]> bbs.cooldavid.org Git - net-next-2.6.git/blame - include/linux/fs.h
[PATCH] dm-md-dependency-tree-in-sysfs-holders-slaves-subdirectory-tidy
[net-next-2.6.git] / include / linux / fs.h
CommitLineData
1da177e4
LT
1#ifndef _LINUX_FS_H
2#define _LINUX_FS_H
3
4/*
5 * This file has definitions for some important file table
6 * structures etc.
7 */
8
9#include <linux/config.h>
10#include <linux/limits.h>
11#include <linux/ioctl.h>
12
13/*
14 * It's silly to have NR_OPEN bigger than NR_FILE, but you can change
15 * the file limit at runtime and only root can increase the per-process
16 * nr_file rlimit, so it's safe to set up a ridiculously high absolute
17 * upper limit on files-per-process.
18 *
19 * Some programs (notably those using select()) may have to be
20 * recompiled to take full advantage of the new limits..
21 */
22
23/* Fixed constants first: */
24#undef NR_OPEN
25#define NR_OPEN (1024*1024) /* Absolute upper limit on fd num */
26#define INR_OPEN 1024 /* Initial setting for nfile rlimits */
27
28#define BLOCK_SIZE_BITS 10
29#define BLOCK_SIZE (1<<BLOCK_SIZE_BITS)
30
31/* And dynamically-tunable limits and defaults: */
32struct files_stat_struct {
33 int nr_files; /* read only */
34 int nr_free_files; /* read only */
35 int max_files; /* tunable */
36};
37extern struct files_stat_struct files_stat;
529bf6be 38extern int get_max_files(void);
1da177e4
LT
39
40struct inodes_stat_t {
41 int nr_inodes;
42 int nr_unused;
43 int dummy[5];
44};
45extern struct inodes_stat_t inodes_stat;
46
47extern int leases_enable, lease_break_time;
48
49#ifdef CONFIG_DNOTIFY
50extern int dir_notify_enable;
51#endif
52
53#define NR_FILE 8192 /* this can well be larger on a larger system */
54
55#define MAY_EXEC 1
56#define MAY_WRITE 2
57#define MAY_READ 4
58#define MAY_APPEND 8
59
60#define FMODE_READ 1
61#define FMODE_WRITE 2
62
63/* Internal kernel extensions */
64#define FMODE_LSEEK 4
65#define FMODE_PREAD 8
66#define FMODE_PWRITE FMODE_PREAD /* These go hand in hand */
67
b500531e
OD
68/* File is being opened for execution. Primary users of this flag are
69 distributed filesystems that can use it to achieve correct ETXTBUSY
70 behavior for cross-node execution/opening_for_writing of files */
71#define FMODE_EXEC 16
72
1da177e4
LT
73#define RW_MASK 1
74#define RWA_MASK 2
75#define READ 0
76#define WRITE 1
77#define READA 2 /* read-ahead - don't block if no resources */
a7662236 78#define SWRITE 3 /* for ll_rw_block() - wait for buffer lock */
1da177e4
LT
79#define SPECIAL 4 /* For non-blockdevice requests in request queue */
80#define READ_SYNC (READ | (1 << BIO_RW_SYNC))
81#define WRITE_SYNC (WRITE | (1 << BIO_RW_SYNC))
82#define WRITE_BARRIER ((1 << BIO_RW) | (1 << BIO_RW_BARRIER))
83
84#define SEL_IN 1
85#define SEL_OUT 2
86#define SEL_EX 4
87
88/* public flags for file_system_type */
89#define FS_REQUIRES_DEV 1
90#define FS_BINARY_MOUNTDATA 2
91#define FS_REVAL_DOT 16384 /* Check the paths ".", ".." for staleness */
92#define FS_ODD_RENAME 32768 /* Temporary stuff; will go away as soon
93 * as nfs_rename() will be cleaned up
94 */
95/*
96 * These are the fs-independent mount-flags: up to 32 flags are supported
97 */
98#define MS_RDONLY 1 /* Mount read-only */
99#define MS_NOSUID 2 /* Ignore suid and sgid bits */
100#define MS_NODEV 4 /* Disallow access to device special files */
101#define MS_NOEXEC 8 /* Disallow program execution */
102#define MS_SYNCHRONOUS 16 /* Writes are synced at once */
103#define MS_REMOUNT 32 /* Alter flags of a mounted FS */
104#define MS_MANDLOCK 64 /* Allow mandatory locks on an FS */
105#define MS_DIRSYNC 128 /* Directory modifications are synchronous */
106#define MS_NOATIME 1024 /* Do not update access times. */
107#define MS_NODIRATIME 2048 /* Do not update directory access times */
108#define MS_BIND 4096
109#define MS_MOVE 8192
110#define MS_REC 16384
9b04c997
TT
111#define MS_VERBOSE 32768 /* War is peace. Verbosity is silence.
112 MS_VERBOSE is deprecated. */
113#define MS_SILENT 32768
bf066c7d 114#define MS_POSIXACL (1<<16) /* VFS does not apply the umask */
9676f0c6 115#define MS_UNBINDABLE (1<<17) /* change to unbindable */
07b20889 116#define MS_PRIVATE (1<<18) /* change to private */
a58b0eb8 117#define MS_SLAVE (1<<19) /* change to slave */
03e06e68 118#define MS_SHARED (1<<20) /* change to shared */
1da177e4
LT
119#define MS_ACTIVE (1<<30)
120#define MS_NOUSER (1<<31)
121
122/*
123 * Superblock flags that can be altered by MS_REMOUNT
124 */
fc33a7bb 125#define MS_RMT_MASK (MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK)
1da177e4
LT
126
127/*
128 * Old magic mount flag and mask
129 */
130#define MS_MGC_VAL 0xC0ED0000
131#define MS_MGC_MSK 0xffff0000
132
133/* Inode flags - they have nothing to superblock flags now */
134
135#define S_SYNC 1 /* Writes are synced at once */
136#define S_NOATIME 2 /* Do not update access times */
137#define S_APPEND 4 /* Append-only file */
138#define S_IMMUTABLE 8 /* Immutable file */
139#define S_DEAD 16 /* removed, but still open directory */
140#define S_NOQUOTA 32 /* Inode is not counted to quota */
141#define S_DIRSYNC 64 /* Directory modifications are synchronous */
142#define S_NOCMTIME 128 /* Do not update file c/mtime */
143#define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
144#define S_PRIVATE 512 /* Inode is fs-internal */
145
146/*
147 * Note that nosuid etc flags are inode-specific: setting some file-system
148 * flags just means all the inodes inherit those flags by default. It might be
149 * possible to override it selectively if you really wanted to with some
150 * ioctl() that is not currently implemented.
151 *
152 * Exception: MS_RDONLY is always applied to the entire file system.
153 *
154 * Unfortunately, it is possible to change a filesystems flags with it mounted
155 * with files in use. This means that all of the inodes will not have their
156 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
157 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
158 */
159#define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))
160
161#define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
162#define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
163 ((inode)->i_flags & S_SYNC))
164#define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
165 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
166#define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
167
168#define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
169#define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
170#define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1da177e4
LT
171#define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
172
173#define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
174#define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
175#define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
176#define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
177
178/* the read-only stuff doesn't really belong here, but any other place is
179 probably as bad and I don't want to create yet another include file. */
180
181#define BLKROSET _IO(0x12,93) /* set device read-only (0 = read-write) */
182#define BLKROGET _IO(0x12,94) /* get read-only status (0 = read_write) */
183#define BLKRRPART _IO(0x12,95) /* re-read partition table */
184#define BLKGETSIZE _IO(0x12,96) /* return device size /512 (long *arg) */
185#define BLKFLSBUF _IO(0x12,97) /* flush buffer cache */
186#define BLKRASET _IO(0x12,98) /* set read ahead for block device */
187#define BLKRAGET _IO(0x12,99) /* get current read ahead setting */
188#define BLKFRASET _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
189#define BLKFRAGET _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
190#define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
191#define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
192#define BLKSSZGET _IO(0x12,104)/* get block device sector size */
193#if 0
194#define BLKPG _IO(0x12,105)/* See blkpg.h */
195
196/* Some people are morons. Do not use sizeof! */
197
198#define BLKELVGET _IOR(0x12,106,size_t)/* elevator get */
199#define BLKELVSET _IOW(0x12,107,size_t)/* elevator set */
200/* This was here just to show that the number is taken -
201 probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
202#endif
203/* A jump here: 108-111 have been used for various private purposes. */
204#define BLKBSZGET _IOR(0x12,112,size_t)
205#define BLKBSZSET _IOW(0x12,113,size_t)
206#define BLKGETSIZE64 _IOR(0x12,114,size_t) /* return device size in bytes (u64 *arg) */
2056a782
JA
207#define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup)
208#define BLKTRACESTART _IO(0x12,116)
209#define BLKTRACESTOP _IO(0x12,117)
210#define BLKTRACETEARDOWN _IO(0x12,118)
1da177e4
LT
211
212#define BMAP_IOCTL 1 /* obsolete - kept for compatibility */
213#define FIBMAP _IO(0x00,1) /* bmap access */
214#define FIGETBSZ _IO(0x00,2) /* get the block size used for bmap */
215
216#ifdef __KERNEL__
217
218#include <linux/linkage.h>
219#include <linux/wait.h>
220#include <linux/types.h>
221#include <linux/kdev_t.h>
222#include <linux/dcache.h>
223#include <linux/stat.h>
224#include <linux/cache.h>
225#include <linux/kobject.h>
226#include <linux/list.h>
227#include <linux/radix-tree.h>
228#include <linux/prio_tree.h>
229#include <linux/init.h>
22e2c507 230#include <linux/sched.h>
1b1dcc1b 231#include <linux/mutex.h>
1da177e4
LT
232
233#include <asm/atomic.h>
234#include <asm/semaphore.h>
235#include <asm/byteorder.h>
236
a885c8c4 237struct hd_geometry;
1da177e4
LT
238struct iovec;
239struct nameidata;
92198f7e 240struct kiocb;
1da177e4
LT
241struct pipe_inode_info;
242struct poll_table_struct;
243struct kstatfs;
244struct vm_area_struct;
245struct vfsmount;
246
1da177e4
LT
247extern void __init inode_init(unsigned long);
248extern void __init inode_init_early(void);
249extern void __init mnt_init(unsigned long);
250extern void __init files_init(unsigned long);
251
252struct buffer_head;
253typedef int (get_block_t)(struct inode *inode, sector_t iblock,
254 struct buffer_head *bh_result, int create);
92198f7e 255typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
1da177e4
LT
256 ssize_t bytes, void *private);
257
258/*
259 * Attribute flags. These should be or-ed together to figure out what
260 * has been changed!
261 */
262#define ATTR_MODE 1
263#define ATTR_UID 2
264#define ATTR_GID 4
265#define ATTR_SIZE 8
266#define ATTR_ATIME 16
267#define ATTR_MTIME 32
268#define ATTR_CTIME 64
269#define ATTR_ATIME_SET 128
270#define ATTR_MTIME_SET 256
271#define ATTR_FORCE 512 /* Not a change, but a change it */
272#define ATTR_ATTR_FLAG 1024
273#define ATTR_KILL_SUID 2048
274#define ATTR_KILL_SGID 4096
cc4e69de 275#define ATTR_FILE 8192
1da177e4
LT
276
277/*
278 * This is the Inode Attributes structure, used for notify_change(). It
279 * uses the above definitions as flags, to know which values have changed.
280 * Also, in this manner, a Filesystem can look at only the values it cares
281 * about. Basically, these are the attributes that the VFS layer can
282 * request to change from the FS layer.
283 *
284 * Derek Atkins <warlord@MIT.EDU> 94-10-20
285 */
286struct iattr {
287 unsigned int ia_valid;
288 umode_t ia_mode;
289 uid_t ia_uid;
290 gid_t ia_gid;
291 loff_t ia_size;
292 struct timespec ia_atime;
293 struct timespec ia_mtime;
294 struct timespec ia_ctime;
cc4e69de
MS
295
296 /*
297 * Not an attribute, but an auxilary info for filesystems wanting to
298 * implement an ftruncate() like method. NOTE: filesystem should
299 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
300 */
301 struct file *ia_file;
1da177e4
LT
302};
303
1da177e4
LT
304/*
305 * Includes for diskquotas.
306 */
307#include <linux/quota.h>
308
994fc28c
ZB
309/**
310 * enum positive_aop_returns - aop return codes with specific semantics
311 *
312 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
313 * completed, that the page is still locked, and
314 * should be considered active. The VM uses this hint
315 * to return the page to the active list -- it won't
316 * be a candidate for writeback again in the near
317 * future. Other callers must be careful to unlock
318 * the page if they get this return. Returned by
319 * writepage();
320 *
321 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
322 * unlocked it and the page might have been truncated.
323 * The caller should back up to acquiring a new page and
324 * trying again. The aop will be taking reasonable
325 * precautions not to livelock. If the caller held a page
326 * reference, it should drop it before retrying. Returned
327 * by readpage(), prepare_write(), and commit_write().
328 *
329 * address_space_operation functions return these large constants to indicate
330 * special semantics to the caller. These are much larger than the bytes in a
331 * page to allow for functions that return the number of bytes operated on in a
332 * given page.
333 */
334
335enum positive_aop_returns {
336 AOP_WRITEPAGE_ACTIVATE = 0x80000,
337 AOP_TRUNCATED_PAGE = 0x80001,
338};
339
1da177e4
LT
340/*
341 * oh the beauties of C type declarations.
342 */
343struct page;
344struct address_space;
345struct writeback_control;
1da177e4
LT
346
347struct address_space_operations {
348 int (*writepage)(struct page *page, struct writeback_control *wbc);
349 int (*readpage)(struct file *, struct page *);
3978d717 350 void (*sync_page)(struct page *);
1da177e4
LT
351
352 /* Write back some dirty pages from this mapping. */
353 int (*writepages)(struct address_space *, struct writeback_control *);
354
4741c9fd 355 /* Set a page dirty. Return true if this dirtied it */
1da177e4
LT
356 int (*set_page_dirty)(struct page *page);
357
358 int (*readpages)(struct file *filp, struct address_space *mapping,
359 struct list_head *pages, unsigned nr_pages);
360
361 /*
362 * ext3 requires that a successful prepare_write() call be followed
363 * by a commit_write() call - they must be balanced
364 */
365 int (*prepare_write)(struct file *, struct page *, unsigned, unsigned);
366 int (*commit_write)(struct file *, struct page *, unsigned, unsigned);
367 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
368 sector_t (*bmap)(struct address_space *, sector_t);
2ff28e22 369 void (*invalidatepage) (struct page *, unsigned long);
27496a8c 370 int (*releasepage) (struct page *, gfp_t);
1da177e4
LT
371 ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
372 loff_t offset, unsigned long nr_segs);
ceffc078
CO
373 struct page* (*get_xip_page)(struct address_space *, sector_t,
374 int);
e965f963
CL
375 /* migrate the contents of a page to the specified target */
376 int (*migratepage) (struct page *, struct page *);
1da177e4
LT
377};
378
379struct backing_dev_info;
380struct address_space {
381 struct inode *host; /* owner: inode, block_device */
382 struct radix_tree_root page_tree; /* radix tree of all pages */
383 rwlock_t tree_lock; /* and rwlock protecting it */
384 unsigned int i_mmap_writable;/* count VM_SHARED mappings */
385 struct prio_tree_root i_mmap; /* tree of private and shared mappings */
386 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
387 spinlock_t i_mmap_lock; /* protect tree, count, list */
388 unsigned int truncate_count; /* Cover race condition with truncate */
389 unsigned long nrpages; /* number of total pages */
390 pgoff_t writeback_index;/* writeback starts here */
391 struct address_space_operations *a_ops; /* methods */
392 unsigned long flags; /* error bits/gfp mask */
393 struct backing_dev_info *backing_dev_info; /* device readahead, etc */
394 spinlock_t private_lock; /* for use by the address_space */
395 struct list_head private_list; /* ditto */
396 struct address_space *assoc_mapping; /* ditto */
397} __attribute__((aligned(sizeof(long))));
398 /*
399 * On most architectures that alignment is already the case; but
400 * must be enforced here for CRIS, to let the least signficant bit
401 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
402 */
403
404struct block_device {
405 dev_t bd_dev; /* not a kdev_t - it's a search key */
406 struct inode * bd_inode; /* will die */
407 int bd_openers;
c039e313
AV
408 struct mutex bd_mutex; /* open/close mutex */
409 struct mutex bd_mount_mutex; /* mount mutex */
1da177e4
LT
410 struct list_head bd_inodes;
411 void * bd_holder;
412 int bd_holders;
413 struct block_device * bd_contains;
414 unsigned bd_block_size;
415 struct hd_struct * bd_part;
416 /* number of times partitions within this device have been opened. */
417 unsigned bd_part_count;
418 int bd_invalidated;
419 struct gendisk * bd_disk;
420 struct list_head bd_list;
421 struct backing_dev_info *bd_inode_backing_dev_info;
422 /*
423 * Private data. You must have bd_claim'ed the block_device
424 * to use this. NOTE: bd_claim allows an owner to claim
425 * the same device multiple times, the owner must take special
426 * care to not mess up bd_private for that case.
427 */
428 unsigned long bd_private;
429};
430
431/*
432 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
433 * radix trees
434 */
435#define PAGECACHE_TAG_DIRTY 0
436#define PAGECACHE_TAG_WRITEBACK 1
437
438int mapping_tagged(struct address_space *mapping, int tag);
439
440/*
441 * Might pages of this file be mapped into userspace?
442 */
443static inline int mapping_mapped(struct address_space *mapping)
444{
445 return !prio_tree_empty(&mapping->i_mmap) ||
446 !list_empty(&mapping->i_mmap_nonlinear);
447}
448
449/*
450 * Might pages of this file have been modified in userspace?
451 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
452 * marks vma as VM_SHARED if it is shared, and the file was opened for
453 * writing i.e. vma may be mprotected writable even if now readonly.
454 */
455static inline int mapping_writably_mapped(struct address_space *mapping)
456{
457 return mapping->i_mmap_writable != 0;
458}
459
460/*
461 * Use sequence counter to get consistent i_size on 32-bit processors.
462 */
463#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
464#include <linux/seqlock.h>
465#define __NEED_I_SIZE_ORDERED
466#define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
467#else
468#define i_size_ordered_init(inode) do { } while (0)
469#endif
470
471struct inode {
472 struct hlist_node i_hash;
473 struct list_head i_list;
474 struct list_head i_sb_list;
475 struct list_head i_dentry;
476 unsigned long i_ino;
477 atomic_t i_count;
478 umode_t i_mode;
479 unsigned int i_nlink;
480 uid_t i_uid;
481 gid_t i_gid;
482 dev_t i_rdev;
483 loff_t i_size;
484 struct timespec i_atime;
485 struct timespec i_mtime;
486 struct timespec i_ctime;
487 unsigned int i_blkbits;
488 unsigned long i_blksize;
489 unsigned long i_version;
a0f62ac6 490 blkcnt_t i_blocks;
1da177e4
LT
491 unsigned short i_bytes;
492 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
1b1dcc1b 493 struct mutex i_mutex;
1da177e4
LT
494 struct rw_semaphore i_alloc_sem;
495 struct inode_operations *i_op;
496 struct file_operations *i_fop; /* former ->i_op->default_file_ops */
497 struct super_block *i_sb;
498 struct file_lock *i_flock;
499 struct address_space *i_mapping;
500 struct address_space i_data;
501#ifdef CONFIG_QUOTA
502 struct dquot *i_dquot[MAXQUOTAS];
503#endif
504 /* These three should probably be a union */
505 struct list_head i_devices;
506 struct pipe_inode_info *i_pipe;
507 struct block_device *i_bdev;
508 struct cdev *i_cdev;
509 int i_cindex;
510
511 __u32 i_generation;
512
513#ifdef CONFIG_DNOTIFY
514 unsigned long i_dnotify_mask; /* Directory notify events */
515 struct dnotify_struct *i_dnotify; /* for directory notifications */
516#endif
517
0eeca283
RL
518#ifdef CONFIG_INOTIFY
519 struct list_head inotify_watches; /* watches on this inode */
d4f9af9d 520 struct mutex inotify_mutex; /* protects the watches list */
0eeca283
RL
521#endif
522
1da177e4
LT
523 unsigned long i_state;
524 unsigned long dirtied_when; /* jiffies of first dirtying */
525
526 unsigned int i_flags;
527
528 atomic_t i_writecount;
529 void *i_security;
530 union {
531 void *generic_ip;
532 } u;
533#ifdef __NEED_I_SIZE_ORDERED
534 seqcount_t i_size_seqcount;
535#endif
536};
537
538/*
539 * NOTE: in a 32bit arch with a preemptable kernel and
540 * an UP compile the i_size_read/write must be atomic
541 * with respect to the local cpu (unlike with preempt disabled),
542 * but they don't need to be atomic with respect to other cpus like in
543 * true SMP (so they need either to either locally disable irq around
544 * the read or for example on x86 they can be still implemented as a
545 * cmpxchg8b without the need of the lock prefix). For SMP compiles
546 * and 64bit archs it makes no difference if preempt is enabled or not.
547 */
548static inline loff_t i_size_read(struct inode *inode)
549{
550#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
551 loff_t i_size;
552 unsigned int seq;
553
554 do {
555 seq = read_seqcount_begin(&inode->i_size_seqcount);
556 i_size = inode->i_size;
557 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
558 return i_size;
559#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
560 loff_t i_size;
561
562 preempt_disable();
563 i_size = inode->i_size;
564 preempt_enable();
565 return i_size;
566#else
567 return inode->i_size;
568#endif
569}
570
571
572static inline void i_size_write(struct inode *inode, loff_t i_size)
573{
574#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
575 write_seqcount_begin(&inode->i_size_seqcount);
576 inode->i_size = i_size;
577 write_seqcount_end(&inode->i_size_seqcount);
578#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
579 preempt_disable();
580 inode->i_size = i_size;
581 preempt_enable();
582#else
583 inode->i_size = i_size;
584#endif
585}
586
587static inline unsigned iminor(struct inode *inode)
588{
589 return MINOR(inode->i_rdev);
590}
591
592static inline unsigned imajor(struct inode *inode)
593{
594 return MAJOR(inode->i_rdev);
595}
596
597extern struct block_device *I_BDEV(struct inode *inode);
598
599struct fown_struct {
600 rwlock_t lock; /* protects pid, uid, euid fields */
601 int pid; /* pid or -pgrp where SIGIO should be sent */
602 uid_t uid, euid; /* uid/euid of process setting the owner */
603 void *security;
604 int signum; /* posix.1b rt signal to be delivered on IO */
605};
606
607/*
608 * Track a single file's readahead state
609 */
610struct file_ra_state {
611 unsigned long start; /* Current window */
612 unsigned long size;
613 unsigned long flags; /* ra flags RA_FLAG_xxx*/
614 unsigned long cache_hit; /* cache hit count*/
615 unsigned long prev_page; /* Cache last read() position */
616 unsigned long ahead_start; /* Ahead window */
617 unsigned long ahead_size;
618 unsigned long ra_pages; /* Maximum readahead window */
619 unsigned long mmap_hit; /* Cache hit stat for mmap accesses */
620 unsigned long mmap_miss; /* Cache miss stat for mmap accesses */
621};
622#define RA_FLAG_MISS 0x01 /* a cache miss occured against this file */
623#define RA_FLAG_INCACHE 0x02 /* file is already in cache */
624
625struct file {
2f512016
ED
626 /*
627 * fu_list becomes invalid after file_free is called and queued via
628 * fu_rcuhead for RCU freeing
629 */
630 union {
631 struct list_head fu_list;
632 struct rcu_head fu_rcuhead;
633 } f_u;
1da177e4
LT
634 struct dentry *f_dentry;
635 struct vfsmount *f_vfsmnt;
636 struct file_operations *f_op;
637 atomic_t f_count;
638 unsigned int f_flags;
639 mode_t f_mode;
1da177e4
LT
640 loff_t f_pos;
641 struct fown_struct f_owner;
642 unsigned int f_uid, f_gid;
643 struct file_ra_state f_ra;
644
1da177e4
LT
645 unsigned long f_version;
646 void *f_security;
647
648 /* needed for tty driver, and maybe others */
649 void *private_data;
650
651#ifdef CONFIG_EPOLL
652 /* Used by fs/eventpoll.c to link all the hooks to this file */
653 struct list_head f_ep_links;
654 spinlock_t f_ep_lock;
655#endif /* #ifdef CONFIG_EPOLL */
656 struct address_space *f_mapping;
657};
658extern spinlock_t files_lock;
659#define file_list_lock() spin_lock(&files_lock);
660#define file_list_unlock() spin_unlock(&files_lock);
661
095975da 662#define get_file(x) atomic_inc(&(x)->f_count)
1da177e4
LT
663#define file_count(x) atomic_read(&(x)->f_count)
664
665#define MAX_NON_LFS ((1UL<<31) - 1)
666
667/* Page cache limit. The filesystems should put that into their s_maxbytes
668 limits, otherwise bad things can happen in VM. */
669#if BITS_PER_LONG==32
670#define MAX_LFS_FILESIZE (((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
671#elif BITS_PER_LONG==64
672#define MAX_LFS_FILESIZE 0x7fffffffffffffffUL
673#endif
674
675#define FL_POSIX 1
676#define FL_FLOCK 2
677#define FL_ACCESS 8 /* not trying to lock, just looking */
1da177e4
LT
678#define FL_LEASE 32 /* lease held on this file */
679#define FL_SLEEP 128 /* A blocking lock */
680
681/*
682 * The POSIX file lock owner is determined by
683 * the "struct files_struct" in the thread group
684 * (or NULL for no owner - BSD locks).
685 *
686 * Lockd stuffs a "host" pointer into this.
687 */
688typedef struct files_struct *fl_owner_t;
689
690struct file_lock_operations {
691 void (*fl_insert)(struct file_lock *); /* lock insertion callback */
692 void (*fl_remove)(struct file_lock *); /* lock removal callback */
693 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
694 void (*fl_release_private)(struct file_lock *);
695};
696
697struct lock_manager_operations {
698 int (*fl_compare_owner)(struct file_lock *, struct file_lock *);
699 void (*fl_notify)(struct file_lock *); /* unblock callback */
700 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
701 void (*fl_release_private)(struct file_lock *);
702 void (*fl_break)(struct file_lock *);
703 int (*fl_mylease)(struct file_lock *, struct file_lock *);
704 int (*fl_change)(struct file_lock **, int);
705};
706
707/* that will die - we need it for nfs_lock_info */
708#include <linux/nfs_fs_i.h>
709
710struct file_lock {
711 struct file_lock *fl_next; /* singly linked list for this inode */
712 struct list_head fl_link; /* doubly linked list of all locks */
713 struct list_head fl_block; /* circular list of blocked processes */
714 fl_owner_t fl_owner;
715 unsigned int fl_pid;
716 wait_queue_head_t fl_wait;
717 struct file *fl_file;
718 unsigned char fl_flags;
719 unsigned char fl_type;
720 loff_t fl_start;
721 loff_t fl_end;
722
723 struct fasync_struct * fl_fasync; /* for lease break notifications */
724 unsigned long fl_break_time; /* for nonblocking lease breaks */
725
726 struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
727 struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
728 union {
729 struct nfs_lock_info nfs_fl;
8d0a8a9d 730 struct nfs4_lock_info nfs4_fl;
1da177e4
LT
731 } fl_u;
732};
733
734/* The following constant reflects the upper bound of the file/locking space */
735#ifndef OFFSET_MAX
736#define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
737#define OFFSET_MAX INT_LIMIT(loff_t)
738#define OFFT_OFFSET_MAX INT_LIMIT(off_t)
739#endif
740
1da177e4
LT
741#include <linux/fcntl.h>
742
743extern int fcntl_getlk(struct file *, struct flock __user *);
c293621b
PS
744extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
745 struct flock __user *);
1da177e4
LT
746
747#if BITS_PER_LONG == 32
748extern int fcntl_getlk64(struct file *, struct flock64 __user *);
c293621b
PS
749extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
750 struct flock64 __user *);
1da177e4
LT
751#endif
752
753extern void send_sigio(struct fown_struct *fown, int fd, int band);
754extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
755extern int fcntl_getlease(struct file *filp);
756
757/* fs/locks.c */
758extern void locks_init_lock(struct file_lock *);
759extern void locks_copy_lock(struct file_lock *, struct file_lock *);
760extern void locks_remove_posix(struct file *, fl_owner_t);
761extern void locks_remove_flock(struct file *);
8dc7c311 762extern int posix_test_lock(struct file *, struct file_lock *, struct file_lock *);
5842add2 763extern int posix_lock_file_conf(struct file *, struct file_lock *, struct file_lock *);
1da177e4
LT
764extern int posix_lock_file(struct file *, struct file_lock *);
765extern int posix_lock_file_wait(struct file *, struct file_lock *);
64a318ee 766extern int posix_unblock_lock(struct file *, struct file_lock *);
1da177e4
LT
767extern int posix_locks_deadlock(struct file_lock *, struct file_lock *);
768extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
769extern int __break_lease(struct inode *inode, unsigned int flags);
770extern void lease_get_mtime(struct inode *, struct timespec *time);
771extern int setlease(struct file *, long, struct file_lock **);
772extern int lease_modify(struct file_lock **, int);
773extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
774extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
775extern void steal_locks(fl_owner_t from);
776
777struct fasync_struct {
778 int magic;
779 int fa_fd;
780 struct fasync_struct *fa_next; /* singly linked list */
781 struct file *fa_file;
782};
783
784#define FASYNC_MAGIC 0x4601
785
786/* SMP safe fasync helpers: */
787extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
788/* can be called from interrupts */
789extern void kill_fasync(struct fasync_struct **, int, int);
790/* only for net: no internal synchronization */
791extern void __kill_fasync(struct fasync_struct *, int, int);
792
793extern int f_setown(struct file *filp, unsigned long arg, int force);
794extern void f_delown(struct file *filp);
795extern int send_sigurg(struct fown_struct *fown);
796
797/*
798 * Umount options
799 */
800
801#define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
802#define MNT_DETACH 0x00000002 /* Just detach from the tree */
803#define MNT_EXPIRE 0x00000004 /* Mark for expiry */
804
805extern struct list_head super_blocks;
806extern spinlock_t sb_lock;
807
808#define sb_entry(list) list_entry((list), struct super_block, s_list)
809#define S_BIAS (1<<30)
810struct super_block {
811 struct list_head s_list; /* Keep this first */
812 dev_t s_dev; /* search index; _not_ kdev_t */
813 unsigned long s_blocksize;
1da177e4
LT
814 unsigned char s_blocksize_bits;
815 unsigned char s_dirt;
816 unsigned long long s_maxbytes; /* Max file size */
817 struct file_system_type *s_type;
818 struct super_operations *s_op;
819 struct dquot_operations *dq_op;
820 struct quotactl_ops *s_qcop;
821 struct export_operations *s_export_op;
822 unsigned long s_flags;
823 unsigned long s_magic;
824 struct dentry *s_root;
825 struct rw_semaphore s_umount;
7892f2f4 826 struct mutex s_lock;
1da177e4
LT
827 int s_count;
828 int s_syncing;
829 int s_need_sync_fs;
830 atomic_t s_active;
831 void *s_security;
832 struct xattr_handler **s_xattr;
833
834 struct list_head s_inodes; /* all inodes */
835 struct list_head s_dirty; /* dirty inodes */
836 struct list_head s_io; /* parked for writeback */
837 struct hlist_head s_anon; /* anonymous dentries for (nfs) exporting */
838 struct list_head s_files;
839
840 struct block_device *s_bdev;
841 struct list_head s_instances;
842 struct quota_info s_dquot; /* Diskquota specific options */
843
844 int s_frozen;
845 wait_queue_head_t s_wait_unfrozen;
846
847 char s_id[32]; /* Informational name */
848
849 void *s_fs_info; /* Filesystem private info */
850
851 /*
852 * The next field is for VFS *only*. No filesystems have any business
853 * even looking at it. You had been warned.
854 */
a11f3a05 855 struct mutex s_vfs_rename_mutex; /* Kludge */
1da177e4
LT
856
857 /* Granuality of c/m/atime in ns.
858 Cannot be worse than a second */
859 u32 s_time_gran;
860};
861
862extern struct timespec current_fs_time(struct super_block *sb);
863
864/*
865 * Snapshotting support.
866 */
867enum {
868 SB_UNFROZEN = 0,
869 SB_FREEZE_WRITE = 1,
870 SB_FREEZE_TRANS = 2,
871};
872
873#define vfs_check_frozen(sb, level) \
874 wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
875
22e2c507
JA
876static inline void get_fs_excl(void)
877{
878 atomic_inc(&current->fs_excl);
879}
880
881static inline void put_fs_excl(void)
882{
883 atomic_dec(&current->fs_excl);
884}
885
886static inline int has_fs_excl(void)
887{
888 return atomic_read(&current->fs_excl);
889}
890
891
1da177e4
LT
892/*
893 * Superblock locking.
894 */
895static inline void lock_super(struct super_block * sb)
896{
22e2c507 897 get_fs_excl();
7892f2f4 898 mutex_lock(&sb->s_lock);
1da177e4
LT
899}
900
901static inline void unlock_super(struct super_block * sb)
902{
22e2c507 903 put_fs_excl();
7892f2f4 904 mutex_unlock(&sb->s_lock);
1da177e4
LT
905}
906
907/*
908 * VFS helper functions..
909 */
e4543edd 910extern int vfs_permission(struct nameidata *, int);
1da177e4
LT
911extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *);
912extern int vfs_mkdir(struct inode *, struct dentry *, int);
913extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t);
914extern int vfs_symlink(struct inode *, struct dentry *, const char *, int);
915extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
916extern int vfs_rmdir(struct inode *, struct dentry *);
917extern int vfs_unlink(struct inode *, struct dentry *);
918extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
919
920/*
921 * VFS dentry helper functions.
922 */
923extern void dentry_unhash(struct dentry *dentry);
924
8c744fb8
CH
925/*
926 * VFS file helper functions.
927 */
928extern int file_permission(struct file *, int);
929
1da177e4
LT
930/*
931 * File types
932 *
933 * NOTE! These match bits 12..15 of stat.st_mode
934 * (ie "(i_mode >> 12) & 15").
935 */
936#define DT_UNKNOWN 0
937#define DT_FIFO 1
938#define DT_CHR 2
939#define DT_DIR 4
940#define DT_BLK 6
941#define DT_REG 8
942#define DT_LNK 10
943#define DT_SOCK 12
944#define DT_WHT 14
945
946#define OSYNC_METADATA (1<<0)
947#define OSYNC_DATA (1<<1)
948#define OSYNC_INODE (1<<2)
949int generic_osync_inode(struct inode *, struct address_space *, int);
950
951/*
952 * This is the "filldir" function type, used by readdir() to let
953 * the kernel specify what kind of dirent layout it wants to have.
954 * This allows the kernel to read directories into kernel space or
955 * to have different dirent layouts depending on the binary type.
956 */
957typedef int (*filldir_t)(void *, const char *, int, loff_t, ino_t, unsigned);
958
959struct block_device_operations {
960 int (*open) (struct inode *, struct file *);
961 int (*release) (struct inode *, struct file *);
962 int (*ioctl) (struct inode *, struct file *, unsigned, unsigned long);
bb93e3a5 963 long (*unlocked_ioctl) (struct file *, unsigned, unsigned long);
1da177e4 964 long (*compat_ioctl) (struct file *, unsigned, unsigned long);
420edbcc 965 int (*direct_access) (struct block_device *, sector_t, unsigned long *);
1da177e4
LT
966 int (*media_changed) (struct gendisk *);
967 int (*revalidate_disk) (struct gendisk *);
a885c8c4 968 int (*getgeo)(struct block_device *, struct hd_geometry *);
1da177e4
LT
969 struct module *owner;
970};
971
972/*
973 * "descriptor" for what we're up to with a read for sendfile().
974 * This allows us to use the same read code yet
975 * have multiple different users of the data that
976 * we read from a file.
977 *
978 * The simplest case just copies the data to user
979 * mode.
980 */
981typedef struct {
982 size_t written;
983 size_t count;
984 union {
985 char __user * buf;
986 void *data;
987 } arg;
988 int error;
989} read_descriptor_t;
990
991typedef int (*read_actor_t)(read_descriptor_t *, struct page *, unsigned long, unsigned long);
992
993/* These macros are for out of kernel modules to test that
994 * the kernel supports the unlocked_ioctl and compat_ioctl
995 * fields in struct file_operations. */
996#define HAVE_COMPAT_IOCTL 1
997#define HAVE_UNLOCKED_IOCTL 1
998
999/*
1000 * NOTE:
1001 * read, write, poll, fsync, readv, writev, unlocked_ioctl and compat_ioctl
1002 * can be called without the big kernel lock held in all filesystems.
1003 */
1004struct file_operations {
1005 struct module *owner;
1006 loff_t (*llseek) (struct file *, loff_t, int);
1007 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1008 ssize_t (*aio_read) (struct kiocb *, char __user *, size_t, loff_t);
1009 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1010 ssize_t (*aio_write) (struct kiocb *, const char __user *, size_t, loff_t);
1011 int (*readdir) (struct file *, void *, filldir_t);
1012 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1013 int (*ioctl) (struct inode *, struct file *, unsigned int, unsigned long);
1014 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1015 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1016 int (*mmap) (struct file *, struct vm_area_struct *);
1017 int (*open) (struct inode *, struct file *);
1018 int (*flush) (struct file *);
1019 int (*release) (struct inode *, struct file *);
1020 int (*fsync) (struct file *, struct dentry *, int datasync);
1021 int (*aio_fsync) (struct kiocb *, int datasync);
1022 int (*fasync) (int, struct file *, int);
1023 int (*lock) (struct file *, int, struct file_lock *);
1024 ssize_t (*readv) (struct file *, const struct iovec *, unsigned long, loff_t *);
1025 ssize_t (*writev) (struct file *, const struct iovec *, unsigned long, loff_t *);
1026 ssize_t (*sendfile) (struct file *, loff_t *, size_t, read_actor_t, void *);
1027 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1028 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1029 int (*check_flags)(int);
1030 int (*dir_notify)(struct file *filp, unsigned long arg);
1031 int (*flock) (struct file *, int, struct file_lock *);
1032};
1033
1034struct inode_operations {
1035 int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
1036 struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
1037 int (*link) (struct dentry *,struct inode *,struct dentry *);
1038 int (*unlink) (struct inode *,struct dentry *);
1039 int (*symlink) (struct inode *,struct dentry *,const char *);
1040 int (*mkdir) (struct inode *,struct dentry *,int);
1041 int (*rmdir) (struct inode *,struct dentry *);
1042 int (*mknod) (struct inode *,struct dentry *,int,dev_t);
1043 int (*rename) (struct inode *, struct dentry *,
1044 struct inode *, struct dentry *);
1045 int (*readlink) (struct dentry *, char __user *,int);
cc314eef
LT
1046 void * (*follow_link) (struct dentry *, struct nameidata *);
1047 void (*put_link) (struct dentry *, struct nameidata *, void *);
1da177e4
LT
1048 void (*truncate) (struct inode *);
1049 int (*permission) (struct inode *, int, struct nameidata *);
1050 int (*setattr) (struct dentry *, struct iattr *);
1051 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1052 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1053 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1054 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1055 int (*removexattr) (struct dentry *, const char *);
f6b3ec23 1056 void (*truncate_range)(struct inode *, loff_t, loff_t);
1da177e4
LT
1057};
1058
1059struct seq_file;
1060
1061extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1062extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1063extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1064 unsigned long, loff_t *);
1065extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1066 unsigned long, loff_t *);
1067
1068/*
1069 * NOTE: write_inode, delete_inode, clear_inode, put_inode can be called
1070 * without the big kernel lock held in all filesystems.
1071 */
1072struct super_operations {
1073 struct inode *(*alloc_inode)(struct super_block *sb);
1074 void (*destroy_inode)(struct inode *);
1075
1076 void (*read_inode) (struct inode *);
1077
1078 void (*dirty_inode) (struct inode *);
1079 int (*write_inode) (struct inode *, int);
1080 void (*put_inode) (struct inode *);
1081 void (*drop_inode) (struct inode *);
1082 void (*delete_inode) (struct inode *);
1083 void (*put_super) (struct super_block *);
1084 void (*write_super) (struct super_block *);
1085 int (*sync_fs)(struct super_block *sb, int wait);
1086 void (*write_super_lockfs) (struct super_block *);
1087 void (*unlockfs) (struct super_block *);
1088 int (*statfs) (struct super_block *, struct kstatfs *);
1089 int (*remount_fs) (struct super_block *, int *, char *);
1090 void (*clear_inode) (struct inode *);
1091 void (*umount_begin) (struct super_block *);
1092
1093 int (*show_options)(struct seq_file *, struct vfsmount *);
b4629fe2 1094 int (*show_stats)(struct seq_file *, struct vfsmount *);
1da177e4
LT
1095
1096 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1097 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1098};
1099
1100/* Inode state bits. Protected by inode_lock. */
1101#define I_DIRTY_SYNC 1 /* Not dirty enough for O_DATASYNC */
1102#define I_DIRTY_DATASYNC 2 /* Data-related inode changes pending */
1103#define I_DIRTY_PAGES 4 /* Data-related inode changes pending */
1104#define __I_LOCK 3
1105#define I_LOCK (1 << __I_LOCK)
1106#define I_FREEING 16
1107#define I_CLEAR 32
1108#define I_NEW 64
991114c6 1109#define I_WILL_FREE 128
1da177e4
LT
1110
1111#define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1112
1113extern void __mark_inode_dirty(struct inode *, int);
1114static inline void mark_inode_dirty(struct inode *inode)
1115{
1116 __mark_inode_dirty(inode, I_DIRTY);
1117}
1118
1119static inline void mark_inode_dirty_sync(struct inode *inode)
1120{
1121 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1122}
1123
66d21730
AD
1124static inline void inode_inc_link_count(struct inode *inode)
1125{
1126 inode->i_nlink++;
1127 mark_inode_dirty(inode);
1128}
1129
1130static inline void inode_dec_link_count(struct inode *inode)
1131{
1132 inode->i_nlink--;
1133 mark_inode_dirty(inode);
1134}
1135
869243a0 1136extern void touch_atime(struct vfsmount *mnt, struct dentry *dentry);
1da177e4
LT
1137static inline void file_accessed(struct file *file)
1138{
1139 if (!(file->f_flags & O_NOATIME))
1140 touch_atime(file->f_vfsmnt, file->f_dentry);
1141}
1142
1143int sync_inode(struct inode *inode, struct writeback_control *wbc);
1144
1145/**
4dc3b16b
PP
1146 * struct export_operations - for nfsd to communicate with file systems
1147 * @decode_fh: decode a file handle fragment and return a &struct dentry
1148 * @encode_fh: encode a file handle fragment from a dentry
1149 * @get_name: find the name for a given inode in a given directory
1150 * @get_parent: find the parent of a given directory
1151 * @get_dentry: find a dentry for the inode given a file handle sub-fragment
8f2709b5
RD
1152 * @find_exported_dentry:
1153 * set by the exporting module to a standard helper function.
1da177e4
LT
1154 *
1155 * Description:
1156 * The export_operations structure provides a means for nfsd to communicate
1157 * with a particular exported file system - particularly enabling nfsd and
1158 * the filesystem to co-operate when dealing with file handles.
1159 *
67be2dd1
MW
1160 * export_operations contains two basic operation for dealing with file
1161 * handles, decode_fh() and encode_fh(), and allows for some other
1162 * operations to be defined which standard helper routines use to get
1163 * specific information from the filesystem.
1da177e4
LT
1164 *
1165 * nfsd encodes information use to determine which filesystem a filehandle
67be2dd1
MW
1166 * applies to in the initial part of the file handle. The remainder, termed
1167 * a file handle fragment, is controlled completely by the filesystem. The
1168 * standard helper routines assume that this fragment will contain one or
1169 * two sub-fragments, one which identifies the file, and one which may be
1170 * used to identify the (a) directory containing the file.
1da177e4
LT
1171 *
1172 * In some situations, nfsd needs to get a dentry which is connected into a
67be2dd1
MW
1173 * specific part of the file tree. To allow for this, it passes the
1174 * function acceptable() together with a @context which can be used to see
1175 * if the dentry is acceptable. As there can be multiple dentrys for a
1176 * given file, the filesystem should check each one for acceptability before
1177 * looking for the next. As soon as an acceptable one is found, it should
1178 * be returned.
1da177e4
LT
1179 *
1180 * decode_fh:
67be2dd1
MW
1181 * @decode_fh is given a &struct super_block (@sb), a file handle fragment
1182 * (@fh, @fh_len) and an acceptability testing function (@acceptable,
1183 * @context). It should return a &struct dentry which refers to the same
1184 * file that the file handle fragment refers to, and which passes the
1185 * acceptability test. If it cannot, it should return a %NULL pointer if
1186 * the file was found but no acceptable &dentries were available, or a
1187 * %ERR_PTR error code indicating why it couldn't be found (e.g. %ENOENT or
1188 * %ENOMEM).
1da177e4
LT
1189 *
1190 * encode_fh:
67be2dd1
MW
1191 * @encode_fh should store in the file handle fragment @fh (using at most
1192 * @max_len bytes) information that can be used by @decode_fh to recover the
1193 * file refered to by the &struct dentry @de. If the @connectable flag is
1194 * set, the encode_fh() should store sufficient information so that a good
1195 * attempt can be made to find not only the file but also it's place in the
1196 * filesystem. This typically means storing a reference to de->d_parent in
1197 * the filehandle fragment. encode_fh() should return the number of bytes
1198 * stored or a negative error code such as %-ENOSPC
1da177e4
LT
1199 *
1200 * get_name:
67be2dd1
MW
1201 * @get_name should find a name for the given @child in the given @parent
1202 * directory. The name should be stored in the @name (with the
1203 * understanding that it is already pointing to a a %NAME_MAX+1 sized
1204 * buffer. get_name() should return %0 on success, a negative error code
1b1dcc1b 1205 * or error. @get_name will be called without @parent->i_mutex held.
1da177e4
LT
1206 *
1207 * get_parent:
67be2dd1
MW
1208 * @get_parent should find the parent directory for the given @child which
1209 * is also a directory. In the event that it cannot be found, or storage
1210 * space cannot be allocated, a %ERR_PTR should be returned.
1da177e4
LT
1211 *
1212 * get_dentry:
67be2dd1
MW
1213 * Given a &super_block (@sb) and a pointer to a file-system specific inode
1214 * identifier, possibly an inode number, (@inump) get_dentry() should find
1215 * the identified inode and return a dentry for that inode. Any suitable
1216 * dentry can be returned including, if necessary, a new dentry created with
1217 * d_alloc_root. The caller can then find any other extant dentrys by
1218 * following the d_alias links. If a new dentry was created using
1219 * d_alloc_root, DCACHE_NFSD_DISCONNECTED should be set, and the dentry
1220 * should be d_rehash()ed.
1da177e4 1221 *
67be2dd1
MW
1222 * If the inode cannot be found, either a %NULL pointer or an %ERR_PTR code
1223 * can be returned. The @inump will be whatever was passed to
1224 * nfsd_find_fh_dentry() in either the @obj or @parent parameters.
1da177e4
LT
1225 *
1226 * Locking rules:
1b1dcc1b 1227 * get_parent is called with child->d_inode->i_mutex down
67be2dd1 1228 * get_name is not (which is possibly inconsistent)
1da177e4
LT
1229 */
1230
1231struct export_operations {
1232 struct dentry *(*decode_fh)(struct super_block *sb, __u32 *fh, int fh_len, int fh_type,
1233 int (*acceptable)(void *context, struct dentry *de),
1234 void *context);
1235 int (*encode_fh)(struct dentry *de, __u32 *fh, int *max_len,
1236 int connectable);
1237
1238 /* the following are only called from the filesystem itself */
1239 int (*get_name)(struct dentry *parent, char *name,
1240 struct dentry *child);
1241 struct dentry * (*get_parent)(struct dentry *child);
1242 struct dentry * (*get_dentry)(struct super_block *sb, void *inump);
1243
1244 /* This is set by the exporting module to a standard helper */
1245 struct dentry * (*find_exported_dentry)(
1246 struct super_block *sb, void *obj, void *parent,
1247 int (*acceptable)(void *context, struct dentry *de),
1248 void *context);
1249
1250
1251};
1252
1253extern struct dentry *
1254find_exported_dentry(struct super_block *sb, void *obj, void *parent,
1255 int (*acceptable)(void *context, struct dentry *de),
1256 void *context);
1257
1258struct file_system_type {
1259 const char *name;
1260 int fs_flags;
1261 struct super_block *(*get_sb) (struct file_system_type *, int,
1262 const char *, void *);
1263 void (*kill_sb) (struct super_block *);
1264 struct module *owner;
1265 struct file_system_type * next;
1266 struct list_head fs_supers;
1267};
1268
1269struct super_block *get_sb_bdev(struct file_system_type *fs_type,
1270 int flags, const char *dev_name, void *data,
1271 int (*fill_super)(struct super_block *, void *, int));
1272struct super_block *get_sb_single(struct file_system_type *fs_type,
1273 int flags, void *data,
1274 int (*fill_super)(struct super_block *, void *, int));
1275struct super_block *get_sb_nodev(struct file_system_type *fs_type,
1276 int flags, void *data,
1277 int (*fill_super)(struct super_block *, void *, int));
1278void generic_shutdown_super(struct super_block *sb);
1279void kill_block_super(struct super_block *sb);
1280void kill_anon_super(struct super_block *sb);
1281void kill_litter_super(struct super_block *sb);
1282void deactivate_super(struct super_block *sb);
1283int set_anon_super(struct super_block *s, void *data);
1284struct super_block *sget(struct file_system_type *type,
1285 int (*test)(struct super_block *,void *),
1286 int (*set)(struct super_block *,void *),
1287 void *data);
1288struct super_block *get_sb_pseudo(struct file_system_type *, char *,
1289 struct super_operations *ops, unsigned long);
1290int __put_super(struct super_block *sb);
1291int __put_super_and_need_restart(struct super_block *sb);
1292void unnamed_dev_init(void);
1293
1294/* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1295#define fops_get(fops) \
1296 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1297#define fops_put(fops) \
1298 do { if (fops) module_put((fops)->owner); } while(0)
1299
1300extern int register_filesystem(struct file_system_type *);
1301extern int unregister_filesystem(struct file_system_type *);
1302extern struct vfsmount *kern_mount(struct file_system_type *);
1303extern int may_umount_tree(struct vfsmount *);
1304extern int may_umount(struct vfsmount *);
a05964f3 1305extern void umount_tree(struct vfsmount *, int, struct list_head *);
b90fa9ae 1306extern void release_mounts(struct list_head *);
1da177e4 1307extern long do_mount(char *, char *, char *, unsigned long, void *);
b90fa9ae
RP
1308extern struct vfsmount *copy_tree(struct vfsmount *, struct dentry *, int);
1309extern void mnt_set_mountpoint(struct vfsmount *, struct dentry *,
1310 struct vfsmount *);
1da177e4
LT
1311
1312extern int vfs_statfs(struct super_block *, struct kstatfs *);
1313
f87fd4c2
MS
1314/* /sys/fs */
1315extern struct subsystem fs_subsys;
1316
1da177e4
LT
1317#define FLOCK_VERIFY_READ 1
1318#define FLOCK_VERIFY_WRITE 2
1319
1320extern int locks_mandatory_locked(struct inode *);
1321extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1322
1323/*
1324 * Candidates for mandatory locking have the setgid bit set
1325 * but no group execute bit - an otherwise meaningless combination.
1326 */
1327#define MANDATORY_LOCK(inode) \
1328 (IS_MANDLOCK(inode) && ((inode)->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID)
1329
1330static inline int locks_verify_locked(struct inode *inode)
1331{
1332 if (MANDATORY_LOCK(inode))
1333 return locks_mandatory_locked(inode);
1334 return 0;
1335}
1336
1337extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1338
1339static inline int locks_verify_truncate(struct inode *inode,
1340 struct file *filp,
1341 loff_t size)
1342{
1343 if (inode->i_flock && MANDATORY_LOCK(inode))
1344 return locks_mandatory_area(
1345 FLOCK_VERIFY_WRITE, inode, filp,
1346 size < inode->i_size ? size : inode->i_size,
1347 (size < inode->i_size ? inode->i_size - size
1348 : size - inode->i_size)
1349 );
1350 return 0;
1351}
1352
1353static inline int break_lease(struct inode *inode, unsigned int mode)
1354{
1355 if (inode->i_flock)
1356 return __break_lease(inode, mode);
1357 return 0;
1358}
1359
1360/* fs/open.c */
1361
4a30131e
N
1362extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
1363 struct file *filp);
5590ff0d
UD
1364extern long do_sys_open(int fdf, const char __user *filename, int flags,
1365 int mode);
1da177e4
LT
1366extern struct file *filp_open(const char *, int, int);
1367extern struct file * dentry_open(struct dentry *, struct vfsmount *, int);
1368extern int filp_close(struct file *, fl_owner_t id);
1369extern char * getname(const char __user *);
1370
1371/* fs/dcache.c */
1372extern void __init vfs_caches_init_early(void);
1373extern void __init vfs_caches_init(unsigned long);
1374
1375#define __getname() kmem_cache_alloc(names_cachep, SLAB_KERNEL)
1376#define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
1377#ifndef CONFIG_AUDITSYSCALL
1378#define putname(name) __putname(name)
1379#else
1380extern void putname(const char *name);
1381#endif
1382
1383extern int register_blkdev(unsigned int, const char *);
1384extern int unregister_blkdev(unsigned int, const char *);
1385extern struct block_device *bdget(dev_t);
1386extern void bd_set_size(struct block_device *, loff_t size);
1387extern void bd_forget(struct inode *inode);
1388extern void bdput(struct block_device *);
1389extern struct block_device *open_by_devnum(dev_t, unsigned);
1390extern struct file_operations def_blk_fops;
1391extern struct address_space_operations def_blk_aops;
1392extern struct file_operations def_chr_fops;
1393extern struct file_operations bad_sock_fops;
1394extern struct file_operations def_fifo_fops;
1395extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
1396extern int blkdev_ioctl(struct inode *, struct file *, unsigned, unsigned long);
1397extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
1398extern int blkdev_get(struct block_device *, mode_t, unsigned);
1399extern int blkdev_put(struct block_device *);
1400extern int bd_claim(struct block_device *, void *);
1401extern void bd_release(struct block_device *);
1402
1403/* fs/char_dev.c */
1404extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
1405extern int register_chrdev_region(dev_t, unsigned, const char *);
1406extern int register_chrdev(unsigned int, const char *,
1407 struct file_operations *);
1408extern int unregister_chrdev(unsigned int, const char *);
1409extern void unregister_chrdev_region(dev_t, unsigned);
1410extern int chrdev_open(struct inode *, struct file *);
7170be5f
NH
1411extern int get_chrdev_list(char *);
1412extern void *acquire_chrdev_list(void);
1413extern int count_chrdev_list(void);
1414extern void *get_next_chrdev(void *);
1415extern int get_chrdev_info(void *, int *, char **);
1416extern void release_chrdev_list(void *);
1da177e4
LT
1417
1418/* fs/block_dev.c */
1419#define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
1420extern const char *__bdevname(dev_t, char *buffer);
1421extern const char *bdevname(struct block_device *bdev, char *buffer);
1422extern struct block_device *lookup_bdev(const char *);
1423extern struct block_device *open_bdev_excl(const char *, int, void *);
1424extern void close_bdev_excl(struct block_device *);
7170be5f
NH
1425extern void *acquire_blkdev_list(void);
1426extern int count_blkdev_list(void);
1427extern void *get_next_blkdev(void *);
1428extern int get_blkdev_info(void *, int *, char **);
1429extern void release_blkdev_list(void *);
1da177e4
LT
1430
1431extern void init_special_inode(struct inode *, umode_t, dev_t);
1432
1433/* Invalid inode operations -- fs/bad_inode.c */
1434extern void make_bad_inode(struct inode *);
1435extern int is_bad_inode(struct inode *);
1436
1437extern struct file_operations read_fifo_fops;
1438extern struct file_operations write_fifo_fops;
1439extern struct file_operations rdwr_fifo_fops;
1da177e4
LT
1440
1441extern int fs_may_remount_ro(struct super_block *);
1442
1443/*
1444 * return READ, READA, or WRITE
1445 */
1446#define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
1447
1448/*
1449 * return data direction, READ or WRITE
1450 */
1451#define bio_data_dir(bio) ((bio)->bi_rw & 1)
1452
1453extern int check_disk_change(struct block_device *);
1454extern int invalidate_inodes(struct super_block *);
2ef41634 1455extern int __invalidate_device(struct block_device *);
1da177e4
LT
1456extern int invalidate_partition(struct gendisk *, int);
1457unsigned long invalidate_mapping_pages(struct address_space *mapping,
1458 pgoff_t start, pgoff_t end);
1459unsigned long invalidate_inode_pages(struct address_space *mapping);
1460static inline void invalidate_remote_inode(struct inode *inode)
1461{
1462 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
1463 S_ISLNK(inode->i_mode))
1464 invalidate_inode_pages(inode->i_mapping);
1465}
1466extern int invalidate_inode_pages2(struct address_space *mapping);
1467extern int invalidate_inode_pages2_range(struct address_space *mapping,
1468 pgoff_t start, pgoff_t end);
1469extern int write_inode_now(struct inode *, int);
1470extern int filemap_fdatawrite(struct address_space *);
1471extern int filemap_flush(struct address_space *);
1472extern int filemap_fdatawait(struct address_space *);
1473extern int filemap_write_and_wait(struct address_space *mapping);
1474extern int filemap_write_and_wait_range(struct address_space *mapping,
1475 loff_t lstart, loff_t lend);
ebcf28e1
AM
1476extern int wait_on_page_writeback_range(struct address_space *mapping,
1477 pgoff_t start, pgoff_t end);
1478extern int __filemap_fdatawrite_range(struct address_space *mapping,
1479 loff_t start, loff_t end, int sync_mode);
1480
18e79b40 1481extern long do_fsync(struct file *file, int datasync);
1da177e4
LT
1482extern void sync_supers(void);
1483extern void sync_filesystems(int wait);
1484extern void emergency_sync(void);
1485extern void emergency_remount(void);
1486extern int do_remount_sb(struct super_block *sb, int flags,
1487 void *data, int force);
1488extern sector_t bmap(struct inode *, sector_t);
1da177e4
LT
1489extern int notify_change(struct dentry *, struct iattr *);
1490extern int permission(struct inode *, int, struct nameidata *);
1491extern int generic_permission(struct inode *, int,
1492 int (*check_acl)(struct inode *, int));
1493
1494extern int get_write_access(struct inode *);
1495extern int deny_write_access(struct file *);
1496static inline void put_write_access(struct inode * inode)
1497{
1498 atomic_dec(&inode->i_writecount);
1499}
1500static inline void allow_write_access(struct file *file)
1501{
1502 if (file)
1503 atomic_inc(&file->f_dentry->d_inode->i_writecount);
1504}
1505extern int do_pipe(int *);
1506
5590ff0d 1507extern int open_namei(int dfd, const char *, int, int, struct nameidata *);
1da177e4
LT
1508extern int may_open(struct nameidata *, int, int);
1509
1510extern int kernel_read(struct file *, unsigned long, char *, unsigned long);
1511extern struct file * open_exec(const char *);
1512
1513/* fs/dcache.c -- generic fs support functions */
1514extern int is_subdir(struct dentry *, struct dentry *);
1515extern ino_t find_inode_number(struct dentry *, struct qstr *);
1516
1517#include <linux/err.h>
1518
1519/* needed for stackable file system support */
1520extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
1521
1522extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
1523
1524extern void inode_init_once(struct inode *);
1525extern void iput(struct inode *);
1526extern struct inode * igrab(struct inode *);
1527extern ino_t iunique(struct super_block *, ino_t);
1528extern int inode_needs_sync(struct inode *inode);
1529extern void generic_delete_inode(struct inode *inode);
cb2c0233 1530extern void generic_drop_inode(struct inode *inode);
1da177e4 1531
88bd5121
AA
1532extern struct inode *ilookup5_nowait(struct super_block *sb,
1533 unsigned long hashval, int (*test)(struct inode *, void *),
1534 void *data);
1da177e4
LT
1535extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
1536 int (*test)(struct inode *, void *), void *data);
1537extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
1538
1539extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
1540extern struct inode * iget_locked(struct super_block *, unsigned long);
1541extern void unlock_new_inode(struct inode *);
1542
1543static inline struct inode *iget(struct super_block *sb, unsigned long ino)
1544{
1545 struct inode *inode = iget_locked(sb, ino);
1546
1547 if (inode && (inode->i_state & I_NEW)) {
1548 sb->s_op->read_inode(inode);
1549 unlock_new_inode(inode);
1550 }
1551
1552 return inode;
1553}
1554
1555extern void __iget(struct inode * inode);
1556extern void clear_inode(struct inode *);
1557extern void destroy_inode(struct inode *);
1558extern struct inode *new_inode(struct super_block *);
1559extern int remove_suid(struct dentry *);
1560extern void remove_dquot_ref(struct super_block *, int, struct list_head *);
1da177e4
LT
1561
1562extern void __insert_inode_hash(struct inode *, unsigned long hashval);
1563extern void remove_inode_hash(struct inode *);
1564static inline void insert_inode_hash(struct inode *inode) {
1565 __insert_inode_hash(inode, inode->i_ino);
1566}
1567
1568extern struct file * get_empty_filp(void);
1569extern void file_move(struct file *f, struct list_head *list);
1570extern void file_kill(struct file *f);
1571struct bio;
1572extern void submit_bio(int, struct bio *);
1573extern int bdev_read_only(struct block_device *);
1574extern int set_blocksize(struct block_device *, int);
1575extern int sb_set_blocksize(struct super_block *, int);
1576extern int sb_min_blocksize(struct super_block *, int);
1577
1578extern int generic_file_mmap(struct file *, struct vm_area_struct *);
1579extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
1580extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
1581extern int file_send_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
1582extern ssize_t generic_file_read(struct file *, char __user *, size_t, loff_t *);
1583int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
1584extern ssize_t generic_file_write(struct file *, const char __user *, size_t, loff_t *);
1585extern ssize_t generic_file_aio_read(struct kiocb *, char __user *, size_t, loff_t);
1586extern ssize_t __generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t *);
1587extern ssize_t generic_file_aio_write(struct kiocb *, const char __user *, size_t, loff_t);
1588extern ssize_t generic_file_aio_write_nolock(struct kiocb *, const struct iovec *,
1589 unsigned long, loff_t *);
1590extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
1591 unsigned long *, loff_t, loff_t *, size_t, size_t);
1592extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
1593 unsigned long, loff_t, loff_t *, size_t, ssize_t);
1594extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
1595extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
1596ssize_t generic_file_write_nolock(struct file *file, const struct iovec *iov,
1597 unsigned long nr_segs, loff_t *ppos);
1598extern ssize_t generic_file_sendfile(struct file *, loff_t *, size_t, read_actor_t, void *);
1599extern void do_generic_mapping_read(struct address_space *mapping,
1600 struct file_ra_state *, struct file *,
1601 loff_t *, read_descriptor_t *, read_actor_t);
1602extern void
1603file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
1da177e4
LT
1604extern ssize_t generic_file_readv(struct file *filp, const struct iovec *iov,
1605 unsigned long nr_segs, loff_t *ppos);
1606ssize_t generic_file_writev(struct file *filp, const struct iovec *iov,
1607 unsigned long nr_segs, loff_t *ppos);
1608extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
1609extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
1610extern loff_t remote_llseek(struct file *file, loff_t offset, int origin);
1611extern int generic_file_open(struct inode * inode, struct file * filp);
1612extern int nonseekable_open(struct inode * inode, struct file * filp);
1613
ceffc078 1614#ifdef CONFIG_FS_XIP
eb6fe0c3
CO
1615extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
1616 loff_t *ppos);
ceffc078
CO
1617extern ssize_t xip_file_sendfile(struct file *in_file, loff_t *ppos,
1618 size_t count, read_actor_t actor,
1619 void *target);
1620extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
eb6fe0c3
CO
1621extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
1622 size_t len, loff_t *ppos);
ceffc078 1623extern int xip_truncate_page(struct address_space *mapping, loff_t from);
6d79125b
CO
1624#else
1625static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
1626{
1627 return 0;
1628}
ceffc078
CO
1629#endif
1630
1da177e4
LT
1631static inline void do_generic_file_read(struct file * filp, loff_t *ppos,
1632 read_descriptor_t * desc,
1633 read_actor_t actor)
1634{
1635 do_generic_mapping_read(filp->f_mapping,
1636 &filp->f_ra,
1637 filp,
1638 ppos,
1639 desc,
1640 actor);
1641}
1642
1643ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
1644 struct block_device *bdev, const struct iovec *iov, loff_t offset,
1d8fa7a2 1645 unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
1da177e4
LT
1646 int lock_type);
1647
1648enum {
1649 DIO_LOCKING = 1, /* need locking between buffered and direct access */
1650 DIO_NO_LOCKING, /* bdev; no locking at all between buffered/direct */
1651 DIO_OWN_LOCKING, /* filesystem locks buffered and direct internally */
1652};
1653
1654static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
1655 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1d8fa7a2 1656 loff_t offset, unsigned long nr_segs, get_block_t get_block,
1da177e4
LT
1657 dio_iodone_t end_io)
1658{
1659 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1d8fa7a2 1660 nr_segs, get_block, end_io, DIO_LOCKING);
1da177e4
LT
1661}
1662
1663static inline ssize_t blockdev_direct_IO_no_locking(int rw, struct kiocb *iocb,
1664 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1d8fa7a2 1665 loff_t offset, unsigned long nr_segs, get_block_t get_block,
1da177e4
LT
1666 dio_iodone_t end_io)
1667{
1668 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1d8fa7a2 1669 nr_segs, get_block, end_io, DIO_NO_LOCKING);
1da177e4
LT
1670}
1671
1672static inline ssize_t blockdev_direct_IO_own_locking(int rw, struct kiocb *iocb,
1673 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1d8fa7a2 1674 loff_t offset, unsigned long nr_segs, get_block_t get_block,
1da177e4
LT
1675 dio_iodone_t end_io)
1676{
1677 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1d8fa7a2 1678 nr_segs, get_block, end_io, DIO_OWN_LOCKING);
1da177e4
LT
1679}
1680
1681extern struct file_operations generic_ro_fops;
1682
1683#define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
1684
1685extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
1686extern int vfs_follow_link(struct nameidata *, const char *);
1687extern int page_readlink(struct dentry *, char __user *, int);
cc314eef
LT
1688extern void *page_follow_link_light(struct dentry *, struct nameidata *);
1689extern void page_put_link(struct dentry *, struct nameidata *, void *);
0adb25d2
KK
1690extern int __page_symlink(struct inode *inode, const char *symname, int len,
1691 gfp_t gfp_mask);
1da177e4
LT
1692extern int page_symlink(struct inode *inode, const char *symname, int len);
1693extern struct inode_operations page_symlink_inode_operations;
1694extern int generic_readlink(struct dentry *, char __user *, int);
1695extern void generic_fillattr(struct inode *, struct kstat *);
1696extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1697void inode_add_bytes(struct inode *inode, loff_t bytes);
1698void inode_sub_bytes(struct inode *inode, loff_t bytes);
1699loff_t inode_get_bytes(struct inode *inode);
1700void inode_set_bytes(struct inode *inode, loff_t bytes);
1701
1702extern int vfs_readdir(struct file *, filldir_t, void *);
1703
1704extern int vfs_stat(char __user *, struct kstat *);
1705extern int vfs_lstat(char __user *, struct kstat *);
5590ff0d
UD
1706extern int vfs_stat_fd(int dfd, char __user *, struct kstat *);
1707extern int vfs_lstat_fd(int dfd, char __user *, struct kstat *);
1da177e4
LT
1708extern int vfs_fstat(unsigned int, struct kstat *);
1709
1710extern int vfs_ioctl(struct file *, unsigned int, unsigned int, unsigned long);
1711
1712extern struct file_system_type *get_fs_type(const char *name);
1713extern struct super_block *get_super(struct block_device *);
1714extern struct super_block *user_get_super(dev_t);
1715extern void drop_super(struct super_block *sb);
1716
1717extern int dcache_dir_open(struct inode *, struct file *);
1718extern int dcache_dir_close(struct inode *, struct file *);
1719extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
1720extern int dcache_readdir(struct file *, void *, filldir_t);
1721extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1722extern int simple_statfs(struct super_block *, struct kstatfs *);
1723extern int simple_link(struct dentry *, struct inode *, struct dentry *);
1724extern int simple_unlink(struct inode *, struct dentry *);
1725extern int simple_rmdir(struct inode *, struct dentry *);
1726extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1727extern int simple_sync_file(struct file *, struct dentry *, int);
1728extern int simple_empty(struct dentry *);
1729extern int simple_readpage(struct file *file, struct page *page);
1730extern int simple_prepare_write(struct file *file, struct page *page,
1731 unsigned offset, unsigned to);
1732extern int simple_commit_write(struct file *file, struct page *page,
1733 unsigned offset, unsigned to);
1734
1735extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
1736extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
1737extern struct file_operations simple_dir_operations;
1738extern struct inode_operations simple_dir_inode_operations;
1739struct tree_descr { char *name; struct file_operations *ops; int mode; };
1740struct dentry *d_alloc_name(struct dentry *, const char *);
1741extern int simple_fill_super(struct super_block *, int, struct tree_descr *);
1742extern int simple_pin_fs(char *name, struct vfsmount **mount, int *count);
1743extern void simple_release_fs(struct vfsmount **mount, int *count);
1744
1745extern ssize_t simple_read_from_buffer(void __user *, size_t, loff_t *, const void *, size_t);
1746
e965f963
CL
1747#ifdef CONFIG_MIGRATION
1748extern int buffer_migrate_page(struct page *, struct page *);
1749#else
1750#define buffer_migrate_page NULL
1751#endif
1752
1da177e4
LT
1753extern int inode_change_ok(struct inode *, struct iattr *);
1754extern int __must_check inode_setattr(struct inode *, struct iattr *);
1755
870f4817 1756extern void file_update_time(struct file *file);
1da177e4
LT
1757
1758static inline ino_t parent_ino(struct dentry *dentry)
1759{
1760 ino_t res;
1761
1762 spin_lock(&dentry->d_lock);
1763 res = dentry->d_parent->d_inode->i_ino;
1764 spin_unlock(&dentry->d_lock);
1765 return res;
1766}
1767
1768/* kernel/fork.c */
1769extern int unshare_files(void);
1770
1771/* Transaction based IO helpers */
1772
1773/*
1774 * An argresp is stored in an allocated page and holds the
1775 * size of the argument or response, along with its content
1776 */
1777struct simple_transaction_argresp {
1778 ssize_t size;
1779 char data[0];
1780};
1781
1782#define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
1783
1784char *simple_transaction_get(struct file *file, const char __user *buf,
1785 size_t size);
1786ssize_t simple_transaction_read(struct file *file, char __user *buf,
1787 size_t size, loff_t *pos);
1788int simple_transaction_release(struct inode *inode, struct file *file);
1789
1790static inline void simple_transaction_set(struct file *file, size_t n)
1791{
1792 struct simple_transaction_argresp *ar = file->private_data;
1793
1794 BUG_ON(n > SIMPLE_TRANSACTION_LIMIT);
1795
1796 /*
1797 * The barrier ensures that ar->size will really remain zero until
1798 * ar->data is ready for reading.
1799 */
1800 smp_mb();
1801 ar->size = n;
1802}
1803
acaefc25
AB
1804/*
1805 * simple attribute files
1806 *
1807 * These attributes behave similar to those in sysfs:
1808 *
1809 * Writing to an attribute immediately sets a value, an open file can be
1810 * written to multiple times.
1811 *
1812 * Reading from an attribute creates a buffer from the value that might get
1813 * read with multiple read calls. When the attribute has been read
1814 * completely, no further read calls are possible until the file is opened
1815 * again.
1816 *
1817 * All attributes contain a text representation of a numeric value
1818 * that are accessed with the get() and set() functions.
1819 */
1820#define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
1821static int __fops ## _open(struct inode *inode, struct file *file) \
1822{ \
1823 __simple_attr_check_format(__fmt, 0ull); \
1824 return simple_attr_open(inode, file, __get, __set, __fmt); \
1825} \
1826static struct file_operations __fops = { \
1827 .owner = THIS_MODULE, \
1828 .open = __fops ## _open, \
1829 .release = simple_attr_close, \
1830 .read = simple_attr_read, \
1831 .write = simple_attr_write, \
1832};
1833
1834static inline void __attribute__((format(printf, 1, 2)))
1835__simple_attr_check_format(const char *fmt, ...)
1836{
1837 /* don't do anything, just let the compiler check the arguments; */
1838}
1839
1840int simple_attr_open(struct inode *inode, struct file *file,
1841 u64 (*get)(void *), void (*set)(void *, u64),
1842 const char *fmt);
1843int simple_attr_close(struct inode *inode, struct file *file);
1844ssize_t simple_attr_read(struct file *file, char __user *buf,
1845 size_t len, loff_t *ppos);
1846ssize_t simple_attr_write(struct file *file, const char __user *buf,
1847 size_t len, loff_t *ppos);
1848
1849
1da177e4
LT
1850#ifdef CONFIG_SECURITY
1851static inline char *alloc_secdata(void)
1852{
1853 return (char *)get_zeroed_page(GFP_KERNEL);
1854}
1855
1856static inline void free_secdata(void *secdata)
1857{
1858 free_page((unsigned long)secdata);
1859}
1860#else
1861static inline char *alloc_secdata(void)
1862{
1863 return (char *)1;
1864}
1865
1866static inline void free_secdata(void *secdata)
1867{ }
1868#endif /* CONFIG_SECURITY */
1869
1870#endif /* __KERNEL__ */
1871#endif /* _LINUX_FS_H */