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