can be returned including, if necessary, a new dentry created with
1207 * d_alloc_root. The caller can then find any other extant dentrys by
1208 * following the d_alias links. If a new dentry was created using
1209 * d_alloc_root, DCACHE_NFSD_DISCONNECTED should be set, and the dentry
1210 * should be d_rehash()ed.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001211 *
Martin Waitz67be2dd2005-05-01 08:59:26 -07001212 * If the inode cannot be found, either a %NULL pointer or an %ERR_PTR code
1213 * can be returned. The @inump will be whatever was passed to
1214 * nfsd_find_fh_dentry() in either the @obj or @parent parameters.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001215 *
1216 * Locking rules:
Martin Waitz67be2dd2005-05-01 08:59:26 -07001217 * get_parent is called with child->d_inode->i_sem down
1218 * get_name is not (which is possibly inconsistent)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001219 */
1220
1221struct export_operations {
1222 struct dentry *(*decode_fh)(struct super_block *sb, __u32 *fh, int fh_len, int fh_type,
1223 int (*acceptable)(void *context, struct dentry *de),
1224 void *context);
1225 int (*encode_fh)(struct dentry *de, __u32 *fh, int *max_len,
1226 int connectable);
1227
1228 /* the following are only called from the filesystem itself */
1229 int (*get_name)(struct dentry *parent, char *name,
1230 struct dentry *child);
1231 struct dentry * (*get_parent)(struct dentry *child);
1232 struct dentry * (*get_dentry)(struct super_block *sb, void *inump);
1233
1234 /* This is set by the exporting module to a standard helper */
1235 struct dentry * (*find_exported_dentry)(
1236 struct super_block *sb, void *obj, void *parent,
1237 int (*acceptable)(void *context, struct dentry *de),
1238 void *context);
1239
1240
1241};
1242
1243extern struct dentry *
1244find_exported_dentry(struct super_block *sb, void *obj, void *parent,
1245 int (*acceptable)(void *context, struct dentry *de),
1246 void *context);
1247
1248struct file_system_type {
1249 const char *name;
1250 int fs_flags;
1251 struct super_block *(*get_sb) (struct file_system_type *, int,
1252 const char *, void *);
1253 void (*kill_sb) (struct super_block *);
1254 struct module *owner;
1255 struct file_system_type * next;
1256 struct list_head fs_supers;
1257};
1258
1259struct super_block *get_sb_bdev(struct file_system_type *fs_type,
1260 int flags, const char *dev_name, void *data,
1261 int (*fill_super)(struct super_block *, void *, int));
1262struct super_block *get_sb_single(struct file_system_type *fs_type,
1263 int flags, void *data,
1264 int (*fill_super)(struct super_block *, void *, int));
1265struct super_block *get_sb_nodev(struct file_system_type *fs_type,
1266 int flags, void *data,
1267 int (*fill_super)(struct super_block *, void *, int));
1268void generic_shutdown_super(struct super_block *sb);
1269void kill_block_super(struct super_block *sb);
1270void kill_anon_super(struct super_block *sb);
1271void kill_litter_super(struct super_block *sb);
1272void deactivate_super(struct super_block *sb);
1273int set_anon_super(struct super_block *s, void *data);
1274struct super_block *sget(struct file_system_type *type,
1275 int (*test)(struct super_block *,void *),
1276 int (*set)(struct super_block *,void *),
1277 void *data);
1278struct super_block *get_sb_pseudo(struct file_system_type *, char *,
1279 struct super_operations *ops, unsigned long);
1280int __put_super(struct super_block *sb);
1281int __put_super_and_need_restart(struct super_block *sb);
1282void unnamed_dev_init(void);
1283
1284/* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1285#define fops_get(fops) \
1286 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1287#define fops_put(fops) \
1288 do { if (fops) module_put((fops)->owner); } while(0)
1289
1290extern int register_filesystem(struct file_system_type *);
1291extern int unregister_filesystem(struct file_system_type *);
1292extern struct vfsmount *kern_mount(struct file_system_type *);
1293extern int may_umount_tree(struct vfsmount *);
1294extern int may_umount(struct vfsmount *);
Ram Paia05964f2005-11-07 17:20:17 -05001295extern void umount_tree(struct vfsmount *, int, struct list_head *);
Ram Paib90fa9a2005-11-07 17:19:50 -05001296extern void release_mounts(struct list_head *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001297extern long do_mount(char *, char *, char *, unsigned long, void *);
Ram Paib90fa9a2005-11-07 17:19:50 -05001298extern struct vfsmount *copy_tree(struct vfsmount *, struct dentry *, int);
1299extern void mnt_set_mountpoint(struct vfsmount *, struct dentry *,
1300 struct vfsmount *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001301
1302extern int vfs_statfs(struct super_block *, struct kstatfs *);
1303
1304#define FLOCK_VERIFY_READ 1
1305#define FLOCK_VERIFY_WRITE 2
1306
1307extern int locks_mandatory_locked(struct inode *);
1308extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1309
1310/*
1311 * Candidates for mandatory locking have the setgid bit set
1312 * but no group execute bit - an otherwise meaningless combination.
1313 */
1314#define MANDATORY_LOCK(inode) \
1315 (IS_MANDLOCK(inode) && ((inode)->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID)
1316
1317static inline int locks_verify_locked(struct inode *inode)
1318{
1319 if (MANDATORY_LOCK(inode))
1320 return locks_mandatory_locked(inode);
1321 return 0;
1322}
1323
1324extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1325
1326static inline int locks_verify_truncate(struct inode *inode,
1327 struct file *filp,
1328 loff_t size)
1329{
1330 if (inode->i_flock && MANDATORY_LOCK(inode))
1331 return locks_mandatory_area(
1332 FLOCK_VERIFY_WRITE, inode, filp,
1333 size < inode->i_size ? size : inode->i_size,
1334 (size < inode->i_size ? inode->i_size - size
1335 : size - inode->i_size)
1336 );
1337 return 0;
1338}
1339
1340static inline int break_lease(struct inode *inode, unsigned int mode)
1341{
1342 if (inode->i_flock)
1343 return __break_lease(inode, mode);
1344 return 0;
1345}
1346
1347/* fs/open.c */
1348
NeilBrown4a301312006-01-08 01:02:39 -08001349extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
1350 struct file *filp);
Miklos Szeredie922efc2005-09-06 15:18:25 -07001351extern long do_sys_open(const char __user *filename, int flags, int mode);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001352extern struct file *filp_open(const char *, int, int);
1353extern struct file * dentry_open(struct dentry *, struct vfsmount *, int);
1354extern int filp_close(struct file *, fl_owner_t id);
1355extern char * getname(const char __user *);
1356
1357/* fs/dcache.c */
1358extern void __init vfs_caches_init_early(void);
1359extern void __init vfs_caches_init(unsigned long);
1360
1361#define __getname() kmem_cache_alloc(names_cachep, SLAB_KERNEL)
1362#define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
1363#ifndef CONFIG_AUDITSYSCALL
1364#define putname(name) __putname(name)
1365#else
1366extern void putname(const char *name);
1367#endif
1368
1369extern int register_blkdev(unsigned int, const char *);
1370extern int unregister_blkdev(unsigned int, const char *);
1371extern struct block_device *bdget(dev_t);
1372extern void bd_set_size(struct block_device *, loff_t size);
1373extern void bd_forget(struct inode *inode);
1374extern void bdput(struct block_device *);
1375extern struct block_device *open_by_devnum(dev_t, unsigned);
1376extern struct file_operations def_blk_fops;
1377extern struct address_space_operations def_blk_aops;
1378extern struct file_operations def_chr_fops;
1379extern struct file_operations bad_sock_fops;
1380extern struct file_operations def_fifo_fops;
1381extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
1382extern int blkdev_ioctl(struct inode *, struct file *, unsigned, unsigned long);
1383extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
1384extern int blkdev_get(struct block_device *, mode_t, unsigned);
1385extern int blkdev_put(struct block_device *);
1386extern int bd_claim(struct block_device *, void *);
1387extern void bd_release(struct block_device *);
1388
1389/* fs/char_dev.c */
1390extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
1391extern int register_chrdev_region(dev_t, unsigned, const char *);
1392extern int register_chrdev(unsigned int, const char *,
1393 struct file_operations *);
1394extern int unregister_chrdev(unsigned int, const char *);
1395extern void unregister_chrdev_region(dev_t, unsigned);
1396extern int chrdev_open(struct inode *, struct file *);
1397
1398/* fs/block_dev.c */
1399#define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
1400extern const char *__bdevname(dev_t, char *buffer);
1401extern const char *bdevname(struct block_device *bdev, char *buffer);
1402extern struct block_device *lookup_bdev(const char *);
1403extern struct block_device *open_bdev_excl(const char *, int, void *);
1404extern void close_bdev_excl(struct block_device *);
1405
1406extern void init_special_inode(struct inode *, umode_t, dev_t);
1407
1408/* Invalid inode operations -- fs/bad_inode.c */
1409extern void make_bad_inode(struct inode *);
1410extern int is_bad_inode(struct inode *);
1411
1412extern struct file_operations read_fifo_fops;
1413extern struct file_operations write_fifo_fops;
1414extern struct file_operations rdwr_fifo_fops;
1415extern struct file_operations read_pipe_fops;
1416extern struct file_operations write_pipe_fops;
1417extern struct file_operations rdwr_pipe_fops;
1418
1419extern int fs_may_remount_ro(struct super_block *);
1420
1421/*
1422 * return READ, READA, or WRITE
1423 */
1424#define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
1425
1426/*
1427 * return data direction, READ or WRITE
1428 */
1429#define bio_data_dir(bio) ((bio)->bi_rw & 1)
1430
1431extern int check_disk_change(struct block_device *);
1432extern int invalidate_inodes(struct super_block *);
Christoph Hellwig2ef41632005-05-05 16:15:59 -07001433extern int __invalidate_device(struct block_device *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001434extern int invalidate_partition(struct gendisk *, int);
1435unsigned long invalidate_mapping_pages(struct address_space *mapping,
1436 pgoff_t start, pgoff_t end);
1437unsigned long invalidate_inode_pages(struct address_space *mapping);
1438static inline void invalidate_remote_inode(struct inode *inode)
1439{
1440 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
1441 S_ISLNK(inode->i_mode))
1442 invalidate_inode_pages(inode->i_mapping);
1443}
1444extern int invalidate_inode_pages2(struct address_space *mapping);
1445extern int invalidate_inode_pages2_range(struct address_space *mapping,
1446 pgoff_t start, pgoff_t end);
1447extern int write_inode_now(struct inode *, int);
1448extern int filemap_fdatawrite(struct address_space *);
1449extern int filemap_flush(struct address_space *);
1450extern int filemap_fdatawait(struct address_space *);
1451extern int filemap_write_and_wait(struct address_space *mapping);
1452extern int filemap_write_and_wait_range(struct address_space *mapping,
1453 loff_t lstart, loff_t lend);
1454extern void sync_supers(void);
1455extern void sync_filesystems(int wait);
1456extern void emergency_sync(void);
1457extern void emergency_remount(void);
1458extern int do_remount_sb(struct super_block *sb, int flags,
1459 void *data, int force);
1460extern sector_t bmap(struct inode *, sector_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001461extern int notify_change(struct dentry *, struct iattr *);
1462extern int permission(struct inode *, int, struct nameidata *);
1463extern int generic_permission(struct inode *, int,
1464 int (*check_acl)(struct inode *, int));
1465
1466extern int get_write_access(struct inode *);
1467extern int deny_write_access(struct file *);
1468static inline void put_write_access(struct inode * inode)
1469{
1470 atomic_dec(&inode->i_writecount);
1471}
1472static inline void allow_write_access(struct file *file)
1473{
1474 if (file)
1475 atomic_inc(&file->f_dentry->d_inode->i_writecount);
1476}
1477extern int do_pipe(int *);
1478
1479extern int open_namei(const char *, int, int, struct nameidata *);
1480extern int may_open(struct nameidata *, int, int);
1481
1482extern int kernel_read(struct file *, unsigned long, char *, unsigned long);
1483extern struct file * open_exec(const char *);
1484
1485/* fs/dcache.c -- generic fs support functions */
1486extern int is_subdir(struct dentry *, struct dentry *);
1487extern ino_t find_inode_number(struct dentry *, struct qstr *);
1488
1489#include <linux/err.h>
1490
1491/* needed for stackable file system support */
1492extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
1493
1494extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
1495
1496extern void inode_init_once(struct inode *);
1497extern void iput(struct inode *);
1498extern struct inode * igrab(struct inode *);
1499extern ino_t iunique(struct super_block *, ino_t);
1500extern int inode_needs_sync(struct inode *inode);
1501extern void generic_delete_inode(struct inode *inode);
Mark Fashehcb2c0232005-07-07 17:56:03 -07001502extern void generic_drop_inode(struct inode *inode);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001503
Anton Altaparmakov88bd5122005-07-13 01:10:44 -07001504extern struct inode *ilookup5_nowait(struct super_block *sb,
1505 unsigned long hashval, int (*test)(struct inode *, void *),
1506 void *data);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001507extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
1508 int (*test)(struct inode *, void *), void *data);
1509extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
1510
1511extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
1512extern struct inode * iget_locked(struct super_block *, unsigned long);
1513extern void unlock_new_inode(struct inode *);
1514
1515static inline struct inode *iget(struct super_block *sb, unsigned long ino)
1516{
1517 struct inode *inode = iget_locked(sb, ino);
1518
1519 if (inode && (inode->i_state & I_NEW)) {
1520 sb->s_op->read_inode(inode);
1521 unlock_new_inode(inode);
1522 }
1523
1524 return inode;
1525}
1526
1527extern void __iget(struct inode * inode);
1528extern void clear_inode(struct inode *);
1529extern void destroy_inode(struct inode *);
1530extern struct inode *new_inode(struct super_block *);
1531extern int remove_suid(struct dentry *);
1532extern void remove_dquot_ref(struct super_block *, int, struct list_head *);
1533extern struct semaphore iprune_sem;
1534
1535extern void __insert_inode_hash(struct inode *, unsigned long hashval);
1536extern void remove_inode_hash(struct inode *);
1537static inline void insert_inode_hash(struct inode *inode) {
1538 __insert_inode_hash(inode, inode->i_ino);
1539}
1540
1541extern struct file * get_empty_filp(void);
1542extern void file_move(struct file *f, struct list_head *list);
1543extern void file_kill(struct file *f);
1544struct bio;
1545extern void submit_bio(int, struct bio *);
1546extern int bdev_read_only(struct block_device *);
1547extern int set_blocksize(struct block_device *, int);
1548extern int sb_set_blocksize(struct super_block *, int);
1549extern int sb_min_blocksize(struct super_block *, int);
1550
1551extern int generic_file_mmap(struct file *, struct vm_area_struct *);
1552extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
1553extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
1554extern int file_send_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
1555extern ssize_t generic_file_read(struct file *, char __user *, size_t, loff_t *);
1556int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
1557extern ssize_t generic_file_write(struct file *, const char __user *, size_t, loff_t *);
1558extern ssize_t generic_file_aio_read(struct kiocb *, char __user *, size_t, loff_t);
1559extern ssize_t __generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t *);
1560extern ssize_t generic_file_aio_write(struct kiocb *, const char __user *, size_t, loff_t);
1561extern ssize_t generic_file_aio_write_nolock(struct kiocb *, const struct iovec *,
1562 unsigned long, loff_t *);
1563extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
1564 unsigned long *, loff_t, loff_t *, size_t, size_t);
1565extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
1566 unsigned long, loff_t, loff_t *, size_t, ssize_t);
1567extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
1568extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
1569ssize_t generic_file_write_nolock(struct file *file, const struct iovec *iov,
1570 unsigned long nr_segs, loff_t *ppos);
1571extern ssize_t generic_file_sendfile(struct file *, loff_t *, size_t, read_actor_t, void *);
1572extern void do_generic_mapping_read(struct address_space *mapping,
1573 struct file_ra_state *, struct file *,
1574 loff_t *, read_descriptor_t *, read_actor_t);
1575extern void
1576file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001577extern ssize_t generic_file_readv(struct file *filp, const struct iovec *iov,
1578 unsigned long nr_segs, loff_t *ppos);
1579ssize_t generic_file_writev(struct file *filp, const struct iovec *iov,
1580 unsigned long nr_segs, loff_t *ppos);
1581extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
1582extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
1583extern loff_t remote_llseek(struct file *file, loff_t offset, int origin);
1584extern int generic_file_open(struct inode * inode, struct file * filp);
1585extern int nonseekable_open(struct inode * inode, struct file * filp);
1586
Carsten Otteceffc072005-06-23 22:05:25 -07001587#ifdef CONFIG_FS_XIP
Carsten Otteeb6fe0c2005-06-23 22:05:28 -07001588extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
1589 loff_t *ppos);
Carsten Otteceffc072005-06-23 22:05:25 -07001590extern ssize_t xip_file_sendfile(struct file *in_file, loff_t *ppos,
1591 size_t count, read_actor_t actor,
1592 void *target);
1593extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
Carsten Otteeb6fe0c2005-06-23 22:05:28 -07001594extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
1595 size_t len, loff_t *ppos);
Carsten Otteceffc072005-06-23 22:05:25 -07001596extern int xip_truncate_page(struct address_space *mapping, loff_t from);
Carsten Otte6d791252005-06-23 22:05:26 -07001597#else
1598static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
1599{
1600 return 0;
1601}
Carsten Otteceffc072005-06-23 22:05:25 -07001602#endif
1603
Linus Torvalds1da177e2005-04-16 15:20:36 -07001604static inline void do_generic_file_read(struct file * filp, loff_t *ppos,
1605 read_descriptor_t * desc,
1606 read_actor_t actor)
1607{
1608 do_generic_mapping_read(filp->f_mapping,
1609 &filp->f_ra,
1610 filp,
1611 ppos,
1612 desc,
1613 actor);
1614}
1615
1616ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
1617 struct block_device *bdev, const struct iovec *iov, loff_t offset,
1618 unsigned long nr_segs, get_blocks_t get_blocks, dio_iodone_t end_io,
1619 int lock_type);
1620
1621enum {
1622 DIO_LOCKING = 1, /* need locking between buffered and direct access */
1623 DIO_NO_LOCKING, /* bdev; no locking at all between buffered/direct */
1624 DIO_OWN_LOCKING, /* filesystem locks buffered and direct internally */
1625};
1626
1627static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
1628 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1629 loff_t offset, unsigned long nr_segs, get_blocks_t get_blocks,
1630 dio_iodone_t end_io)
1631{
1632 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1633 nr_segs, get_blocks, end_io, DIO_LOCKING);
1634}
1635
1636static inline ssize_t blockdev_direct_IO_no_locking(int rw, struct kiocb *iocb,
1637 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1638 loff_t offset, unsigned long nr_segs, get_blocks_t get_blocks,
1639 dio_iodone_t end_io)
1640{
1641 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1642 nr_segs, get_blocks, end_io, DIO_NO_LOCKING);
1643}
1644
1645static inline ssize_t blockdev_direct_IO_own_locking(int rw, struct kiocb *iocb,
1646 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1647 loff_t offset, unsigned long nr_segs, get_blocks_t get_blocks,
1648 dio_iodone_t end_io)
1649{
1650 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1651 nr_segs, get_blocks, end_io, DIO_OWN_LOCKING);
1652}
1653
1654extern struct file_operations generic_ro_fops;
1655
1656#define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
1657
1658extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
1659extern int vfs_follow_link(struct nameidata *, const char *);
1660extern int page_readlink(struct dentry *, char __user *, int);
Linus Torvaldscc314ee2005-08-19 18:02:56 -07001661extern void *page_follow_link_light(struct dentry *, struct nameidata *);
1662extern void page_put_link(struct dentry *, struct nameidata *, void *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001663extern int page_symlink(struct inode *inode, const char *symname, int len);
1664extern struct inode_operations page_symlink_inode_operations;
1665extern int generic_readlink(struct dentry *, char __user *, int);
1666extern void generic_fillattr(struct inode *, struct kstat *);
1667extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1668void inode_add_bytes(struct inode *inode, loff_t bytes);
1669void inode_sub_bytes(struct inode *inode, loff_t bytes);
1670loff_t inode_get_bytes(struct inode *inode);
1671void inode_set_bytes(struct inode *inode, loff_t bytes);
1672
1673extern int vfs_readdir(struct file *, filldir_t, void *);
1674
1675extern int vfs_stat(char __user *, struct kstat *);
1676extern int vfs_lstat(char __user *, struct kstat *);
1677extern int vfs_fstat(unsigned int, struct kstat *);
1678
1679extern int vfs_ioctl(struct file *, unsigned int, unsigned int, unsigned long);
1680
1681extern struct file_system_type *get_fs_type(const char *name);
1682extern struct super_block *get_super(struct block_device *);
1683extern struct super_block *user_get_super(dev_t);
1684extern void drop_super(struct super_block *sb);
1685
1686extern int dcache_dir_open(struct inode *, struct file *);
1687extern int dcache_dir_close(struct inode *, struct file *);
1688extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
1689extern int dcache_readdir(struct file *, void *, filldir_t);
1690extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1691extern int simple_statfs(struct super_block *, struct kstatfs *);
1692extern int simple_link(struct dentry *, struct inode *, struct dentry *);
1693extern int simple_unlink(struct inode *, struct dentry *);
1694extern int simple_rmdir(struct inode *, struct dentry *);
1695extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1696extern int simple_sync_file(struct file *, struct dentry *, int);
1697extern int simple_empty(struct dentry *);
1698extern int simple_readpage(struct file *file, struct page *page);
1699extern int simple_prepare_write(struct file *file, struct page *page,
1700 unsigned offset, unsigned to);
1701extern int simple_commit_write(struct file *file, struct page *page,
1702 unsigned offset, unsigned to);
1703
1704extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
1705extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
1706extern struct file_operations simple_dir_operations;
1707extern struct inode_operations simple_dir_inode_operations;
1708struct tree_descr { char *name; struct file_operations *ops; int mode; };
1709struct dentry *d_alloc_name(struct dentry *, const char *);
1710extern int simple_fill_super(struct super_block *, int, struct tree_descr *);
1711extern int simple_pin_fs(char *name, struct vfsmount **mount, int *count);
1712extern void simple_release_fs(struct vfsmount **mount, int *count);
1713
1714extern ssize_t simple_read_from_buffer(void __user *, size_t, loff_t *, const void *, size_t);
1715
1716extern int inode_change_ok(struct inode *, struct iattr *);
1717extern int __must_check inode_setattr(struct inode *, struct iattr *);
1718
1719extern void inode_update_time(struct inode *inode, int ctime_too);
1720
1721static inline ino_t parent_ino(struct dentry *dentry)
1722{
1723 ino_t res;
1724
1725 spin_lock(&dentry->d_lock);
1726 res = dentry->d_parent->d_inode->i_ino;
1727 spin_unlock(&dentry->d_lock);
1728 return res;
1729}
1730
1731/* kernel/fork.c */
1732extern int unshare_files(void);
1733
1734/* Transaction based IO helpers */
1735
1736/*
1737 * An argresp is stored in an allocated page and holds the
1738 * size of the argument or response, along with its content
1739 */
1740struct simple_transaction_argresp {
1741 ssize_t size;
1742 char data[0];
1743};
1744
1745#define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
1746
1747char *simple_transaction_get(struct file *file, const char __user *buf,
1748 size_t size);
1749ssize_t simple_transaction_read(struct file *file, char __user *buf,
1750 size_t size, loff_t *pos);
1751int simple_transaction_release(struct inode *inode, struct file *file);
1752
1753static inline void simple_transaction_set(struct file *file, size_t n)
1754{
1755 struct simple_transaction_argresp *ar = file->private_data;
1756
1757 BUG_ON(n > SIMPLE_TRANSACTION_LIMIT);
1758
1759 /*
1760 * The barrier ensures that ar->size will really remain zero until
1761 * ar->data is ready for reading.
1762 */
1763 smp_mb();
1764 ar->size = n;
1765}
1766
Arnd Bergmannacaefc22005-05-18 14:40:59 +02001767/*
1768 * simple attribute files
1769 *
1770 * These attributes behave similar to those in sysfs:
1771 *
1772 * Writing to an attribute immediately sets a value, an open file can be
1773 * written to multiple times.
1774 *
1775 * Reading from an attribute creates a buffer from the value that might get
1776 * read with multiple read calls. When the attribute has been read
1777 * completely, no further read calls are possible until the file is opened
1778 * again.
1779 *
1780 * All attributes contain a text representation of a numeric value
1781 * that are accessed with the get() and set() functions.
1782 */
1783#define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
1784static int __fops ## _open(struct inode *inode, struct file *file) \
1785{ \
1786 __simple_attr_check_format(__fmt, 0ull); \
1787 return simple_attr_open(inode, file, __get, __set, __fmt); \
1788} \
1789static struct file_operations __fops = { \
1790 .owner = THIS_MODULE, \
1791 .open = __fops ## _open, \
1792 .release = simple_attr_close, \
1793 .read = simple_attr_read, \
1794 .write = simple_attr_write, \
1795};
1796
1797static inline void __attribute__((format(printf, 1, 2)))
1798__simple_attr_check_format(const char *fmt, ...)
1799{
1800 /* don't do anything, just let the compiler check the arguments; */
1801}
1802
1803int simple_attr_open(struct inode *inode, struct file *file,
1804 u64 (*get)(void *), void (*set)(void *, u64),
1805 const char *fmt);
1806int simple_attr_close(struct inode *inode, struct file *file);
1807ssize_t simple_attr_read(struct file *file, char __user *buf,
1808 size_t len, loff_t *ppos);
1809ssize_t simple_attr_write(struct file *file, const char __user *buf,
1810 size_t len, loff_t *ppos);
1811
1812
Linus Torvalds1da177e2005-04-16 15:20:36 -07001813#ifdef CONFIG_SECURITY
1814static inline char *alloc_secdata(void)
1815{
1816 return (char *)get_zeroed_page(GFP_KERNEL);
1817}
1818
1819static inline void free_secdata(void *secdata)
1820{
1821 free_page((unsigned long)secdata);
1822}
1823#else
1824static inline char *alloc_secdata(void)
1825{
1826 return (char *)1;
1827}
1828
1829static inline void free_secdata(void *secdata)
1830{ }
1831#endif /* CONFIG_SECURITY */
1832
1833#endif /* __KERNEL__ */
1834#endif /* _LINUX_FS_H */