5 * This file has definitions for some important file table
9 #include <linux/config.h>
10 #include <linux/linkage.h>
11 #include <linux/limits.h>
12 #include <linux/wait.h>
13 #include <linux/types.h>
14 #include <linux/vfs.h>
15 #include <linux/net.h>
16 #include <linux/kdev_t.h>
17 #include <linux/ioctl.h>
18 #include <linux/list.h>
19 #include <linux/dcache.h>
20 #include <linux/stat.h>
21 #include <linux/cache.h>
22 #include <linux/stddef.h>
23 #include <linux/string.h>
25 #include <asm/atomic.h>
26 #include <asm/bitops.h>
28 struct poll_table_struct
;
32 * It's silly to have NR_OPEN bigger than NR_FILE, but you can change
33 * the file limit at runtime and only root can increase the per-process
34 * nr_file rlimit, so it's safe to set up a ridiculously high absolute
35 * upper limit on files-per-process.
37 * Some programs (notably those using select()) may have to be
38 * recompiled to take full advantage of the new limits..
41 /* Fixed constants first: */
43 #define NR_OPEN (1024*1024) /* Absolute upper limit on fd num */
44 #define INR_OPEN 1024 /* Initial setting for nfile rlimits */
46 #define BLOCK_SIZE_BITS 10
47 #define BLOCK_SIZE (1<<BLOCK_SIZE_BITS)
49 /* And dynamically-tunable limits and defaults: */
50 extern int max_files
, nr_files
, nr_free_files
;
51 extern int max_super_blocks
, nr_super_blocks
;
53 #define NR_FILE 8192 /* this can well be larger on a larger system */
54 #define NR_RESERVED_FILES 10 /* reserved for root */
66 #define READA 2 /* read-ahead - don't block if no resources */
67 #define SPECIAL 4 /* For non-blockdevice requests in request queue */
69 #define WRITERAW 5 /* raw write - don't play with buffer lists */
71 #define NIL_FILP ((struct file *)0)
76 /* public flags for file_system_type */
77 #define FS_REQUIRES_DEV 1
78 #define FS_NO_DCACHE 2 /* Only dcache the necessary things. */
79 #define FS_NO_PRELIM 4 /* prevent preloading of dentries, even if
80 * FS_NO_DCACHE is not set.
84 * These are the fs-independent mount-flags: up to 16 flags are supported
86 #define MS_RDONLY 1 /* Mount read-only */
87 #define MS_NOSUID 2 /* Ignore suid and sgid bits */
88 #define MS_NODEV 4 /* Disallow access to device special files */
89 #define MS_NOEXEC 8 /* Disallow program execution */
90 #define MS_SYNCHRONOUS 16 /* Writes are synced at once */
91 #define MS_REMOUNT 32 /* Alter flags of a mounted FS */
92 #define MS_MANDLOCK 64 /* Allow mandatory locks on an FS */
93 #define S_QUOTA 128 /* Quota initialized for file/directory/symlink */
94 #define S_APPEND 256 /* Append-only file */
95 #define S_IMMUTABLE 512 /* Immutable file */
96 #define MS_NOATIME 1024 /* Do not update access times. */
97 #define MS_NODIRATIME 2048 /* Do not update directory access times */
99 #define MS_ODD_RENAME 32768 /* Temporary stuff; will go away as soon
100 * as nfs_rename() will be cleaned up
104 * Flags that can be altered by MS_REMOUNT
106 #define MS_RMT_MASK (MS_RDONLY|MS_NOSUID|MS_NODEV|MS_NOEXEC|\
107 MS_SYNCHRONOUS|MS_MANDLOCK|MS_NOATIME|MS_NODIRATIME)
110 * Magic mount flag number. Has to be or-ed to the flag values.
112 #define MS_MGC_VAL 0xC0ED0000 /* magic flag number to indicate "new" flags */
113 #define MS_MGC_MSK 0xffff0000 /* magic flag number mask */
116 * Note that nosuid etc flags are inode-specific: setting some file-system
117 * flags just means all the inodes inherit those flags by default. It might be
118 * possible to override it selectively if you really wanted to with some
119 * ioctl() that is not currently implemented.
121 * Exception: MS_RDONLY is always applied to the entire file system.
123 * Unfortunately, it is possible to change a filesystems flags with it mounted
124 * with files in use. This means that all of the inodes will not have their
125 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
126 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
128 #define __IS_FLG(inode,flg) (((inode)->i_sb && (inode)->i_sb->s_flags & (flg)) \
129 || (inode)->i_flags & (flg))
131 #define IS_RDONLY(inode) (((inode)->i_sb) && ((inode)->i_sb->s_flags & MS_RDONLY))
132 #define IS_NOSUID(inode) __IS_FLG(inode, MS_NOSUID)
133 #define IS_NODEV(inode) __IS_FLG(inode, MS_NODEV)
134 #define IS_NOEXEC(inode) __IS_FLG(inode, MS_NOEXEC)
135 #define IS_SYNC(inode) __IS_FLG(inode, MS_SYNCHRONOUS)
136 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
138 #define IS_QUOTAINIT(inode) ((inode)->i_flags & S_QUOTA)
139 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
140 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
141 #define IS_NOATIME(inode) __IS_FLG(inode, MS_NOATIME)
142 #define IS_NODIRATIME(inode) __IS_FLG(inode, MS_NODIRATIME)
145 /* the read-only stuff doesn't really belong here, but any other place is
146 probably as bad and I don't want to create yet another include file. */
148 #define BLKROSET _IO(0x12,93) /* set device read-only (0 = read-write) */
149 #define BLKROGET _IO(0x12,94) /* get read-only status (0 = read_write) */
150 #define BLKRRPART _IO(0x12,95) /* re-read partition table */
151 #define BLKGETSIZE _IO(0x12,96) /* return device size */
152 #define BLKFLSBUF _IO(0x12,97) /* flush buffer cache */
153 #define BLKRASET _IO(0x12,98) /* Set read ahead for block device */
154 #define BLKRAGET _IO(0x12,99) /* get current read ahead setting */
155 #define BLKFRASET _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
156 #define BLKFRAGET _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
157 #define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
158 #define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
159 #define BLKSSZGET _IO(0x12,104)/* get block device sector size */
161 #define BLKPG _IO(0x12,105)/* See blkpg.h */
162 #define BLKELVGET _IO(0x12,106)/* elevator get */
163 #define BLKELVSET _IO(0x12,107)/* elevator set */
164 /* This was here just to show that the number is taken -
165 probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
169 #define BMAP_IOCTL 1 /* obsolete - kept for compatibility */
170 #define FIBMAP _IO(0x00,1) /* bmap access */
171 #define FIGETBSZ _IO(0x00,2) /* get the block size used for bmap */
175 #include <asm/semaphore.h>
176 #include <asm/byteorder.h>
178 extern void update_atime (struct inode
*);
179 #define UPDATE_ATIME(inode) update_atime (inode)
181 extern void buffer_init(unsigned long);
182 extern void inode_init(void);
183 extern void file_table_init(void);
184 extern void dcache_init(void);
187 #define BH_Uptodate 0 /* 1 if the buffer contains valid data */
188 #define BH_Dirty 1 /* 1 if the buffer is dirty */
189 #define BH_Lock 2 /* 1 if the buffer is locked */
190 #define BH_Req 3 /* 0 if the buffer has been invalidated */
191 #define BH_Mapped 4 /* 1 if the buffer has a disk mapping */
192 #define BH_New 5 /* 1 if the buffer is new and not yet written out */
193 #define BH_Protected 6 /* 1 if the buffer is protected */
196 * Try to keep the most commonly used fields in single cache lines (16
197 * bytes) to improve performance. This ordering should be
198 * particularly beneficial on 32-bit processors.
200 * We use the first 16 bytes for the data which is used in searches
201 * over the block hash lists (ie. getblk() and friends).
203 * The second 16 bytes we use for lru buffer scans, as used by
204 * sync_buffers() and refill_freelist(). -- sct
207 /* First cache line: */
208 struct buffer_head
*b_next
; /* Hash queue list */
209 unsigned long b_blocknr
; /* block number */
210 unsigned short b_size
; /* block size */
211 unsigned short b_list
; /* List that this buffer appears */
212 kdev_t b_dev
; /* device (B_FREE = free) */
214 atomic_t b_count
; /* users using this block */
215 kdev_t b_rdev
; /* Real device */
216 unsigned long b_state
; /* buffer state bitmap (see above) */
217 unsigned long b_flushtime
; /* Time when (dirty) buffer should be written */
219 struct buffer_head
*b_next_free
;/* lru/free list linkage */
220 struct buffer_head
*b_prev_free
;/* doubly linked list of buffers */
221 struct buffer_head
*b_this_page
;/* circular list of buffers in one page */
222 struct buffer_head
*b_reqnext
; /* request queue */
224 struct buffer_head
**b_pprev
; /* doubly linked list of hash-queue */
225 char * b_data
; /* pointer to data block (512 byte) */
226 struct page
*b_page
; /* the page this bh is mapped to */
227 void (*b_end_io
)(struct buffer_head
*bh
, int uptodate
); /* I/O completion */
230 unsigned long b_rsector
; /* Real buffer location on disk */
231 wait_queue_head_t b_wait
;
232 struct kiobuf
* b_kiobuf
; /* kiobuf which owns this IO */
235 typedef void (bh_end_io_t
)(struct buffer_head
*bh
, int uptodate
);
236 void init_buffer(struct buffer_head
*, bh_end_io_t
*, void *);
238 #define __buffer_state(bh, state) (((bh)->b_state & (1UL << BH_##state)) != 0)
240 #define buffer_uptodate(bh) __buffer_state(bh,Uptodate)
241 #define buffer_dirty(bh) __buffer_state(bh,Dirty)
242 #define buffer_locked(bh) __buffer_state(bh,Lock)
243 #define buffer_req(bh) __buffer_state(bh,Req)
244 #define buffer_mapped(bh) __buffer_state(bh,Mapped)
245 #define buffer_new(bh) __buffer_state(bh,New)
246 #define buffer_protected(bh) __buffer_state(bh,Protected)
248 #define bh_offset(bh) ((unsigned long)(bh)->b_data & ~PAGE_MASK)
250 extern void set_bh_page(struct buffer_head
*bh
, struct page
*page
, unsigned long offset
);
252 #define touch_buffer(bh) set_bit(PG_referenced, &bh->b_page->flags)
255 #include <linux/pipe_fs_i.h>
256 #include <linux/minix_fs_i.h>
257 #include <linux/ext2_fs_i.h>
258 #include <linux/hpfs_fs_i.h>
259 #include <linux/ntfs_fs_i.h>
260 #include <linux/msdos_fs_i.h>
261 #include <linux/umsdos_fs_i.h>
262 #include <linux/iso_fs_i.h>
263 #include <linux/nfs_fs_i.h>
264 #include <linux/sysv_fs_i.h>
265 #include <linux/affs_fs_i.h>
266 #include <linux/ufs_fs_i.h>
267 #include <linux/efs_fs_i.h>
268 #include <linux/coda_fs_i.h>
269 #include <linux/romfs_fs_i.h>
270 #include <linux/smb_fs_i.h>
271 #include <linux/hfs_fs_i.h>
272 #include <linux/adfs_fs_i.h>
273 #include <linux/qnx4_fs_i.h>
274 #include <linux/bfs_fs_i.h>
275 #include <linux/udf_fs_i.h>
276 #include <linux/ncp_fs_i.h>
277 #include <linux/proc_fs_i.h>
278 #include <linux/usbdev_fs_i.h>
281 * Attribute flags. These should be or-ed together to figure out what
288 #define ATTR_ATIME 16
289 #define ATTR_MTIME 32
290 #define ATTR_CTIME 64
291 #define ATTR_ATIME_SET 128
292 #define ATTR_MTIME_SET 256
293 #define ATTR_FORCE 512 /* Not a change, but a change it */
294 #define ATTR_ATTR_FLAG 1024
297 * This is the Inode Attributes structure, used for notify_change(). It
298 * uses the above definitions as flags, to know which values have changed.
299 * Also, in this manner, a Filesystem can look at only the values it cares
300 * about. Basically, these are the attributes that the VFS layer can
301 * request to change from the FS layer.
303 * Derek Atkins <warlord@MIT.EDU> 94-10-20
306 unsigned int ia_valid
;
314 unsigned int ia_attr_flags
;
318 * This is the inode attributes flag definitions
320 #define ATTR_FLAG_SYNCRONOUS 1 /* Syncronous write */
321 #define ATTR_FLAG_NOATIME 2 /* Don't update atime */
322 #define ATTR_FLAG_APPEND 4 /* Append-only file */
323 #define ATTR_FLAG_IMMUTABLE 8 /* Immutable file */
324 #define ATTR_FLAG_NODIRATIME 16 /* Don't update atime for directory */
327 * Includes for diskquotas and mount structures.
329 #include <linux/quota.h>
330 #include <linux/mount.h>
333 * oh the beauties of C type declarations.
336 struct address_space
;
338 struct address_space_operations
{
339 int (*writepage
) (struct dentry
*, struct page
*);
340 int (*readpage
)(struct dentry
*, struct page
*);
341 int (*prepare_write
)(struct file
*, struct page
*, unsigned, unsigned);
342 int (*commit_write
)(struct file
*, struct page
*, unsigned, unsigned);
343 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
344 int (*bmap
)(struct address_space
*, long);
347 struct address_space
{
348 struct list_head pages
; /* list of pages */
349 unsigned long nrpages
; /* number of pages */
350 struct address_space_operations
*a_ops
; /* methods */
351 void *host
; /* owner: inode, block_device */
352 struct vm_area_struct
*i_mmap
; /* list of mappings */
353 spinlock_t i_shared_lock
; /* and spinlock protecting it */
356 struct block_device
{
357 struct list_head bd_hash
;
359 /* struct address_space bd_data; */
360 dev_t bd_dev
; /* not a kdev_t - it's a search key */
362 const struct block_device_operations
*bd_op
;
363 struct semaphore bd_sem
; /* open/close mutex */
367 struct list_head i_hash
;
368 struct list_head i_list
;
369 struct list_head i_dentry
;
372 unsigned int i_count
;
383 unsigned long i_blksize
;
384 unsigned long i_blocks
;
385 unsigned long i_version
;
386 struct semaphore i_sem
;
387 struct semaphore i_zombie
;
388 struct inode_operations
*i_op
;
389 struct file_operations
*i_fop
; /* former ->i_op->default_file_ops */
390 struct super_block
*i_sb
;
391 wait_queue_head_t i_wait
;
392 struct file_lock
*i_flock
;
393 struct address_space
*i_mapping
;
394 struct address_space i_data
;
395 struct dquot
*i_dquot
[MAXQUOTAS
];
396 struct pipe_inode_info
*i_pipe
;
397 struct block_device
*i_bdev
;
399 unsigned long i_state
;
401 unsigned int i_flags
;
402 unsigned char i_sock
;
404 atomic_t i_writecount
;
405 unsigned int i_attr_flags
;
408 struct minix_inode_info minix_i
;
409 struct ext2_inode_info ext2_i
;
410 struct hpfs_inode_info hpfs_i
;
411 struct ntfs_inode_info ntfs_i
;
412 struct msdos_inode_info msdos_i
;
413 struct umsdos_inode_info umsdos_i
;
414 struct iso_inode_info isofs_i
;
415 struct nfs_inode_info nfs_i
;
416 struct sysv_inode_info sysv_i
;
417 struct affs_inode_info affs_i
;
418 struct ufs_inode_info ufs_i
;
419 struct efs_inode_info efs_i
;
420 struct romfs_inode_info romfs_i
;
421 struct coda_inode_info coda_i
;
422 struct smb_inode_info smbfs_i
;
423 struct hfs_inode_info hfs_i
;
424 struct adfs_inode_info adfs_i
;
425 struct qnx4_inode_info qnx4_i
;
426 struct bfs_inode_info bfs_i
;
427 struct udf_inode_info udf_i
;
428 struct ncp_inode_info ncpfs_i
;
429 struct proc_inode_info proc_i
;
430 struct socket socket_i
;
431 struct usbdev_inode_info usbdev_i
;
436 /* Inode state bits.. */
442 extern void __mark_inode_dirty(struct inode
*);
443 static inline void mark_inode_dirty(struct inode
*inode
)
445 if (!(inode
->i_state
& I_DIRTY
))
446 __mark_inode_dirty(inode
);
450 int pid
; /* pid or -pgrp where SIGIO should be sent */
451 uid_t uid
, euid
; /* uid/euid of process setting the owner */
452 int signum
; /* posix.1b rt signal to be delivered on IO */
456 struct list_head f_list
;
457 struct dentry
*f_dentry
;
458 struct vfsmount
*f_vfsmnt
;
459 struct file_operations
*f_op
;
461 unsigned int f_flags
;
464 unsigned long f_reada
, f_ramax
, f_raend
, f_ralen
, f_rawin
;
465 struct fown_struct f_owner
;
466 unsigned int f_uid
, f_gid
;
469 unsigned long f_version
;
471 /* needed for tty driver, and maybe others */
474 extern spinlock_t files_lock
;
475 #define file_list_lock() spin_lock(&files_lock);
476 #define file_list_unlock() spin_unlock(&files_lock);
478 #define get_file(x) atomic_inc(&(x)->f_count)
479 #define file_count(x) atomic_read(&(x)->f_count)
481 extern int init_private_file(struct file
*, struct dentry
*, int);
485 #define FL_BROKEN 4 /* broken flock() emulation */
486 #define FL_ACCESS 8 /* for processes suspended by mandatory locking */
487 #define FL_LOCKD 16 /* lock held by rpc.lockd */
490 * The POSIX file lock owner is determined by
491 * the "struct files_struct" in the thread group
492 * (or NULL for no owner - BSD locks).
494 * Lockd stuffs a "host" pointer into this.
496 typedef struct files_struct
*fl_owner_t
;
499 struct file_lock
*fl_next
; /* singly linked list for this inode */
500 struct file_lock
*fl_nextlink
; /* doubly linked list of all locks */
501 struct file_lock
*fl_prevlink
; /* used to simplify lock removal */
502 struct file_lock
*fl_nextblock
; /* circular list of blocked processes */
503 struct file_lock
*fl_prevblock
;
506 wait_queue_head_t fl_wait
;
507 struct file
*fl_file
;
508 unsigned char fl_flags
;
509 unsigned char fl_type
;
513 void (*fl_notify
)(struct file_lock
*); /* unblock callback */
514 void (*fl_insert
)(struct file_lock
*); /* lock insertion callback */
515 void (*fl_remove
)(struct file_lock
*); /* lock removal callback */
518 struct nfs_lock_info nfs_fl
;
522 /* The following constant reflects the upper bound of the file/locking space */
524 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
525 #define OFFSET_MAX INT_LIMIT(loff_t)
528 extern struct file_lock
*file_lock_table
;
530 #include <linux/fcntl.h>
532 extern int fcntl_getlk(unsigned int, struct flock
*);
533 extern int fcntl_setlk(unsigned int, unsigned int, struct flock
*);
536 extern void locks_remove_posix(struct file
*, fl_owner_t
);
537 extern void locks_remove_flock(struct file
*);
538 extern struct file_lock
*posix_test_lock(struct file
*, struct file_lock
*);
539 extern int posix_lock_file(struct file
*, struct file_lock
*, unsigned int);
540 extern void posix_block_lock(struct file_lock
*, struct file_lock
*);
541 extern void posix_unblock_lock(struct file_lock
*);
543 struct fasync_struct
{
546 struct fasync_struct
*fa_next
; /* singly linked list */
547 struct file
*fa_file
;
551 struct dentry
*dentry
;
552 struct vfsmount
*mnt
;
556 #define FASYNC_MAGIC 0x4601
558 extern int fasync_helper(int, struct file
*, int, struct fasync_struct
**);
560 #define DQUOT_USR_ENABLED 0x01 /* User diskquotas enabled */
561 #define DQUOT_GRP_ENABLED 0x02 /* Group diskquotas enabled */
563 struct quota_mount_options
565 unsigned int flags
; /* Flags for diskquotas on this device */
566 struct semaphore dqio_sem
; /* lock device while I/O in progress */
567 struct semaphore dqoff_sem
; /* serialize quota_off() and quota_on() on device */
568 struct file
*files
[MAXQUOTAS
]; /* fp's to quotafiles */
569 time_t inode_expire
[MAXQUOTAS
]; /* expiretime for inode-quota */
570 time_t block_expire
[MAXQUOTAS
]; /* expiretime for block-quota */
571 char rsquash
[MAXQUOTAS
]; /* for quotas threat root as any other user */
578 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
580 #include <linux/minix_fs_sb.h>
581 #include <linux/ext2_fs_sb.h>
582 #include <linux/hpfs_fs_sb.h>
583 #include <linux/ntfs_fs_sb.h>
584 #include <linux/msdos_fs_sb.h>
585 #include <linux/iso_fs_sb.h>
586 #include <linux/nfs_fs_sb.h>
587 #include <linux/sysv_fs_sb.h>
588 #include <linux/affs_fs_sb.h>
589 #include <linux/ufs_fs_sb.h>
590 #include <linux/efs_fs_sb.h>
591 #include <linux/romfs_fs_sb.h>
592 #include <linux/smb_fs_sb.h>
593 #include <linux/hfs_fs_sb.h>
594 #include <linux/adfs_fs_sb.h>
595 #include <linux/qnx4_fs_sb.h>
596 #include <linux/bfs_fs_sb.h>
597 #include <linux/udf_fs_sb.h>
598 #include <linux/ncp_fs_sb.h>
599 #include <linux/usbdev_fs_sb.h>
601 extern struct list_head super_blocks
;
603 #define sb_entry(list) list_entry((list), struct super_block, s_list)
605 struct list_head s_list
; /* Keep this first */
607 unsigned long s_blocksize
;
608 unsigned char s_blocksize_bits
;
609 unsigned char s_lock
;
610 unsigned char s_dirt
;
611 struct file_system_type
*s_type
;
612 struct super_operations
*s_op
;
613 struct dquot_operations
*dq_op
;
614 unsigned long s_flags
;
615 unsigned long s_magic
;
616 struct dentry
*s_root
;
617 wait_queue_head_t s_wait
;
619 struct list_head s_dirty
; /* dirty inodes */
620 struct list_head s_files
;
622 struct block_device
*s_bdev
;
623 struct quota_mount_options s_dquot
; /* Diskquota specific options */
626 struct minix_sb_info minix_sb
;
627 struct ext2_sb_info ext2_sb
;
628 struct hpfs_sb_info hpfs_sb
;
629 struct ntfs_sb_info ntfs_sb
;
630 struct msdos_sb_info msdos_sb
;
631 struct isofs_sb_info isofs_sb
;
632 struct nfs_sb_info nfs_sb
;
633 struct sysv_sb_info sysv_sb
;
634 struct affs_sb_info affs_sb
;
635 struct ufs_sb_info ufs_sb
;
636 struct efs_sb_info efs_sb
;
637 struct romfs_sb_info romfs_sb
;
638 struct smb_sb_info smbfs_sb
;
639 struct hfs_sb_info hfs_sb
;
640 struct adfs_sb_info adfs_sb
;
641 struct qnx4_sb_info qnx4_sb
;
642 struct bfs_sb_info bfs_sb
;
643 struct udf_sb_info udf_sb
;
644 struct ncp_sb_info ncpfs_sb
;
645 struct usbdev_sb_info usbdevfs_sb
;
649 * The next field is for VFS *only*. No filesystems have any business
650 * even looking at it. You had been warned.
652 struct semaphore s_vfs_rename_sem
; /* Kludge */
654 /* The next field is used by knfsd when converting a (inode number based)
655 * file handle into a dentry. As it builds a path in the dcache tree from
656 * the bottom up, there may for a time be a subpath of dentrys which is not
657 * connected to the main tree. This semaphore ensure that there is only ever
658 * one such free path per filesystem. Note that unconnected files (or other
659 * non-directories) are allowed, but not unconnected diretories.
661 struct semaphore s_nfsd_free_path_sem
;
665 * VFS helper functions..
667 extern int vfs_create(struct inode
*, struct dentry
*, int);
668 extern int vfs_mkdir(struct inode
*, struct dentry
*, int);
669 extern int vfs_mknod(struct inode
*, struct dentry
*, int, dev_t
);
670 extern int vfs_symlink(struct inode
*, struct dentry
*, const char *);
671 extern int vfs_link(struct dentry
*, struct inode
*, struct dentry
*);
672 extern int vfs_rmdir(struct inode
*, struct dentry
*);
673 extern int vfs_unlink(struct inode
*, struct dentry
*);
674 extern int vfs_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*);
677 * This is the "filldir" function type, used by readdir() to let
678 * the kernel specify what kind of dirent layout it wants to have.
679 * This allows the kernel to read directories into kernel space or
680 * to have different dirent layouts depending on the binary type.
682 typedef int (*filldir_t
)(void *, const char *, int, off_t
, ino_t
);
684 struct block_device_operations
{
685 int (*open
) (struct inode
*, struct file
*);
686 int (*release
) (struct inode
*, struct file
*);
687 int (*ioctl
) (struct inode
*, struct file
*, unsigned, unsigned long);
688 int (*check_media_change
) (kdev_t
);
689 int (*revalidate
) (kdev_t
);
692 struct file_operations
{
693 loff_t (*llseek
) (struct file
*, loff_t
, int);
694 ssize_t (*read
) (struct file
*, char *, size_t, loff_t
*);
695 ssize_t (*write
) (struct file
*, const char *, size_t, loff_t
*);
696 int (*readdir
) (struct file
*, void *, filldir_t
);
697 unsigned int (*poll
) (struct file
*, struct poll_table_struct
*);
698 int (*ioctl
) (struct inode
*, struct file
*, unsigned int, unsigned long);
699 int (*mmap
) (struct file
*, struct vm_area_struct
*);
700 int (*open
) (struct inode
*, struct file
*);
701 int (*flush
) (struct file
*);
702 int (*release
) (struct inode
*, struct file
*);
703 int (*fsync
) (struct file
*, struct dentry
*);
704 int (*fasync
) (int, struct file
*, int);
705 int (*lock
) (struct file
*, int, struct file_lock
*);
706 ssize_t (*readv
) (struct file
*, const struct iovec
*, unsigned long, loff_t
*);
707 ssize_t (*writev
) (struct file
*, const struct iovec
*, unsigned long, loff_t
*);
710 struct inode_operations
{
711 int (*create
) (struct inode
*,struct dentry
*,int);
712 struct dentry
* (*lookup
) (struct inode
*,struct dentry
*);
713 int (*link
) (struct dentry
*,struct inode
*,struct dentry
*);
714 int (*unlink
) (struct inode
*,struct dentry
*);
715 int (*symlink
) (struct inode
*,struct dentry
*,const char *);
716 int (*mkdir
) (struct inode
*,struct dentry
*,int);
717 int (*rmdir
) (struct inode
*,struct dentry
*);
718 int (*mknod
) (struct inode
*,struct dentry
*,int,int);
719 int (*rename
) (struct inode
*, struct dentry
*,
720 struct inode
*, struct dentry
*);
721 int (*readlink
) (struct dentry
*, char *,int);
722 int (*follow_link
) (struct dentry
*, struct nameidata
*);
723 void (*truncate
) (struct inode
*);
724 int (*permission
) (struct inode
*, int);
725 int (*revalidate
) (struct dentry
*);
726 int (*setattr
) (struct dentry
*, struct iattr
*);
727 int (*getattr
) (struct dentry
*, struct iattr
*);
731 * NOTE: write_inode, delete_inode, clear_inode, put_inode can be called
732 * without the big kernel lock held in all filesystems.
734 struct super_operations
{
735 void (*read_inode
) (struct inode
*);
736 void (*write_inode
) (struct inode
*);
737 void (*put_inode
) (struct inode
*);
738 void (*delete_inode
) (struct inode
*);
739 void (*put_super
) (struct super_block
*);
740 void (*write_super
) (struct super_block
*);
741 int (*statfs
) (struct super_block
*, struct statfs
*);
742 int (*remount_fs
) (struct super_block
*, int *, char *);
743 void (*clear_inode
) (struct inode
*);
744 void (*umount_begin
) (struct super_block
*);
747 struct dquot_operations
{
748 void (*initialize
) (struct inode
*, short);
749 void (*drop
) (struct inode
*);
750 int (*alloc_block
) (const struct inode
*, unsigned long, char);
751 int (*alloc_inode
) (const struct inode
*, unsigned long);
752 void (*free_block
) (const struct inode
*, unsigned long);
753 void (*free_inode
) (const struct inode
*, unsigned long);
754 int (*transfer
) (struct dentry
*, struct iattr
*);
757 struct file_system_type
{
760 struct super_block
*(*read_super
) (struct super_block
*, void *, int);
761 struct module
*owner
;
762 struct file_system_type
* next
;
765 #define DECLARE_FSTYPE(var,type,read,flags) \
766 struct file_system_type var = { \
770 owner: THIS_MODULE, \
773 #define DECLARE_FSTYPE_DEV(var,type,read) \
774 DECLARE_FSTYPE(var,type,read,FS_REQUIRES_DEV)
776 extern int register_filesystem(struct file_system_type
*);
777 extern int unregister_filesystem(struct file_system_type
*);
779 static inline int vfs_statfs(struct super_block
*sb
, struct statfs
*buf
)
783 if (!sb
->s_op
|| !sb
->s_op
->statfs
)
785 memset(buf
, 0, sizeof(struct statfs
));
786 return sb
->s_op
->statfs(sb
, buf
);
789 /* Return value for VFS lock functions - tells locks.c to lock conventionally
790 * REALLY kosha for root NFS and nfs_lock
792 #define LOCK_USE_CLNT 1
794 #define FLOCK_VERIFY_READ 1
795 #define FLOCK_VERIFY_WRITE 2
797 extern int locks_mandatory_locked(struct inode
*);
798 extern int locks_mandatory_area(int, struct inode
*, struct file
*, loff_t
, size_t);
801 * Candidates for mandatory locking have the setgid bit set
802 * but no group execute bit - an otherwise meaningless combination.
804 #define MANDATORY_LOCK(inode) \
805 (IS_MANDLOCK(inode) && ((inode)->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID)
807 static inline int locks_verify_locked(struct inode
*inode
)
809 if (MANDATORY_LOCK(inode
))
810 return locks_mandatory_locked(inode
);
814 extern inline int locks_verify_area(int read_write
, struct inode
*inode
,
815 struct file
*filp
, loff_t offset
,
818 if (inode
->i_flock
&& MANDATORY_LOCK(inode
))
819 return locks_mandatory_area(read_write
, inode
, filp
, offset
, count
);
823 extern inline int locks_verify_truncate(struct inode
*inode
,
827 if (inode
->i_flock
&& MANDATORY_LOCK(inode
))
828 return locks_mandatory_area(
829 FLOCK_VERIFY_WRITE
, inode
, filp
,
830 size
< inode
->i_size
? size
: inode
->i_size
,
831 abs(inode
->i_size
- size
)
839 asmlinkage
long sys_open(const char *, int, int);
840 asmlinkage
long sys_close(unsigned int); /* yes, it's really unsigned */
841 extern int do_close(unsigned int, int); /* yes, it's really unsigned */
842 extern int do_truncate(struct dentry
*, loff_t start
);
843 extern int get_unused_fd(void);
844 extern void put_unused_fd(unsigned int);
846 extern struct file
*filp_open(const char *, int, int);
847 extern struct file
* dentry_open(struct dentry
*, struct vfsmount
*, int);
848 extern int filp_close(struct file
*, fl_owner_t id
);
849 extern char * getname(const char *);
850 #define __getname() ((char *) __get_free_page(GFP_KERNEL))
851 #define putname(name) free_page((unsigned long)(name))
853 enum {BDEV_FILE
, BDEV_SWAP
, BDEV_FS
, BDEV_RAW
};
854 extern void kill_fasync(struct fasync_struct
*, int, int);
855 extern int register_blkdev(unsigned int, const char *, struct block_device_operations
*);
856 extern int unregister_blkdev(unsigned int, const char *);
857 extern struct block_device
*bdget(dev_t
);
858 extern void bdput(struct block_device
*);
859 extern int blkdev_open(struct inode
*, struct file
*);
860 extern struct file_operations def_blk_fops
;
861 extern struct file_operations def_fifo_fops
;
862 extern int ioctl_by_bdev(struct block_device
*, unsigned, unsigned long);
863 extern int blkdev_get(struct block_device
*, mode_t
, unsigned, int);
864 extern int blkdev_put(struct block_device
*, int);
867 extern const struct block_device_operations
*get_blkfops(unsigned int);
868 extern struct file_operations
*get_chrfops(unsigned int, unsigned int);
869 extern int register_chrdev(unsigned int, const char *, struct file_operations
*);
870 extern int unregister_chrdev(unsigned int, const char *);
871 extern int chrdev_open(struct inode
*, struct file
*);
872 extern const char * bdevname(kdev_t
);
873 extern const char * cdevname(kdev_t
);
874 extern const char * kdevname(kdev_t
);
875 extern void init_special_inode(struct inode
*, umode_t
, int);
877 /* Invalid inode operations -- fs/bad_inode.c */
878 extern void make_bad_inode(struct inode
*);
879 extern int is_bad_inode(struct inode
*);
881 extern struct file_operations connecting_fifo_fops
;
882 extern struct file_operations read_fifo_fops
;
883 extern struct file_operations write_fifo_fops
;
884 extern struct file_operations rdwr_fifo_fops
;
885 extern struct file_operations read_pipe_fops
;
886 extern struct file_operations write_pipe_fops
;
887 extern struct file_operations rdwr_pipe_fops
;
889 extern int fs_may_remount_ro(struct super_block
*);
890 extern int fs_may_mount(kdev_t
);
892 extern int try_to_free_buffers(struct page
*);
893 extern void refile_buffer(struct buffer_head
* buf
);
896 #define BUF_LOCKED 1 /* Buffers scheduled for write */
897 #define BUF_DIRTY 2 /* Dirty buffers, not yet scheduled for write */
898 #define BUF_PROTECTED 3 /* Ramdisk persistent storage */
902 * This is called by bh->b_end_io() handlers when I/O has completed.
904 extern inline void mark_buffer_uptodate(struct buffer_head
* bh
, int on
)
907 set_bit(BH_Uptodate
, &bh
->b_state
);
909 clear_bit(BH_Uptodate
, &bh
->b_state
);
912 #define atomic_set_buffer_clean(bh) test_and_clear_bit(BH_Dirty, &(bh)->b_state)
914 extern inline void __mark_buffer_clean(struct buffer_head
*bh
)
919 extern inline void mark_buffer_clean(struct buffer_head
* bh
)
921 if (atomic_set_buffer_clean(bh
))
922 __mark_buffer_clean(bh
);
925 #define atomic_set_buffer_protected(bh) test_and_set_bit(BH_Protected, &(bh)->b_state)
927 extern inline void __mark_buffer_protected(struct buffer_head
*bh
)
932 extern inline void mark_buffer_protected(struct buffer_head
* bh
)
934 if (!atomic_set_buffer_protected(bh
))
935 __mark_buffer_protected(bh
);
938 extern void FASTCALL(__mark_buffer_dirty(struct buffer_head
*bh
, int flag
));
939 extern void FASTCALL(mark_buffer_dirty(struct buffer_head
*bh
, int flag
));
941 #define atomic_set_buffer_dirty(bh) test_and_set_bit(BH_Dirty, &(bh)->b_state)
943 extern void balance_dirty(kdev_t
);
944 extern int check_disk_change(kdev_t
);
945 extern int invalidate_inodes(struct super_block
*);
946 extern void invalidate_inode_pages(struct inode
*);
947 #define invalidate_buffers(dev) __invalidate_buffers((dev), 0)
948 #define destroy_buffers(dev) __invalidate_buffers((dev), 1)
949 extern void __invalidate_buffers(kdev_t dev
, int);
950 extern int floppy_is_wp(int);
951 extern void sync_inodes(kdev_t
);
952 extern void write_inode_now(struct inode
*);
953 extern void sync_dev(kdev_t
);
954 extern int fsync_dev(kdev_t
);
955 extern void sync_supers(kdev_t
);
956 extern int bmap(struct inode
*, int);
957 extern int notify_change(struct dentry
*, struct iattr
*);
958 extern int permission(struct inode
*, int);
959 extern int get_write_access(struct inode
*);
960 extern void put_write_access(struct inode
*);
961 extern struct dentry
* do_mknod(const char *, int, dev_t
);
962 extern int do_pipe(int *);
964 extern int open_namei(const char *, int, int, struct nameidata
*);
966 extern int kernel_read(struct file
*, unsigned long, char *, unsigned long);
967 extern struct file
* open_exec(const char *);
969 /* fs/dcache.c -- generic fs support functions */
970 extern int is_subdir(struct dentry
*, struct dentry
*);
971 extern ino_t
find_inode_number(struct dentry
*, struct qstr
*);
974 * Kernel pointers have redundant information, so we can use a
975 * scheme where we can return either an error code or a dentry
976 * pointer with the same return value.
978 * This should be a per-architecture thing, to allow different
979 * error and pointer decisions.
981 #define ERR_PTR(err) ((void *)((long)(err)))
982 #define PTR_ERR(ptr) ((long)(ptr))
983 #define IS_ERR(ptr) ((unsigned long)(ptr) > (unsigned long)(-1000))
986 * The bitmask for a lookup event:
987 * - follow links at the end
988 * - require a directory
989 * - ending slashes ok even for nonexistent files
990 * - internal "there are more path compnents" flag
992 #define LOOKUP_FOLLOW (1)
993 #define LOOKUP_DIRECTORY (2)
994 #define LOOKUP_SLASHOK (4)
995 #define LOOKUP_CONTINUE (8)
996 #define LOOKUP_POSITIVE (16)
997 #define LOOKUP_PARENT (32)
1000 * "descriptor" for what we're up to with a read for sendfile().
1001 * This allows us to use the same read code yet
1002 * have multiple different users of the data that
1003 * we read from a file.
1005 * The simplest case just copies the data to user
1013 } read_descriptor_t
;
1015 typedef int (*read_actor_t
)(read_descriptor_t
*, struct page
*, unsigned long, unsigned long);
1017 extern struct dentry
* lookup_dentry(const char *, unsigned int);
1018 extern int walk_init(const char *, unsigned, struct nameidata
*);
1019 extern int walk_name(const char *, unsigned, struct nameidata
*);
1020 extern struct dentry
* lookup_one(const char *, struct dentry
*);
1021 extern struct dentry
* __namei(const char *, unsigned int);
1023 #define namei(pathname) __namei(pathname, LOOKUP_FOLLOW)
1024 #define lnamei(pathname) __namei(pathname, 0)
1026 extern void iput(struct inode
*);
1027 extern struct inode
* igrab(struct inode
*);
1028 extern ino_t
iunique(struct super_block
*, ino_t
);
1030 typedef int (*find_inode_t
)(struct inode
*, unsigned long, void *);
1031 extern struct inode
* iget4(struct super_block
*, unsigned long, find_inode_t
, void *);
1032 static inline struct inode
*iget(struct super_block
*sb
, unsigned long ino
)
1034 return iget4(sb
, ino
, NULL
, NULL
);
1037 extern void clear_inode(struct inode
*);
1038 extern struct inode
* get_empty_inode(void);
1040 extern void insert_inode_hash(struct inode
*);
1041 extern void remove_inode_hash(struct inode
*);
1042 extern struct file
* get_empty_filp(void);
1043 extern void file_move(struct file
*f
, struct list_head
*list
);
1044 extern void file_moveto(struct file
*new, struct file
*old
);
1045 extern struct buffer_head
* get_hash_table(kdev_t
, int, int);
1046 extern struct buffer_head
* getblk(kdev_t
, int, int);
1047 extern void ll_rw_block(int, int, struct buffer_head
* bh
[]);
1048 extern int is_read_only(kdev_t
);
1049 extern void __brelse(struct buffer_head
*);
1050 extern inline void brelse(struct buffer_head
*buf
)
1055 extern void __bforget(struct buffer_head
*);
1056 extern inline void bforget(struct buffer_head
*buf
)
1061 extern void set_blocksize(kdev_t
, int);
1062 extern unsigned int get_hardblocksize(kdev_t
);
1063 extern struct buffer_head
* bread(kdev_t
, int, int);
1064 extern struct buffer_head
* breada(kdev_t
, int, int, unsigned int, unsigned int);
1065 extern void wakeup_bdflush(int wait
);
1067 extern int brw_page(int, struct page
*, kdev_t
, int [], int);
1069 typedef int (*writepage_t
)(struct file
*, struct page
*, unsigned long, unsigned long, const char *);
1070 typedef int (get_block_t
)(struct inode
*,long,struct buffer_head
*,int);
1072 /* Generic buffer handling for block filesystems.. */
1073 extern int block_flushpage(struct page
*, unsigned long);
1074 extern int block_symlink(struct inode
*, const char *, int);
1075 extern int block_write_full_page(struct page
*, get_block_t
*);
1076 extern int block_read_full_page(struct page
*, get_block_t
*);
1077 extern int block_prepare_write(struct page
*, unsigned, unsigned, get_block_t
*);
1078 extern int cont_prepare_write(struct page
*, unsigned, unsigned, get_block_t
*,
1080 int generic_block_bmap(struct address_space
*, long, get_block_t
*);
1081 int generic_commit_write(struct file
*, struct page
*, unsigned, unsigned);
1083 extern int generic_file_mmap(struct file
*, struct vm_area_struct
*);
1084 extern ssize_t
generic_file_read(struct file
*, char *, size_t, loff_t
*);
1085 extern ssize_t
generic_file_write(struct file
*, const char *, size_t, loff_t
*);
1086 extern void do_generic_file_read(struct file
*, loff_t
*, read_descriptor_t
*, read_actor_t
);
1088 extern ssize_t
generic_read_dir(struct file
*, char *, size_t, loff_t
*);
1090 extern struct file_operations generic_ro_fops
;
1092 extern int vfs_readlink(struct dentry
*, char *, int, const char *);
1093 extern int vfs_follow_link(struct nameidata
*, const char *);
1094 extern int page_readlink(struct dentry
*, char *, int);
1095 extern int page_follow_link(struct dentry
*, struct nameidata
*);
1096 extern struct inode_operations page_symlink_inode_operations
;
1098 extern int vfs_readdir(struct file
*, filldir_t
, void *);
1100 extern struct super_block
*get_super(kdev_t
);
1101 struct super_block
*get_empty_super(void);
1102 void remove_vfsmnt(kdev_t dev
);
1103 extern void put_super(kdev_t
);
1104 unsigned long generate_cluster(kdev_t
, int b
[], int);
1105 unsigned long generate_cluster_swab32(kdev_t
, int b
[], int);
1106 extern kdev_t ROOT_DEV
;
1107 extern char root_device_name
[];
1110 extern void show_buffers(void);
1111 extern void mount_root(void);
1113 #ifdef CONFIG_BLK_DEV_INITRD
1114 extern kdev_t real_root_dev
;
1115 extern int change_root(kdev_t
, const char *);
1118 extern ssize_t
char_read(struct file
*, char *, size_t, loff_t
*);
1119 extern ssize_t
block_read(struct file
*, char *, size_t, loff_t
*);
1120 extern int read_ahead
[];
1122 extern ssize_t
char_write(struct file
*, const char *, size_t, loff_t
*);
1123 extern ssize_t
block_write(struct file
*, const char *, size_t, loff_t
*);
1125 extern int file_fsync(struct file
*, struct dentry
*);
1126 extern int generic_buffer_fdatasync(struct inode
*inode
, unsigned long start_idx
, unsigned long end_idx
);
1128 extern int inode_change_ok(struct inode
*, struct iattr
*);
1129 extern void inode_setattr(struct inode
*, struct iattr
*);
1132 * Common dentry functions for inclusion in the VFS
1133 * or in other stackable file systems. Some of these
1134 * functions were in linux/fs/ C (VFS) files.
1139 * We need to do a check-parent every time
1140 * after we have locked the parent - to verify
1141 * that the parent is still our parent and
1142 * that we are still hashed onto it..
1144 * This is required in case two processes race
1145 * on removing (or moving) the same entry: the
1146 * parent lock will serialize them, but the
1147 * other process will be too late..
1149 #define check_parent(dir, dentry) \
1150 ((dir) == (dentry)->d_parent && !d_unhashed(dentry))
1153 * Locking the parent is needed to:
1154 * - serialize directory operations
1155 * - make sure the parent doesn't change from
1156 * under us in the middle of an operation.
1158 * NOTE! Right now we'd rather use a "struct inode"
1159 * for this, but as I expect things to move toward
1160 * using dentries instead for most things it is
1161 * probably better to start with the conceptually
1162 * better interface of relying on a path of dentries.
1164 static inline struct dentry
*lock_parent(struct dentry
*dentry
)
1166 struct dentry
*dir
= dget(dentry
->d_parent
);
1168 down(&dir
->d_inode
->i_sem
);
1172 static inline struct dentry
*get_parent(struct dentry
*dentry
)
1174 return dget(dentry
->d_parent
);
1177 static inline void unlock_dir(struct dentry
*dir
)
1179 up(&dir
->d_inode
->i_sem
);
1184 * Whee.. Deadlock country. Happily there are only two VFS
1185 * operations that does this..
1187 static inline void double_down(struct semaphore
*s1
, struct semaphore
*s2
)
1190 if ((unsigned long) s1
< (unsigned long) s2
) {
1191 struct semaphore
*tmp
= s2
;
1200 * Ewwwwwwww... _triple_ lock. We are guaranteed that the 3rd argument is
1201 * not equal to 1st and not equal to 2nd - the first case (target is parent of
1202 * source) would be already caught, the second is plain impossible (target is
1203 * its own parent and that case would be caught even earlier). Very messy.
1204 * I _think_ that it works, but no warranties - please, look it through.
1205 * Pox on bloody lusers who mandated overwriting rename() for directories...
1208 static inline void triple_down(struct semaphore
*s1
,
1209 struct semaphore
*s2
,
1210 struct semaphore
*s3
)
1213 if ((unsigned long) s1
< (unsigned long) s2
) {
1214 if ((unsigned long) s1
< (unsigned long) s3
) {
1215 struct semaphore
*tmp
= s3
;
1218 if ((unsigned long) s1
< (unsigned long) s2
) {
1219 struct semaphore
*tmp
= s2
;
1223 if ((unsigned long) s1
< (unsigned long) s3
) {
1224 struct semaphore
*tmp
= s3
;
1227 if ((unsigned long) s2
< (unsigned long) s3
) {
1228 struct semaphore
*tmp
= s3
;
1233 } else if ((unsigned long) s2
< (unsigned long) s3
) {
1234 struct semaphore
*tmp
= s3
;
1241 static inline void double_up(struct semaphore
*s1
, struct semaphore
*s2
)
1248 static inline void triple_up(struct semaphore
*s1
,
1249 struct semaphore
*s2
,
1250 struct semaphore
*s3
)
1258 static inline void double_lock(struct dentry
*d1
, struct dentry
*d2
)
1260 double_down(&d1
->d_inode
->i_sem
, &d2
->d_inode
->i_sem
);
1263 static inline void double_unlock(struct dentry
*d1
, struct dentry
*d2
)
1265 double_up(&d1
->d_inode
->i_sem
,&d2
->d_inode
->i_sem
);
1270 #endif /* __KERNEL__ */
1272 #endif /* _LINUX_FS_H */