5 * This file has definitions for some important file table
9 #include <linux/limits.h>
10 #include <linux/ioctl.h>
13 * It's silly to have NR_OPEN bigger than NR_FILE, but you can change
14 * the file limit at runtime and only root can increase the per-process
15 * nr_file rlimit, so it's safe to set up a ridiculously high absolute
16 * upper limit on files-per-process.
18 * Some programs (notably those using select()) may have to be
19 * recompiled to take full advantage of the new limits..
22 /* Fixed constants first: */
24 #define INR_OPEN 1024 /* Initial setting for nfile rlimits */
26 #define BLOCK_SIZE_BITS 10
27 #define BLOCK_SIZE (1<<BLOCK_SIZE_BITS)
29 #define SEEK_SET 0 /* seek relative to beginning of file */
30 #define SEEK_CUR 1 /* seek relative to current file position */
31 #define SEEK_END 2 /* seek relative to end of file */
32 #define SEEK_MAX SEEK_END
34 /* And dynamically-tunable limits and defaults: */
35 struct files_stat_struct {
36 int nr_files; /* read only */
37 int nr_free_files; /* read only */
38 int max_files; /* tunable */
41 struct inodes_stat_t {
44 int dummy[5]; /* padding for sysctl ABI compatibility */
48 #define NR_FILE 8192 /* this can well be larger on a larger system */
58 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
59 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
62 /* file is open for reading */
63 #define FMODE_READ ((__force fmode_t)0x1)
64 /* file is open for writing */
65 #define FMODE_WRITE ((__force fmode_t)0x2)
66 /* file is seekable */
67 #define FMODE_LSEEK ((__force fmode_t)0x4)
68 /* file can be accessed using pread */
69 #define FMODE_PREAD ((__force fmode_t)0x8)
70 /* file can be accessed using pwrite */
71 #define FMODE_PWRITE ((__force fmode_t)0x10)
72 /* File is opened for execution with sys_execve / sys_uselib */
73 #define FMODE_EXEC ((__force fmode_t)0x20)
74 /* File is opened with O_NDELAY (only set for block devices) */
75 #define FMODE_NDELAY ((__force fmode_t)0x40)
76 /* File is opened with O_EXCL (only set for block devices) */
77 #define FMODE_EXCL ((__force fmode_t)0x80)
78 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
79 (specialy hack for floppy.c) */
80 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
83 * Don't update ctime and mtime.
85 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
86 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
88 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
90 /* Expect random access pattern */
91 #define FMODE_RANDOM ((__force fmode_t)0x1000)
94 * The below are the various read and write types that we support. Some of
95 * them include behavioral modifiers that send information down to the
96 * block layer and IO scheduler. Terminology:
98 * The block layer uses device plugging to defer IO a little bit, in
99 * the hope that we will see more IO very shortly. This increases
100 * coalescing of adjacent IO and thus reduces the number of IOs we
101 * have to send to the device. It also allows for better queuing,
102 * if the IO isn't mergeable. If the caller is going to be waiting
103 * for the IO, then he must ensure that the device is unplugged so
104 * that the IO is dispatched to the driver.
106 * All IO is handled async in Linux. This is fine for background
107 * writes, but for reads or writes that someone waits for completion
108 * on, we want to notify the block layer and IO scheduler so that they
109 * know about it. That allows them to make better scheduling
110 * decisions. So when the below references 'sync' and 'async', it
111 * is referencing this priority hint.
113 * With that in mind, the available types are:
115 * READ A normal read operation. Device will be plugged.
116 * READ_SYNC A synchronous read. Device is not plugged, caller can
117 * immediately wait on this read without caring about
119 * READA Used for read-ahead operations. Lower priority, and the
120 * block layer could (in theory) choose to ignore this
121 * request if it runs into resource problems.
122 * WRITE A normal async write. Device will be plugged.
123 * SWRITE Like WRITE, but a special case for ll_rw_block() that
124 * tells it to lock the buffer first. Normally a buffer
125 * must be locked before doing IO.
126 * WRITE_SYNC_PLUG Synchronous write. Identical to WRITE, but passes down
127 * the hint that someone will be waiting on this IO
128 * shortly. The device must still be unplugged explicitly,
129 * WRITE_SYNC_PLUG does not do this as we could be
130 * submitting more writes before we actually wait on any
132 * WRITE_SYNC Like WRITE_SYNC_PLUG, but also unplugs the device
133 * immediately after submission. The write equivalent
135 * WRITE_ODIRECT_PLUG Special case write for O_DIRECT only.
137 * SWRITE_SYNC_PLUG Like WRITE_SYNC/WRITE_SYNC_PLUG, but locks the buffer.
139 * WRITE_BARRIER Like WRITE, but tells the block layer that all
140 * previously submitted writes must be safely on storage
141 * before this one is started. Also guarantees that when
142 * this write is complete, it itself is also safely on
143 * storage. Prevents reordering of writes on both sides
151 #define READA 2 /* read-ahead - don't block if no resources */
152 #define SWRITE 3 /* for ll_rw_block() - wait for buffer lock */
153 #define READ_SYNC (READ | (1 << BIO_RW_SYNCIO) | (1 << BIO_RW_UNPLUG))
154 #define READ_META (READ | (1 << BIO_RW_META))
155 #define WRITE_SYNC_PLUG (WRITE | (1 << BIO_RW_SYNCIO) | (1 << BIO_RW_NOIDLE))
156 #define WRITE_SYNC (WRITE_SYNC_PLUG | (1 << BIO_RW_UNPLUG))
157 #define WRITE_ODIRECT_PLUG (WRITE | (1 << BIO_RW_SYNCIO))
158 #define WRITE_META (WRITE | (1 << BIO_RW_META))
159 #define SWRITE_SYNC_PLUG \
160 (SWRITE | (1 << BIO_RW_SYNCIO) | (1 << BIO_RW_NOIDLE))
161 #define SWRITE_SYNC (SWRITE_SYNC_PLUG | (1 << BIO_RW_UNPLUG))
162 #define WRITE_BARRIER (WRITE | (1 << BIO_RW_BARRIER))
165 * These aren't really reads or writes, they pass down information about
166 * parts of device that are now unused by the file system.
168 #define DISCARD_NOBARRIER (WRITE | (1 << BIO_RW_DISCARD))
169 #define DISCARD_BARRIER (DISCARD_NOBARRIER | (1 << BIO_RW_BARRIER))
175 /* public flags for file_system_type */
176 #define FS_REQUIRES_DEV 1
177 #define FS_BINARY_MOUNTDATA 2
178 #define FS_HAS_SUBTYPE 4
179 #define FS_REVAL_DOT 16384 /* Check the paths ".", ".." for staleness */
180 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move()
181 * during rename() internally.
185 * These are the fs-independent mount-flags: up to 32 flags are supported
187 #define MS_RDONLY 1 /* Mount read-only */
188 #define MS_NOSUID 2 /* Ignore suid and sgid bits */
189 #define MS_NODEV 4 /* Disallow access to device special files */
190 #define MS_NOEXEC 8 /* Disallow program execution */
191 #define MS_SYNCHRONOUS 16 /* Writes are synced at once */
192 #define MS_REMOUNT 32 /* Alter flags of a mounted FS */
193 #define MS_MANDLOCK 64 /* Allow mandatory locks on an FS */
194 #define MS_DIRSYNC 128 /* Directory modifications are synchronous */
195 #define MS_NOATIME 1024 /* Do not update access times. */
196 #define MS_NODIRATIME 2048 /* Do not update directory access times */
200 #define MS_VERBOSE 32768 /* War is peace. Verbosity is silence.
201 MS_VERBOSE is deprecated. */
202 #define MS_SILENT 32768
203 #define MS_POSIXACL (1<<16) /* VFS does not apply the umask */
204 #define MS_UNBINDABLE (1<<17) /* change to unbindable */
205 #define MS_PRIVATE (1<<18) /* change to private */
206 #define MS_SLAVE (1<<19) /* change to slave */
207 #define MS_SHARED (1<<20) /* change to shared */
208 #define MS_RELATIME (1<<21) /* Update atime relative to mtime/ctime. */
209 #define MS_KERNMOUNT (1<<22) /* this is a kern_mount call */
210 #define MS_I_VERSION (1<<23) /* Update inode I_version field */
211 #define MS_STRICTATIME (1<<24) /* Always perform atime updates */
212 #define MS_WITHAPPEND (1<<25) /* iop->permission() understands MAY_APPEND */
213 #define MS_ACTIVE (1<<30)
214 #define MS_NOUSER (1<<31)
217 * Superblock flags that can be altered by MS_REMOUNT
219 #define MS_RMT_MASK (MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK|MS_I_VERSION)
222 * Old magic mount flag and mask
224 #define MS_MGC_VAL 0xC0ED0000
225 #define MS_MGC_MSK 0xffff0000
227 /* Inode flags - they have nothing to superblock flags now */
229 #define S_SYNC 1 /* Writes are synced at once */
230 #define S_NOATIME 2 /* Do not update access times */
231 #define S_APPEND 4 /* Append-only file */
232 #define S_IMMUTABLE 8 /* Immutable file */
233 #define S_DEAD 16 /* removed, but still open directory */
234 #define S_NOQUOTA 32 /* Inode is not counted to quota */
235 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
236 #define S_NOCMTIME 128 /* Do not update file c/mtime */
237 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
238 #define S_PRIVATE 512 /* Inode is fs-internal */
241 * Note that nosuid etc flags are inode-specific: setting some file-system
242 * flags just means all the inodes inherit those flags by default. It might be
243 * possible to override it selectively if you really wanted to with some
244 * ioctl() that is not currently implemented.
246 * Exception: MS_RDONLY is always applied to the entire file system.
248 * Unfortunately, it is possible to change a filesystems flags with it mounted
249 * with files in use. This means that all of the inodes will not have their
250 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
251 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
253 #define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))
255 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
256 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
257 ((inode)->i_flags & S_SYNC))
258 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
259 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
260 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
261 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
262 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
263 #define IS_WITHAPPEND(inode) __IS_FLG(inode, MS_WITHAPPEND)
265 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
266 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
267 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
268 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
270 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
271 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
272 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
273 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
275 /* the read-only stuff doesn't really belong here, but any other place is
276 probably as bad and I don't want to create yet another include file. */
278 #define BLKROSET _IO(0x12,93) /* set device read-only (0 = read-write) */
279 #define BLKROGET _IO(0x12,94) /* get read-only status (0 = read_write) */
280 #define BLKRRPART _IO(0x12,95) /* re-read partition table */
281 #define BLKGETSIZE _IO(0x12,96) /* return device size /512 (long *arg) */
282 #define BLKFLSBUF _IO(0x12,97) /* flush buffer cache */
283 #define BLKRASET _IO(0x12,98) /* set read ahead for block device */
284 #define BLKRAGET _IO(0x12,99) /* get current read ahead setting */
285 #define BLKFRASET _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
286 #define BLKFRAGET _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
287 #define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
288 #define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
289 #define BLKSSZGET _IO(0x12,104)/* get block device sector size */
291 #define BLKPG _IO(0x12,105)/* See blkpg.h */
293 /* Some people are morons. Do not use sizeof! */
295 #define BLKELVGET _IOR(0x12,106,size_t)/* elevator get */
296 #define BLKELVSET _IOW(0x12,107,size_t)/* elevator set */
297 /* This was here just to show that the number is taken -
298 probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
300 /* A jump here: 108-111 have been used for various private purposes. */
301 #define BLKBSZGET _IOR(0x12,112,size_t)
302 #define BLKBSZSET _IOW(0x12,113,size_t)
303 #define BLKGETSIZE64 _IOR(0x12,114,size_t) /* return device size in bytes (u64 *arg) */
304 #define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup)
305 #define BLKTRACESTART _IO(0x12,116)
306 #define BLKTRACESTOP _IO(0x12,117)
307 #define BLKTRACETEARDOWN _IO(0x12,118)
308 #define BLKDISCARD _IO(0x12,119)
309 #define BLKIOMIN _IO(0x12,120)
310 #define BLKIOOPT _IO(0x12,121)
311 #define BLKALIGNOFF _IO(0x12,122)
312 #define BLKPBSZGET _IO(0x12,123)
313 #define BLKDISCARDZEROES _IO(0x12,124)
315 #define BMAP_IOCTL 1 /* obsolete - kept for compatibility */
316 #define FIBMAP _IO(0x00,1) /* bmap access */
317 #define FIGETBSZ _IO(0x00,2) /* get the block size used for bmap */
318 #define FIFREEZE _IOWR('X', 119, int) /* Freeze */
319 #define FITHAW _IOWR('X', 120, int) /* Thaw */
321 #define FS_IOC_GETFLAGS _IOR('f', 1, long)
322 #define FS_IOC_SETFLAGS _IOW('f', 2, long)
323 #define FS_IOC_GETVERSION _IOR('v', 1, long)
324 #define FS_IOC_SETVERSION _IOW('v', 2, long)
325 #define FS_IOC_FIEMAP _IOWR('f', 11, struct fiemap)
326 #define FS_IOC32_GETFLAGS _IOR('f', 1, int)
327 #define FS_IOC32_SETFLAGS _IOW('f', 2, int)
328 #define FS_IOC32_GETVERSION _IOR('v', 1, int)
329 #define FS_IOC32_SETVERSION _IOW('v', 2, int)
332 * Inode flags (FS_IOC_GETFLAGS / FS_IOC_SETFLAGS)
334 #define FS_SECRM_FL 0x00000001 /* Secure deletion */
335 #define FS_UNRM_FL 0x00000002 /* Undelete */
336 #define FS_COMPR_FL 0x00000004 /* Compress file */
337 #define FS_SYNC_FL 0x00000008 /* Synchronous updates */
338 #define FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
339 #define FS_APPEND_FL 0x00000020 /* writes to file may only append */
340 #define FS_NODUMP_FL 0x00000040 /* do not dump file */
341 #define FS_NOATIME_FL 0x00000080 /* do not update atime */
342 /* Reserved for compression usage... */
343 #define FS_DIRTY_FL 0x00000100
344 #define FS_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
345 #define FS_NOCOMP_FL 0x00000400 /* Don't compress */
346 #define FS_ECOMPR_FL 0x00000800 /* Compression error */
347 /* End compression flags --- maybe not all used */
348 #define FS_BTREE_FL 0x00001000 /* btree format dir */
349 #define FS_INDEX_FL 0x00001000 /* hash-indexed directory */
350 #define FS_IMAGIC_FL 0x00002000 /* AFS directory */
351 #define FS_JOURNAL_DATA_FL 0x00004000 /* Reserved for ext3 */
352 #define FS_NOTAIL_FL 0x00008000 /* file tail should not be merged */
353 #define FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
354 #define FS_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
355 #define FS_EXTENT_FL 0x00080000 /* Extents */
356 #define FS_DIRECTIO_FL 0x00100000 /* Use direct i/o */
357 #define FS_RESERVED_FL 0x80000000 /* reserved for ext2 lib */
359 #define FS_FL_USER_VISIBLE 0x0003DFFF /* User visible flags */
360 #define FS_FL_USER_MODIFIABLE 0x000380FF /* User modifiable flags */
363 #define SYNC_FILE_RANGE_WAIT_BEFORE 1
364 #define SYNC_FILE_RANGE_WRITE 2
365 #define SYNC_FILE_RANGE_WAIT_AFTER 4
369 #include <linux/linkage.h>
370 #include <linux/wait.h>
371 #include <linux/types.h>
372 #include <linux/kdev_t.h>
373 #include <linux/dcache.h>
374 #include <linux/path.h>
375 #include <linux/stat.h>
376 #include <linux/cache.h>
377 #include <linux/kobject.h>
378 #include <linux/list.h>
379 #include <linux/radix-tree.h>
380 #include <linux/prio_tree.h>
381 #include <linux/init.h>
382 #include <linux/pid.h>
383 #include <linux/mutex.h>
384 #include <linux/capability.h>
385 #include <linux/semaphore.h>
386 #include <linux/fiemap.h>
388 #include <asm/atomic.h>
389 #include <asm/byteorder.h>
391 struct export_operations;
396 struct pipe_inode_info;
397 struct poll_table_struct;
399 struct vm_area_struct;
403 extern void __init inode_init(void);
404 extern void __init inode_init_early(void);
405 extern void __init files_init(unsigned long);
407 extern struct files_stat_struct files_stat;
408 extern int get_max_files(void);
409 extern int sysctl_nr_open;
410 extern struct inodes_stat_t inodes_stat;
411 extern int leases_enable, lease_break_time;
412 #ifdef CONFIG_DNOTIFY
413 extern int dir_notify_enable;
417 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
418 struct buffer_head *bh_result, int create);
419 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
420 ssize_t bytes, void *private);
423 * Attribute flags. These should be or-ed together to figure out what
426 #define ATTR_MODE (1 << 0)
427 #define ATTR_UID (1 << 1)
428 #define ATTR_GID (1 << 2)
429 #define ATTR_SIZE (1 << 3)
430 #define ATTR_ATIME (1 << 4)
431 #define ATTR_MTIME (1 << 5)
432 #define ATTR_CTIME (1 << 6)
433 #define ATTR_ATIME_SET (1 << 7)
434 #define ATTR_MTIME_SET (1 << 8)
435 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
436 #define ATTR_ATTR_FLAG (1 << 10)
437 #define ATTR_KILL_SUID (1 << 11)
438 #define ATTR_KILL_SGID (1 << 12)
439 #define ATTR_FILE (1 << 13)
440 #define ATTR_KILL_PRIV (1 << 14)
441 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
442 #define ATTR_TIMES_SET (1 << 16)
443 #define ATTR_NO_BLOCK (1 << 17) /* Return EAGAIN and don't block on long truncates */
446 * This is the Inode Attributes structure, used for notify_change(). It
447 * uses the above definitions as flags, to know which values have changed.
448 * Also, in this manner, a Filesystem can look at only the values it cares
449 * about. Basically, these are the attributes that the VFS layer can
450 * request to change from the FS layer.
452 * Derek Atkins <warlord@MIT.EDU> 94-10-20
455 unsigned int ia_valid;
460 struct timespec ia_atime;
461 struct timespec ia_mtime;
462 struct timespec ia_ctime;
465 * Not an attribute, but an auxilary info for filesystems wanting to
466 * implement an ftruncate() like method. NOTE: filesystem should
467 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
469 struct file *ia_file;
473 * Includes for diskquotas.
475 #include <linux/quota.h>
478 * enum positive_aop_returns - aop return codes with specific semantics
480 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
481 * completed, that the page is still locked, and
482 * should be considered active. The VM uses this hint
483 * to return the page to the active list -- it won't
484 * be a candidate for writeback again in the near
485 * future. Other callers must be careful to unlock
486 * the page if they get this return. Returned by
489 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
490 * unlocked it and the page might have been truncated.
491 * The caller should back up to acquiring a new page and
492 * trying again. The aop will be taking reasonable
493 * precautions not to livelock. If the caller held a page
494 * reference, it should drop it before retrying. Returned
497 * address_space_operation functions return these large constants to indicate
498 * special semantics to the caller. These are much larger than the bytes in a
499 * page to allow for functions that return the number of bytes operated on in a
503 enum positive_aop_returns {
504 AOP_WRITEPAGE_ACTIVATE = 0x80000,
505 AOP_TRUNCATED_PAGE = 0x80001,
508 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
509 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
510 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
511 * helper code (eg buffer layer)
512 * to clear GFP_FS from alloc */
515 * oh the beauties of C type declarations.
518 struct address_space;
519 struct writeback_control;
522 const struct iovec *iov;
523 unsigned long nr_segs;
528 size_t iov_iter_copy_from_user_atomic(struct page *page,
529 struct iov_iter *i, unsigned long offset, size_t bytes);
530 size_t iov_iter_copy_from_user(struct page *page,
531 struct iov_iter *i, unsigned long offset, size_t bytes);
532 void iov_iter_advance(struct iov_iter *i, size_t bytes);
533 int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes);
534 size_t iov_iter_single_seg_count(struct iov_iter *i);
536 static inline void iov_iter_init(struct iov_iter *i,
537 const struct iovec *iov, unsigned long nr_segs,
538 size_t count, size_t written)
541 i->nr_segs = nr_segs;
543 i->count = count + written;
545 iov_iter_advance(i, written);
548 static inline size_t iov_iter_count(struct iov_iter *i)
554 * "descriptor" for what we're up to with a read.
555 * This allows us to use the same read code yet
556 * have multiple different users of the data that
557 * we read from a file.
559 * The simplest case just copies the data to user
572 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
573 unsigned long, unsigned long);
575 struct address_space_operations {
576 int (*writepage)(struct page *page, struct writeback_control *wbc);
577 int (*readpage)(struct file *, struct page *);
578 void (*sync_page)(struct page *);
580 /* Write back some dirty pages from this mapping. */
581 int (*writepages)(struct address_space *, struct writeback_control *);
583 /* Set a page dirty. Return true if this dirtied it */
584 int (*set_page_dirty)(struct page *page);
586 int (*readpages)(struct file *filp, struct address_space *mapping,
587 struct list_head *pages, unsigned nr_pages);
589 int (*write_begin)(struct file *, struct address_space *mapping,
590 loff_t pos, unsigned len, unsigned flags,
591 struct page **pagep, void **fsdata);
592 int (*write_end)(struct file *, struct address_space *mapping,
593 loff_t pos, unsigned len, unsigned copied,
594 struct page *page, void *fsdata);
596 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
597 sector_t (*bmap)(struct address_space *, sector_t);
598 void (*invalidatepage) (struct page *, unsigned long);
599 int (*releasepage) (struct page *, gfp_t);
600 ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
601 loff_t offset, unsigned long nr_segs);
602 int (*get_xip_mem)(struct address_space *, pgoff_t, int,
603 void **, unsigned long *);
604 /* migrate the contents of a page to the specified target */
605 int (*migratepage) (struct address_space *,
606 struct page *, struct page *);
607 int (*launder_page) (struct page *);
608 int (*is_partially_uptodate) (struct page *, read_descriptor_t *,
610 int (*error_remove_page)(struct address_space *, struct page *);
615 int (*swapon)(struct file *file);
616 int (*swapoff)(struct file *file);
617 int (*swap_out)(struct file *file, struct page *page,
618 struct writeback_control *wbc);
619 int (*swap_in)(struct file *file, struct page *page);
623 * pagecache_write_begin/pagecache_write_end must be used by general code
624 * to write into the pagecache.
626 int pagecache_write_begin(struct file *, struct address_space *mapping,
627 loff_t pos, unsigned len, unsigned flags,
628 struct page **pagep, void **fsdata);
630 int pagecache_write_end(struct file *, struct address_space *mapping,
631 loff_t pos, unsigned len, unsigned copied,
632 struct page *page, void *fsdata);
634 struct backing_dev_info;
635 struct address_space {
636 struct inode *host; /* owner: inode, block_device */
637 struct radix_tree_root page_tree; /* radix tree of all pages */
638 spinlock_t tree_lock; /* and lock protecting it */
639 unsigned int i_mmap_writable;/* count VM_SHARED mappings */
640 struct prio_tree_root i_mmap; /* tree of private and shared mappings */
641 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
642 spinlock_t i_mmap_lock; /* protect tree, count, list */
643 unsigned int truncate_count; /* Cover race condition with truncate */
644 unsigned long nrpages; /* number of total pages */
645 pgoff_t writeback_index;/* writeback starts here */
646 const struct address_space_operations *a_ops; /* methods */
647 unsigned long flags; /* error bits/gfp mask */
648 struct backing_dev_info *backing_dev_info; /* device readahead, etc */
649 spinlock_t private_lock; /* for use by the address_space */
650 struct list_head private_list; /* ditto */
651 struct address_space *assoc_mapping; /* ditto */
652 } __attribute__((aligned(sizeof(long))));
654 * On most architectures that alignment is already the case; but
655 * must be enforced here for CRIS, to let the least signficant bit
656 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
659 struct block_device {
660 dev_t bd_dev; /* not a kdev_t - it's a search key */
661 struct inode * bd_inode; /* will die */
662 struct super_block * bd_super;
664 struct mutex bd_mutex; /* open/close mutex */
665 struct list_head bd_inodes;
670 struct list_head bd_holder_list;
672 struct block_device * bd_contains;
673 unsigned bd_block_size;
674 struct hd_struct * bd_part;
675 /* number of times partitions within this device have been opened. */
676 unsigned bd_part_count;
678 struct gendisk * bd_disk;
679 struct list_head bd_list;
681 * Private data. You must have bd_claim'ed the block_device
682 * to use this. NOTE: bd_claim allows an owner to claim
683 * the same device multiple times, the owner must take special
684 * care to not mess up bd_private for that case.
686 unsigned long bd_private;
688 /* The counter of freeze processes */
689 int bd_fsfreeze_count;
690 /* Mutex for freeze */
691 struct mutex bd_fsfreeze_mutex;
695 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
698 #define PAGECACHE_TAG_DIRTY 0
699 #define PAGECACHE_TAG_WRITEBACK 1
701 int mapping_tagged(struct address_space *mapping, int tag);
704 * Might pages of this file be mapped into userspace?
706 static inline int mapping_mapped(struct address_space *mapping)
708 return !prio_tree_empty(&mapping->i_mmap) ||
709 !list_empty(&mapping->i_mmap_nonlinear);
713 * Might pages of this file have been modified in userspace?
714 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
715 * marks vma as VM_SHARED if it is shared, and the file was opened for
716 * writing i.e. vma may be mprotected writable even if now readonly.
718 static inline int mapping_writably_mapped(struct address_space *mapping)
720 return mapping->i_mmap_writable != 0;
724 * Use sequence counter to get consistent i_size on 32-bit processors.
726 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
727 #include <linux/seqlock.h>
728 #define __NEED_I_SIZE_ORDERED
729 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
731 #define i_size_ordered_init(inode) do { } while (0)
735 #define ACL_NOT_CACHED ((void *)(-1))
738 struct hlist_node i_hash;
739 struct list_head i_list; /* backing dev IO list */
740 struct list_head i_sb_list;
741 struct list_head i_dentry;
744 unsigned int i_nlink;
748 unsigned int i_blkbits;
751 #ifdef __NEED_I_SIZE_ORDERED
752 seqcount_t i_size_seqcount;
754 struct timespec i_atime;
755 struct timespec i_mtime;
756 struct timespec i_ctime;
758 unsigned short i_bytes;
760 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
761 struct mutex i_mutex;
762 struct rw_semaphore i_alloc_sem;
763 const struct inode_operations *i_op;
764 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
765 struct super_block *i_sb;
766 struct file_lock *i_flock;
767 struct address_space *i_mapping;
768 struct address_space i_data;
770 struct dquot *i_dquot[MAXQUOTAS];
772 struct list_head i_devices;
774 struct pipe_inode_info *i_pipe;
775 struct block_device *i_bdev;
781 #ifdef CONFIG_FSNOTIFY
782 __u32 i_fsnotify_mask; /* all events this inode cares about */
783 struct hlist_head i_fsnotify_mark_entries; /* fsnotify mark entries */
786 #ifdef CONFIG_INOTIFY
787 struct list_head inotify_watches; /* watches on this inode */
788 struct mutex inotify_mutex; /* protects the watches list */
791 unsigned long i_state;
792 unsigned long dirtied_when; /* jiffies of first dirtying */
794 unsigned int i_flags;
796 atomic_t i_writecount;
797 #ifdef CONFIG_SECURITY
800 #ifdef CONFIG_FS_POSIX_ACL
801 struct posix_acl *i_acl;
802 struct posix_acl *i_default_acl;
804 void *i_private; /* fs or device private pointer */
808 * inode->i_mutex nesting subclasses for the lock validator:
810 * 0: the object of the current VFS operation
815 * The locking order between these classes is
816 * parent -> child -> normal -> xattr -> quota
818 enum inode_i_mutex_lock_class
828 * NOTE: in a 32bit arch with a preemptable kernel and
829 * an UP compile the i_size_read/write must be atomic
830 * with respect to the local cpu (unlike with preempt disabled),
831 * but they don't need to be atomic with respect to other cpus like in
832 * true SMP (so they need either to either locally disable irq around
833 * the read or for example on x86 they can be still implemented as a
834 * cmpxchg8b without the need of the lock prefix). For SMP compiles
835 * and 64bit archs it makes no difference if preempt is enabled or not.
837 static inline loff_t i_size_read(const struct inode *inode)
839 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
844 seq = read_seqcount_begin(&inode->i_size_seqcount);
845 i_size = inode->i_size;
846 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
848 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
852 i_size = inode->i_size;
856 return inode->i_size;
861 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
862 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
863 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
865 static inline void i_size_write(struct inode *inode, loff_t i_size)
867 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
868 write_seqcount_begin(&inode->i_size_seqcount);
869 inode->i_size = i_size;
870 write_seqcount_end(&inode->i_size_seqcount);
871 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
873 inode->i_size = i_size;
876 inode->i_size = i_size;
880 static inline unsigned iminor(const struct inode *inode)
882 return MINOR(inode->i_rdev);
885 static inline unsigned imajor(const struct inode *inode)
887 return MAJOR(inode->i_rdev);
890 extern struct block_device *I_BDEV(struct inode *inode);
893 rwlock_t lock; /* protects pid, uid, euid fields */
894 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
895 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
896 uid_t uid, euid; /* uid/euid of process setting the owner */
897 int signum; /* posix.1b rt signal to be delivered on IO */
901 * Track a single file's readahead state
903 struct file_ra_state {
904 pgoff_t start; /* where readahead started */
905 unsigned int size; /* # of readahead pages */
906 unsigned int async_size; /* do asynchronous readahead when
907 there are only # of pages ahead */
909 unsigned int ra_pages; /* Maximum readahead window */
910 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
911 loff_t prev_pos; /* Cache last read() position */
915 * Check if @index falls in the readahead windows.
917 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
919 return (index >= ra->start &&
920 index < ra->start + ra->size);
923 #define FILE_MNT_WRITE_TAKEN 1
924 #define FILE_MNT_WRITE_RELEASED 2
928 * fu_list becomes invalid after file_free is called and queued via
929 * fu_rcuhead for RCU freeing
932 struct list_head fu_list;
933 struct rcu_head fu_rcuhead;
936 #define f_dentry f_path.dentry
937 #define f_vfsmnt f_path.mnt
938 const struct file_operations *f_op;
939 spinlock_t f_lock; /* f_ep_links, f_flags, no IRQ */
940 atomic_long_t f_count;
941 unsigned int f_flags;
944 struct fown_struct f_owner;
945 const struct cred *f_cred;
946 struct file_ra_state f_ra;
949 #ifdef CONFIG_SECURITY
952 /* needed for tty driver, and maybe others */
956 /* Used by fs/eventpoll.c to link all the hooks to this file */
957 struct list_head f_ep_links;
958 #endif /* #ifdef CONFIG_EPOLL */
959 struct address_space *f_mapping;
960 #ifdef CONFIG_DEBUG_WRITECOUNT
961 unsigned long f_mnt_write_state;
964 extern spinlock_t files_lock;
965 #define file_list_lock() spin_lock(&files_lock);
966 #define file_list_unlock() spin_unlock(&files_lock);
968 #define get_file(x) atomic_long_inc(&(x)->f_count)
969 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
970 #define file_count(x) atomic_long_read(&(x)->f_count)
972 #ifdef CONFIG_DEBUG_WRITECOUNT
973 static inline void file_take_write(struct file *f)
975 WARN_ON(f->f_mnt_write_state != 0);
976 f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN;
978 static inline void file_release_write(struct file *f)
980 f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED;
982 static inline void file_reset_write(struct file *f)
984 f->f_mnt_write_state = 0;
986 static inline void file_check_state(struct file *f)
989 * At this point, either both or neither of these bits
992 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN);
993 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED);
995 static inline int file_check_writeable(struct file *f)
997 if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN)
999 printk(KERN_WARNING "writeable file with no "
1000 "mnt_want_write()\n");
1004 #else /* !CONFIG_DEBUG_WRITECOUNT */
1005 static inline void file_take_write(struct file *filp) {}
1006 static inline void file_release_write(struct file *filp) {}
1007 static inline void file_reset_write(struct file *filp) {}
1008 static inline void file_check_state(struct file *filp) {}
1009 static inline int file_check_writeable(struct file *filp)
1013 #endif /* CONFIG_DEBUG_WRITECOUNT */
1015 #define MAX_NON_LFS ((1UL<<31) - 1)
1017 /* Page cache limit. The filesystems should put that into their s_maxbytes
1018 limits, otherwise bad things can happen in VM. */
1019 #if BITS_PER_LONG==32
1020 #define MAX_LFS_FILESIZE (((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
1021 #elif BITS_PER_LONG==64
1022 #define MAX_LFS_FILESIZE 0x7fffffffffffffffUL
1027 #define FL_ACCESS 8 /* not trying to lock, just looking */
1028 #define FL_EXISTS 16 /* when unlocking, test for existence */
1029 #define FL_LEASE 32 /* lease held on this file */
1030 #define FL_CLOSE 64 /* unlock on close */
1031 #define FL_SLEEP 128 /* A blocking lock */
1034 * Special return value from posix_lock_file() and vfs_lock_file() for
1035 * asynchronous locking.
1037 #define FILE_LOCK_DEFERRED 1
1040 * The POSIX file lock owner is determined by
1041 * the "struct files_struct" in the thread group
1042 * (or NULL for no owner - BSD locks).
1044 * Lockd stuffs a "host" pointer into this.
1046 typedef struct files_struct *fl_owner_t;
1048 struct file_lock_operations {
1049 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1050 void (*fl_release_private)(struct file_lock *);
1053 struct lock_manager_operations {
1054 int (*fl_compare_owner)(struct file_lock *, struct file_lock *);
1055 void (*fl_notify)(struct file_lock *); /* unblock callback */
1056 int (*fl_grant)(struct file_lock *, struct file_lock *, int);
1057 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1058 void (*fl_release_private)(struct file_lock *);
1059 void (*fl_break)(struct file_lock *);
1060 int (*fl_mylease)(struct file_lock *, struct file_lock *);
1061 int (*fl_change)(struct file_lock **, int);
1064 struct lock_manager {
1065 struct list_head list;
1068 void locks_start_grace(struct lock_manager *);
1069 void locks_end_grace(struct lock_manager *);
1070 int locks_in_grace(void);
1072 /* that will die - we need it for nfs_lock_info */
1073 #include <linux/nfs_fs_i.h>
1076 struct file_lock *fl_next; /* singly linked list for this inode */
1077 struct list_head fl_link; /* doubly linked list of all locks */
1078 struct list_head fl_block; /* circular list of blocked processes */
1079 fl_owner_t fl_owner;
1080 unsigned char fl_flags;
1081 unsigned char fl_type;
1082 unsigned int fl_pid;
1083 struct pid *fl_nspid;
1084 wait_queue_head_t fl_wait;
1085 struct file *fl_file;
1089 struct fasync_struct * fl_fasync; /* for lease break notifications */
1090 unsigned long fl_break_time; /* for nonblocking lease breaks */
1092 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1093 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1095 struct nfs_lock_info nfs_fl;
1096 struct nfs4_lock_info nfs4_fl;
1098 struct list_head link; /* link in AFS vnode's pending_locks list */
1099 int state; /* state of grant or error if -ve */
1104 /* The following constant reflects the upper bound of the file/locking space */
1106 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1107 #define OFFSET_MAX INT_LIMIT(loff_t)
1108 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1111 #include <linux/fcntl.h>
1113 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1115 #ifdef CONFIG_FILE_LOCKING
1116 extern int fcntl_getlk(struct file *, struct flock __user *);
1117 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1118 struct flock __user *);
1120 #if BITS_PER_LONG == 32
1121 extern int fcntl_getlk64(struct file *, struct flock64 __user *);
1122 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1123 struct flock64 __user *);
1126 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1127 extern int fcntl_getlease(struct file *filp);
1130 extern void locks_init_lock(struct file_lock *);
1131 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1132 extern void __locks_copy_lock(struct file_lock *, const struct file_lock *);
1133 extern void locks_remove_posix(struct file *, fl_owner_t);
1134 extern void locks_remove_flock(struct file *);
1135 extern void locks_release_private(struct file_lock *);
1136 extern void posix_test_lock(struct file *, struct file_lock *);
1137 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1138 extern int posix_lock_file_wait(struct file *, struct file_lock *);
1139 extern int posix_unblock_lock(struct file *, struct file_lock *);
1140 extern int vfs_test_lock(struct file *, struct file_lock *);
1141 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1142 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1143 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
1144 extern int __break_lease(struct inode *inode, unsigned int flags);
1145 extern void lease_get_mtime(struct inode *, struct timespec *time);
1146 extern int generic_setlease(struct file *, long, struct file_lock **);
1147 extern int vfs_setlease(struct file *, long, struct file_lock **);
1148 extern int lease_modify(struct file_lock **, int);
1149 extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
1150 extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
1151 #else /* !CONFIG_FILE_LOCKING */
1152 static inline int fcntl_getlk(struct file *file, struct flock __user *user)
1157 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1158 unsigned int cmd, struct flock __user *user)
1163 #if BITS_PER_LONG == 32
1164 static inline int fcntl_getlk64(struct file *file, struct flock64 __user *user)
1169 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1170 unsigned int cmd, struct flock64 __user *user)
1175 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1180 static inline int fcntl_getlease(struct file *filp)
1185 static inline void locks_init_lock(struct file_lock *fl)
1190 static inline void __locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1195 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1200 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1205 static inline void locks_remove_flock(struct file *filp)
1210 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1215 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1216 struct file_lock *conflock)
1221 static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1226 static inline int posix_unblock_lock(struct file *filp,
1227 struct file_lock *waiter)
1232 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1237 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1238 struct file_lock *fl, struct file_lock *conf)
1243 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1248 static inline int flock_lock_file_wait(struct file *filp,
1249 struct file_lock *request)
1254 static inline int __break_lease(struct inode *inode, unsigned int mode)
1259 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1264 static inline int generic_setlease(struct file *filp, long arg,
1265 struct file_lock **flp)
1270 static inline int vfs_setlease(struct file *filp, long arg,
1271 struct file_lock **lease)
1276 static inline int lease_modify(struct file_lock **before, int arg)
1281 static inline int lock_may_read(struct inode *inode, loff_t start,
1287 static inline int lock_may_write(struct inode *inode, loff_t start,
1293 #endif /* !CONFIG_FILE_LOCKING */
1296 struct fasync_struct {
1300 struct fasync_struct *fa_next; /* singly linked list */
1301 struct file *fa_file;
1302 struct rcu_head fa_rcu;
1305 #define FASYNC_MAGIC 0x4601
1307 /* SMP safe fasync helpers: */
1308 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1309 /* can be called from interrupts */
1310 extern void kill_fasync(struct fasync_struct **, int, int);
1312 extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1313 extern int f_setown(struct file *filp, unsigned long arg, int force);
1314 extern void f_delown(struct file *filp);
1315 extern pid_t f_getown(struct file *filp);
1316 extern int send_sigurg(struct fown_struct *fown);
1322 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1323 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1324 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1325 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1326 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1328 extern struct list_head super_blocks;
1329 extern spinlock_t sb_lock;
1331 struct super_block {
1332 struct list_head s_list; /* Keep this first */
1333 dev_t s_dev; /* search index; _not_ kdev_t */
1334 unsigned char s_dirt;
1335 unsigned char s_blocksize_bits;
1336 unsigned long s_blocksize;
1337 loff_t s_maxbytes; /* Max file size */
1338 struct file_system_type *s_type;
1339 const struct super_operations *s_op;
1340 const struct dquot_operations *dq_op;
1341 const struct quotactl_ops *s_qcop;
1342 const struct export_operations *s_export_op;
1343 unsigned long s_flags;
1344 unsigned long s_magic;
1345 struct dentry *s_root;
1346 struct rw_semaphore s_umount;
1347 struct mutex s_lock;
1350 #ifdef CONFIG_SECURITY
1353 const struct xattr_handler **s_xattr;
1355 struct list_head s_inodes; /* all inodes */
1356 struct hlist_head s_anon; /* anonymous dentries for (nfs) exporting */
1357 struct list_head s_files;
1358 /* s_dentry_lru and s_nr_dentry_unused are protected by dcache_lock */
1359 struct list_head s_dentry_lru; /* unused dentry lru */
1360 int s_nr_dentry_unused; /* # of dentry on lru */
1362 struct block_device *s_bdev;
1363 struct backing_dev_info *s_bdi;
1364 struct mtd_info *s_mtd;
1365 struct list_head s_instances;
1366 struct quota_info s_dquot; /* Diskquota specific options */
1369 wait_queue_head_t s_wait_unfrozen;
1371 char s_id[32]; /* Informational name */
1373 void *s_fs_info; /* Filesystem private info */
1376 /* Granularity of c/m/atime in ns.
1377 Cannot be worse than a second */
1381 * The next field is for VFS *only*. No filesystems have any business
1382 * even looking at it. You had been warned.
1384 struct mutex s_vfs_rename_mutex; /* Kludge */
1387 * Filesystem subtype. If non-empty the filesystem type field
1388 * in /proc/mounts will be "type.subtype"
1393 * Saved mount options for lazy filesystems using
1394 * generic_show_options()
1399 extern struct timespec current_fs_time(struct super_block *sb);
1402 * Snapshotting support.
1406 SB_FREEZE_WRITE = 1,
1407 SB_FREEZE_TRANS = 2,
1410 #define vfs_check_frozen(sb, level) \
1411 wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
1413 #define get_fs_excl() atomic_inc(¤t->fs_excl)
1414 #define put_fs_excl() atomic_dec(¤t->fs_excl)
1415 #define has_fs_excl() atomic_read(¤t->fs_excl)
1417 #define is_owner_or_cap(inode) \
1418 ((current_fsuid() == (inode)->i_uid) || capable(CAP_FOWNER))
1420 /* not quite ready to be deprecated, but... */
1421 extern void lock_super(struct super_block *);
1422 extern void unlock_super(struct super_block *);
1425 * VFS helper functions..
1427 extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *);
1428 extern int vfs_mkdir(struct inode *, struct dentry *, int);
1429 extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t);
1430 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1431 extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
1432 extern int vfs_rmdir(struct inode *, struct dentry *);
1433 extern int vfs_unlink(struct inode *, struct dentry *);
1434 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1437 * VFS dentry helper functions.
1439 extern void dentry_unhash(struct dentry *dentry);
1442 * VFS file helper functions.
1444 extern int file_permission(struct file *, int);
1445 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1448 * VFS FS_IOC_FIEMAP helper definitions.
1450 struct fiemap_extent_info {
1451 unsigned int fi_flags; /* Flags as passed from user */
1452 unsigned int fi_extents_mapped; /* Number of mapped extents */
1453 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1454 struct fiemap_extent *fi_extents_start; /* Start of fiemap_extent
1457 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1458 u64 phys, u64 len, u32 flags);
1459 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1464 * NOTE! These match bits 12..15 of stat.st_mode
1465 * (ie "(i_mode >> 12) & 15").
1467 #define DT_UNKNOWN 0
1478 * This is the "filldir" function type, used by readdir() to let
1479 * the kernel specify what kind of dirent layout it wants to have.
1480 * This allows the kernel to read directories into kernel space or
1481 * to have different dirent layouts depending on the binary type.
1483 typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
1484 struct block_device_operations;
1486 /* These macros are for out of kernel modules to test that
1487 * the kernel supports the unlocked_ioctl and compat_ioctl
1488 * fields in struct file_operations. */
1489 #define HAVE_COMPAT_IOCTL 1
1490 #define HAVE_UNLOCKED_IOCTL 1
1494 * read, write, poll, fsync, readv, writev, unlocked_ioctl and compat_ioctl
1495 * can be called without the big kernel lock held in all filesystems.
1497 struct file_operations {
1498 struct module *owner;
1499 loff_t (*llseek) (struct file *, loff_t, int);
1500 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1501 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1502 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1503 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1504 int (*readdir) (struct file *, void *, filldir_t);
1505 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1506 int (*ioctl) (struct inode *, struct file *, unsigned int, unsigned long);
1507 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1508 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1509 int (*mmap) (struct file *, struct vm_area_struct *);
1510 int (*open) (struct inode *, struct file *);
1511 int (*flush) (struct file *, fl_owner_t id);
1512 int (*release) (struct inode *, struct file *);
1513 int (*fsync) (struct file *, int datasync);
1514 int (*aio_fsync) (struct kiocb *, int datasync);
1515 int (*fasync) (int, struct file *, int);
1516 int (*lock) (struct file *, int, struct file_lock *);
1517 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1518 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1519 int (*check_flags)(int);
1520 int (*flock) (struct file *, int, struct file_lock *);
1521 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1522 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1523 #define HAVE_FOP_OPEN_EXEC
1524 int (*open_exec) (struct inode *);
1525 int (*setlease)(struct file *, long, struct file_lock **);
1528 struct inode_operations {
1529 int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
1530 struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
1531 int (*link) (struct dentry *,struct inode *,struct dentry *);
1532 int (*unlink) (struct inode *,struct dentry *);
1533 int (*symlink) (struct inode *,struct dentry *,const char *);
1534 int (*mkdir) (struct inode *,struct dentry *,int);
1535 int (*rmdir) (struct inode *,struct dentry *);
1536 int (*mknod) (struct inode *,struct dentry *,int,dev_t);
1537 int (*rename) (struct inode *, struct dentry *,
1538 struct inode *, struct dentry *);
1539 int (*readlink) (struct dentry *, char __user *,int);
1540 void * (*follow_link) (struct dentry *, struct nameidata *);
1541 void (*put_link) (struct dentry *, struct nameidata *, void *);
1542 void (*truncate) (struct inode *);
1543 int (*permission) (struct inode *, int);
1544 int (*check_acl)(struct inode *, int);
1545 int (*may_create) (struct inode *, int);
1546 int (*may_delete) (struct inode *, struct inode *);
1547 int (*setattr) (struct dentry *, struct iattr *);
1548 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1549 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1550 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1551 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1552 int (*removexattr) (struct dentry *, const char *);
1553 void (*truncate_range)(struct inode *, loff_t, loff_t);
1554 long (*fallocate)(struct inode *inode, int mode, loff_t offset,
1556 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1562 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1563 unsigned long nr_segs, unsigned long fast_segs,
1564 struct iovec *fast_pointer,
1565 struct iovec **ret_pointer);
1567 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1568 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1569 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1570 unsigned long, loff_t *);
1571 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1572 unsigned long, loff_t *);
1574 struct super_operations {
1575 struct inode *(*alloc_inode)(struct super_block *sb);
1576 void (*destroy_inode)(struct inode *);
1578 void (*dirty_inode) (struct inode *);
1579 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1580 void (*drop_inode) (struct inode *);
1581 void (*delete_inode) (struct inode *);
1582 void (*put_super) (struct super_block *);
1583 void (*write_super) (struct super_block *);
1584 int (*sync_fs)(struct super_block *sb, int wait);
1585 int (*freeze_fs) (struct super_block *);
1586 int (*unfreeze_fs) (struct super_block *);
1587 int (*statfs) (struct dentry *, struct kstatfs *);
1588 int (*remount_fs) (struct super_block *, int *, char *);
1589 void (*clear_inode) (struct inode *);
1590 void (*umount_begin) (struct super_block *);
1592 int (*show_options)(struct seq_file *, struct vfsmount *);
1593 int (*show_stats)(struct seq_file *, struct vfsmount *);
1595 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1596 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1598 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1602 * Inode state bits. Protected by inode_lock.
1604 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1605 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1607 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1608 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1609 * various stages of removing an inode.
1611 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1613 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1614 * fdatasync(). i_atime is the usual cause.
1615 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1616 * these changes separately from I_DIRTY_SYNC so that we
1617 * don't have to write inode on fdatasync() when only
1618 * mtime has changed in it.
1619 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1620 * I_NEW Serves as both a mutex and completion notification.
1621 * New inodes set I_NEW. If two processes both create
1622 * the same inode, one of them will release its inode and
1623 * wait for I_NEW to be released before returning.
1624 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1625 * also cause waiting on I_NEW, without I_NEW actually
1626 * being set. find_inode() uses this to prevent returning
1627 * nearly-dead inodes.
1628 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1629 * is zero. I_FREEING must be set when I_WILL_FREE is
1631 * I_FREEING Set when inode is about to be freed but still has dirty
1632 * pages or buffers attached or the inode itself is still
1634 * I_CLEAR Set by clear_inode(). In this state the inode is clean
1635 * and can be destroyed.
1637 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1638 * prohibited for many purposes. iget() must wait for
1639 * the inode to be completely released, then create it
1640 * anew. Other functions will just ignore such inodes,
1641 * if appropriate. I_NEW is used for waiting.
1643 * I_SYNC Synchonized write of dirty inode data. The bits is
1644 * set during data writeback, and cleared with a wakeup
1645 * on the bit address once it is done.
1647 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1649 #define I_DIRTY_SYNC 1
1650 #define I_DIRTY_DATASYNC 2
1651 #define I_DIRTY_PAGES 4
1653 #define I_NEW (1 << __I_NEW)
1654 #define I_WILL_FREE 16
1655 #define I_FREEING 32
1658 #define I_SYNC (1 << __I_SYNC)
1660 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1662 extern void __mark_inode_dirty(struct inode *, int);
1663 static inline void mark_inode_dirty(struct inode *inode)
1665 __mark_inode_dirty(inode, I_DIRTY);
1668 static inline void mark_inode_dirty_sync(struct inode *inode)
1670 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1674 * inc_nlink - directly increment an inode's link count
1677 * This is a low-level filesystem helper to replace any
1678 * direct filesystem manipulation of i_nlink. Currently,
1679 * it is only here for parity with dec_nlink().
1681 static inline void inc_nlink(struct inode *inode)
1686 static inline void inode_inc_link_count(struct inode *inode)
1689 mark_inode_dirty(inode);
1693 * drop_nlink - directly drop an inode's link count
1696 * This is a low-level filesystem helper to replace any
1697 * direct filesystem manipulation of i_nlink. In cases
1698 * where we are attempting to track writes to the
1699 * filesystem, a decrement to zero means an imminent
1700 * write when the file is truncated and actually unlinked
1701 * on the filesystem.
1703 static inline void drop_nlink(struct inode *inode)
1709 * clear_nlink - directly zero an inode's link count
1712 * This is a low-level filesystem helper to replace any
1713 * direct filesystem manipulation of i_nlink. See
1714 * drop_nlink() for why we care about i_nlink hitting zero.
1716 static inline void clear_nlink(struct inode *inode)
1721 static inline void inode_dec_link_count(struct inode *inode)
1724 mark_inode_dirty(inode);
1728 * inode_inc_iversion - increments i_version
1729 * @inode: inode that need to be updated
1731 * Every time the inode is modified, the i_version field will be incremented.
1732 * The filesystem has to be mounted with i_version flag
1735 static inline void inode_inc_iversion(struct inode *inode)
1737 spin_lock(&inode->i_lock);
1739 spin_unlock(&inode->i_lock);
1742 extern void touch_atime(struct vfsmount *mnt, struct dentry *dentry);
1743 static inline void file_accessed(struct file *file)
1745 if (!(file->f_flags & O_NOATIME))
1746 touch_atime(file->f_path.mnt, file->f_path.dentry);
1749 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1751 struct file_system_type {
1754 int (*get_sb) (struct file_system_type *, int,
1755 const char *, void *, struct vfsmount *);
1756 void (*kill_sb) (struct super_block *);
1757 struct module *owner;
1758 struct file_system_type * next;
1759 struct list_head fs_supers;
1761 struct lock_class_key s_lock_key;
1762 struct lock_class_key s_umount_key;
1763 struct lock_class_key s_vfs_rename_key;
1765 struct lock_class_key i_lock_key;
1766 struct lock_class_key i_mutex_key;
1767 struct lock_class_key i_mutex_dir_key;
1768 struct lock_class_key i_alloc_sem_key;
1771 extern int get_sb_ns(struct file_system_type *fs_type, int flags, void *data,
1772 int (*fill_super)(struct super_block *, void *, int),
1773 struct vfsmount *mnt);
1774 extern int get_sb_bdev(struct file_system_type *fs_type,
1775 int flags, const char *dev_name, void *data,
1776 int (*fill_super)(struct super_block *, void *, int),
1777 struct vfsmount *mnt);
1778 extern int get_sb_single(struct file_system_type *fs_type,
1779 int flags, void *data,
1780 int (*fill_super)(struct super_block *, void *, int),
1781 struct vfsmount *mnt);
1782 extern int get_sb_nodev(struct file_system_type *fs_type,
1783 int flags, void *data,
1784 int (*fill_super)(struct super_block *, void *, int),
1785 struct vfsmount *mnt);
1786 void generic_shutdown_super(struct super_block *sb);
1787 void kill_block_super(struct super_block *sb);
1788 void kill_anon_super(struct super_block *sb);
1789 void kill_litter_super(struct super_block *sb);
1790 void deactivate_super(struct super_block *sb);
1791 void deactivate_locked_super(struct super_block *sb);
1792 int set_anon_super(struct super_block *s, void *data);
1793 struct super_block *sget(struct file_system_type *type,
1794 int (*test)(struct super_block *,void *),
1795 int (*set)(struct super_block *,void *),
1797 extern int get_sb_pseudo(struct file_system_type *, char *,
1798 const struct super_operations *ops, unsigned long,
1799 struct vfsmount *mnt);
1800 extern void simple_set_mnt(struct vfsmount *mnt, struct super_block *sb);
1802 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1803 #define fops_get(fops) \
1804 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1805 #define fops_put(fops) \
1806 do { if (fops) module_put((fops)->owner); } while(0)
1808 extern int register_filesystem(struct file_system_type *);
1809 extern int unregister_filesystem(struct file_system_type *);
1810 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1811 #define kern_mount(type) kern_mount_data(type, NULL)
1812 extern int may_umount_tree(struct vfsmount *);
1813 extern int may_umount(struct vfsmount *);
1814 extern long do_mount(char *, char *, char *, unsigned long, void *);
1815 extern struct vfsmount *collect_mounts(struct path *);
1816 extern void drop_collected_mounts(struct vfsmount *);
1817 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1819 extern int vfs_statfs(struct dentry *, struct kstatfs *);
1820 extern int freeze_super(struct super_block *super);
1821 extern int thaw_super(struct super_block *super);
1823 extern int current_umask(void);
1826 extern struct kobject *fs_kobj;
1828 extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1830 #define FLOCK_VERIFY_READ 1
1831 #define FLOCK_VERIFY_WRITE 2
1833 #ifdef CONFIG_FILE_LOCKING
1834 extern int locks_mandatory_locked(struct inode *);
1835 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1838 * Candidates for mandatory locking have the setgid bit set
1839 * but no group execute bit - an otherwise meaningless combination.
1842 static inline int __mandatory_lock(struct inode *ino)
1844 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1848 * ... and these candidates should be on MS_MANDLOCK mounted fs,
1849 * otherwise these will be advisory locks
1852 static inline int mandatory_lock(struct inode *ino)
1854 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1857 static inline int locks_verify_locked(struct inode *inode)
1859 if (mandatory_lock(inode))
1860 return locks_mandatory_locked(inode);
1864 static inline int locks_verify_truncate(struct inode *inode,
1868 if (inode->i_flock && mandatory_lock(inode))
1869 return locks_mandatory_area(
1870 FLOCK_VERIFY_WRITE, inode, filp,
1871 size < inode->i_size ? size : inode->i_size,
1872 (size < inode->i_size ? inode->i_size - size
1873 : size - inode->i_size)
1878 static inline int break_lease(struct inode *inode, unsigned int mode)
1881 return __break_lease(inode, mode);
1884 #else /* !CONFIG_FILE_LOCKING */
1885 static inline int locks_mandatory_locked(struct inode *inode)
1890 static inline int locks_mandatory_area(int rw, struct inode *inode,
1891 struct file *filp, loff_t offset,
1897 static inline int __mandatory_lock(struct inode *inode)
1902 static inline int mandatory_lock(struct inode *inode)
1907 static inline int locks_verify_locked(struct inode *inode)
1912 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
1918 static inline int break_lease(struct inode *inode, unsigned int mode)
1923 #endif /* CONFIG_FILE_LOCKING */
1927 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
1929 extern int do_fallocate(struct file *file, int mode, loff_t offset,
1931 extern long do_sys_open(int dfd, const char __user *filename, int flags,
1933 extern struct file *filp_open(const char *, int, int);
1934 extern struct file * dentry_open(struct dentry *, struct vfsmount *, int,
1935 const struct cred *);
1936 extern int filp_close(struct file *, fl_owner_t id);
1937 extern char * getname(const char __user *);
1941 extern int ioctl_preallocate(struct file *filp, void __user *argp);
1944 extern void __init vfs_caches_init_early(void);
1945 extern void __init vfs_caches_init(unsigned long);
1947 extern struct kmem_cache *names_cachep;
1949 #define __getname_gfp(gfp) kmem_cache_alloc(names_cachep, (gfp))
1950 #define __getname() __getname_gfp(GFP_KERNEL)
1951 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
1952 #ifndef CONFIG_AUDITSYSCALL
1953 #define putname(name) __putname(name)
1955 extern void putname(const char *name);
1959 extern int register_blkdev(unsigned int, const char *);
1960 extern void unregister_blkdev(unsigned int, const char *);
1961 extern struct block_device *bdget(dev_t);
1962 extern struct block_device *bdgrab(struct block_device *bdev);
1963 extern void bd_set_size(struct block_device *, loff_t size);
1964 extern void bd_forget(struct inode *inode);
1965 extern void bdput(struct block_device *);
1966 extern struct block_device *open_by_devnum(dev_t, fmode_t);
1967 extern void invalidate_bdev(struct block_device *);
1968 extern int sync_blockdev(struct block_device *bdev);
1969 extern struct super_block *freeze_bdev(struct block_device *);
1970 extern void emergency_thaw_all(void);
1971 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
1972 extern int fsync_bdev(struct block_device *);
1974 static inline void bd_forget(struct inode *inode) {}
1975 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
1976 static inline void invalidate_bdev(struct block_device *bdev) {}
1978 static inline struct super_block *freeze_bdev(struct block_device *sb)
1983 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
1988 extern int sync_filesystem(struct super_block *);
1989 extern const struct file_operations def_blk_fops;
1990 extern const struct file_operations def_chr_fops;
1991 extern const struct file_operations bad_sock_fops;
1992 extern const struct file_operations def_fifo_fops;
1994 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
1995 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
1996 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
1997 extern int blkdev_get(struct block_device *, fmode_t);
1998 extern int blkdev_put(struct block_device *, fmode_t);
1999 extern int bd_claim(struct block_device *, void *);
2000 extern void bd_release(struct block_device *);
2002 extern int bd_claim_by_disk(struct block_device *, void *, struct gendisk *);
2003 extern void bd_release_from_disk(struct block_device *, struct gendisk *);
2005 #define bd_claim_by_disk(bdev, holder, disk) bd_claim(bdev, holder)
2006 #define bd_release_from_disk(bdev, disk) bd_release(bdev)
2011 #define CHRDEV_MAJOR_HASH_SIZE 255
2012 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2013 extern int register_chrdev_region(dev_t, unsigned, const char *);
2014 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2015 unsigned int count, const char *name,
2016 const struct file_operations *fops);
2017 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2018 unsigned int count, const char *name);
2019 extern void unregister_chrdev_region(dev_t, unsigned);
2020 extern void chrdev_show(struct seq_file *,off_t);
2022 static inline int register_chrdev(unsigned int major, const char *name,
2023 const struct file_operations *fops)
2025 return __register_chrdev(major, 0, 256, name, fops);
2028 static inline void unregister_chrdev(unsigned int major, const char *name)
2030 __unregister_chrdev(major, 0, 256, name);
2033 /* fs/block_dev.c */
2034 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2035 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2038 #define BLKDEV_MAJOR_HASH_SIZE 255
2039 extern const char *__bdevname(dev_t, char *buffer);
2040 extern const char *bdevname(struct block_device *bdev, char *buffer);
2041 extern struct block_device *lookup_bdev(const char *);
2042 extern struct block_device *open_bdev_exclusive(const char *, fmode_t, void *);
2043 extern void close_bdev_exclusive(struct block_device *, fmode_t);
2044 extern void blkdev_show(struct seq_file *,off_t);
2047 #define BLKDEV_MAJOR_HASH_SIZE 0
2050 extern void init_special_inode(struct inode *, umode_t, dev_t);
2052 /* Invalid inode operations -- fs/bad_inode.c */
2053 extern void make_bad_inode(struct inode *);
2054 extern int is_bad_inode(struct inode *);
2056 extern const struct file_operations read_pipefifo_fops;
2057 extern const struct file_operations write_pipefifo_fops;
2058 extern const struct file_operations rdwr_pipefifo_fops;
2060 extern int fs_may_remount_ro(struct super_block *);
2064 * return READ, READA, or WRITE
2066 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2069 * return data direction, READ or WRITE
2071 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2073 extern void check_disk_size_change(struct gendisk *disk,
2074 struct block_device *bdev);
2075 extern int revalidate_disk(struct gendisk *);
2076 extern int check_disk_change(struct block_device *);
2077 extern int __invalidate_device(struct block_device *);
2078 extern int invalidate_partition(struct gendisk *, int);
2080 extern int invalidate_inodes(struct super_block *);
2081 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2082 pgoff_t start, pgoff_t end);
2084 static inline void invalidate_remote_inode(struct inode *inode)
2086 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2087 S_ISLNK(inode->i_mode))
2088 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2090 extern int invalidate_inode_pages2(struct address_space *mapping);
2091 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2092 pgoff_t start, pgoff_t end);
2093 extern int write_inode_now(struct inode *, int);
2094 extern int filemap_fdatawrite(struct address_space *);
2095 extern int filemap_flush(struct address_space *);
2096 extern int filemap_fdatawait(struct address_space *);
2097 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2099 extern int filemap_write_and_wait(struct address_space *mapping);
2100 extern int filemap_write_and_wait_range(struct address_space *mapping,
2101 loff_t lstart, loff_t lend);
2102 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2103 loff_t start, loff_t end, int sync_mode);
2104 extern int filemap_fdatawrite_range(struct address_space *mapping,
2105 loff_t start, loff_t end);
2107 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2109 extern int vfs_fsync(struct file *file, int datasync);
2110 extern int generic_write_sync(struct file *file, loff_t pos, loff_t count);
2111 extern void sync_supers(void);
2112 extern void emergency_sync(void);
2113 extern void emergency_remount(void);
2115 extern sector_t bmap(struct inode *, sector_t);
2117 extern int notify_change(struct dentry *, struct iattr *);
2118 extern int inode_permission(struct inode *, int);
2119 extern int generic_permission(struct inode *, int,
2120 int (*check_acl)(struct inode *, int));
2122 static inline bool execute_ok(struct inode *inode)
2124 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2127 extern int get_write_access(struct inode *);
2128 extern int deny_write_access(struct file *);
2129 static inline void put_write_access(struct inode * inode)
2131 atomic_dec(&inode->i_writecount);
2133 static inline void allow_write_access(struct file *file)
2136 atomic_inc(&file->f_path.dentry->d_inode->i_writecount);
2138 extern int do_pipe_flags(int *, int);
2139 extern struct file *create_read_pipe(struct file *f, int flags);
2140 extern struct file *create_write_pipe(int flags);
2141 extern void free_write_pipe(struct file *);
2143 extern struct file *do_filp_open(int dfd, const char *pathname,
2144 int open_flag, int mode, int acc_mode);
2145 extern int may_open(struct path *, int, int);
2147 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2148 extern struct file * open_exec(const char *);
2150 /* fs/dcache.c -- generic fs support functions */
2151 extern int is_subdir(struct dentry *, struct dentry *);
2152 extern int path_is_under(struct path *, struct path *);
2153 extern ino_t find_inode_number(struct dentry *, struct qstr *);
2155 #include <linux/err.h>
2157 /* needed for stackable file system support */
2158 extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
2160 extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
2162 extern int inode_init_always(struct super_block *, struct inode *);
2163 extern void inode_init_once(struct inode *);
2164 extern void inode_add_to_lists(struct super_block *, struct inode *);
2165 extern void iput(struct inode *);
2166 extern struct inode * igrab(struct inode *);
2167 extern ino_t iunique(struct super_block *, ino_t);
2168 extern int inode_needs_sync(struct inode *inode);
2169 extern void generic_delete_inode(struct inode *inode);
2170 extern void generic_drop_inode(struct inode *inode);
2171 extern int generic_detach_inode(struct inode *inode);
2173 extern struct inode *ilookup5_nowait(struct super_block *sb,
2174 unsigned long hashval, int (*test)(struct inode *, void *),
2176 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2177 int (*test)(struct inode *, void *), void *data);
2178 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2180 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2181 extern struct inode * iget_locked(struct super_block *, unsigned long);
2182 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2183 extern int insert_inode_locked(struct inode *);
2184 extern void unlock_new_inode(struct inode *);
2186 extern void __iget(struct inode * inode);
2187 extern void iget_failed(struct inode *);
2188 extern void clear_inode(struct inode *);
2189 extern void destroy_inode(struct inode *);
2190 extern void __destroy_inode(struct inode *);
2191 extern struct inode *new_inode(struct super_block *);
2192 extern int should_remove_suid(struct dentry *);
2193 extern int file_remove_suid(struct file *);
2195 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2196 extern void remove_inode_hash(struct inode *);
2197 static inline void insert_inode_hash(struct inode *inode) {
2198 __insert_inode_hash(inode, inode->i_ino);
2201 extern void file_move(struct file *f, struct list_head *list);
2202 extern void file_kill(struct file *f);
2205 extern void submit_bio(int, struct bio *);
2206 extern int bdev_read_only(struct block_device *);
2208 extern int set_blocksize(struct block_device *, int);
2209 extern int sb_set_blocksize(struct super_block *, int);
2210 extern int sb_min_blocksize(struct super_block *, int);
2212 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2213 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2214 extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
2215 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
2216 extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2217 extern ssize_t __generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long,
2219 extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2220 extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
2221 unsigned long *, loff_t, loff_t *, size_t, size_t);
2222 extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
2223 unsigned long, loff_t, loff_t *, size_t, ssize_t);
2224 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2225 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2226 extern int generic_segment_checks(const struct iovec *iov,
2227 unsigned long *nr_segs, size_t *count, int access_flags);
2229 /* fs/block_dev.c */
2230 extern ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov,
2231 unsigned long nr_segs, loff_t pos);
2232 extern int blkdev_fsync(struct file *filp, int datasync);
2235 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2236 struct pipe_inode_info *, size_t, unsigned int);
2237 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2238 struct pipe_inode_info *, size_t, unsigned int);
2239 extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
2240 struct file *, loff_t *, size_t, unsigned int);
2241 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2242 struct file *out, loff_t *, size_t len, unsigned int flags);
2243 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2244 size_t len, unsigned int flags);
2247 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2248 extern loff_t noop_llseek(struct file *file, loff_t offset, int origin);
2249 extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
2250 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
2251 extern loff_t generic_file_llseek_unlocked(struct file *file, loff_t offset,
2253 extern int generic_file_open(struct inode * inode, struct file * filp);
2254 extern int nonseekable_open(struct inode * inode, struct file * filp);
2256 #ifdef CONFIG_FS_XIP
2257 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2259 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
2260 extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2261 size_t len, loff_t *ppos);
2262 extern int xip_truncate_page(struct address_space *mapping, loff_t from);
2264 static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2272 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2273 loff_t file_offset);
2274 void dio_end_io(struct bio *bio, int error);
2276 ssize_t __blockdev_direct_IO_newtrunc(int rw, struct kiocb *iocb, struct inode *inode,
2277 struct block_device *bdev, const struct iovec *iov, loff_t offset,
2278 unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
2279 dio_submit_t submit_io, int lock_type);
2280 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2281 struct block_device *bdev, const struct iovec *iov, loff_t offset,
2282 unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
2283 dio_submit_t submit_io, int lock_type);
2286 /* need locking between buffered and direct access */
2289 /* filesystem does not support filling holes */
2290 DIO_SKIP_HOLES = 0x02,
2293 static inline ssize_t blockdev_direct_IO_newtrunc(int rw, struct kiocb *iocb,
2294 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
2295 loff_t offset, unsigned long nr_segs, get_block_t get_block,
2296 dio_iodone_t end_io)
2298 return __blockdev_direct_IO_newtrunc(rw, iocb, inode, bdev, iov, offset,
2299 nr_segs, get_block, end_io, NULL,
2300 DIO_LOCKING | DIO_SKIP_HOLES);
2303 static inline ssize_t blockdev_direct_IO_no_locking_newtrunc(int rw, struct kiocb *iocb,
2304 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
2305 loff_t offset, unsigned long nr_segs, get_block_t get_block,
2306 dio_iodone_t end_io)
2308 return __blockdev_direct_IO_newtrunc(rw, iocb, inode, bdev, iov, offset,
2309 nr_segs, get_block, end_io, NULL, 0);
2311 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
2312 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
2313 loff_t offset, unsigned long nr_segs, get_block_t get_block,
2314 dio_iodone_t end_io)
2316 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
2317 nr_segs, get_block, end_io, NULL,
2318 DIO_LOCKING | DIO_SKIP_HOLES);
2321 static inline ssize_t blockdev_direct_IO_no_locking(int rw, struct kiocb *iocb,
2322 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
2323 loff_t offset, unsigned long nr_segs, get_block_t get_block,
2324 dio_iodone_t end_io)
2326 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
2327 nr_segs, get_block, end_io, NULL, 0);
2331 extern const struct file_operations generic_ro_fops;
2333 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2335 extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
2336 extern int vfs_follow_link(struct nameidata *, const char *);
2337 extern int page_readlink(struct dentry *, char __user *, int);
2338 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2339 extern void page_put_link(struct dentry *, struct nameidata *, void *);
2340 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2342 extern int page_symlink(struct inode *inode, const char *symname, int len);
2343 extern const struct inode_operations page_symlink_inode_operations;
2344 extern int generic_readlink(struct dentry *, char __user *, int);
2345 extern void generic_fillattr(struct inode *, struct kstat *);
2346 extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2347 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2348 void inode_add_bytes(struct inode *inode, loff_t bytes);
2349 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2350 loff_t inode_get_bytes(struct inode *inode);
2351 void inode_set_bytes(struct inode *inode, loff_t bytes);
2353 extern int vfs_readdir(struct file *, filldir_t, void *);
2355 extern int vfs_stat(char __user *, struct kstat *);
2356 extern int vfs_lstat(char __user *, struct kstat *);
2357 extern int vfs_fstat(unsigned int, struct kstat *);
2358 extern int vfs_fstatat(int , char __user *, struct kstat *, int);
2360 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2362 extern int __generic_block_fiemap(struct inode *inode,
2363 struct fiemap_extent_info *fieinfo,
2364 loff_t start, loff_t len,
2365 get_block_t *get_block);
2366 extern int generic_block_fiemap(struct inode *inode,
2367 struct fiemap_extent_info *fieinfo, u64 start,
2368 u64 len, get_block_t *get_block);
2370 extern void get_filesystem(struct file_system_type *fs);
2371 extern void put_filesystem(struct file_system_type *fs);
2372 extern struct file_system_type *get_fs_type(const char *name);
2373 extern struct super_block *get_super(struct block_device *);
2374 extern struct super_block *get_active_super(struct block_device *bdev);
2375 extern struct super_block *user_get_super(dev_t);
2376 extern void drop_super(struct super_block *sb);
2377 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2379 extern int dcache_dir_open(struct inode *, struct file *);
2380 extern int dcache_dir_close(struct inode *, struct file *);
2381 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2382 extern int dcache_readdir(struct file *, void *, filldir_t);
2383 extern int simple_setattr(struct dentry *, struct iattr *);
2384 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2385 extern int simple_statfs(struct dentry *, struct kstatfs *);
2386 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2387 extern int simple_unlink(struct inode *, struct dentry *);
2388 extern int simple_rmdir(struct inode *, struct dentry *);
2389 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2390 extern int simple_setsize(struct inode *, loff_t);
2391 extern int noop_fsync(struct file *, int);
2392 extern int simple_empty(struct dentry *);
2393 extern int simple_readpage(struct file *file, struct page *page);
2394 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2395 loff_t pos, unsigned len, unsigned flags,
2396 struct page **pagep, void **fsdata);
2397 extern int simple_write_end(struct file *file, struct address_space *mapping,
2398 loff_t pos, unsigned len, unsigned copied,
2399 struct page *page, void *fsdata);
2401 extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
2402 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2403 extern const struct file_operations simple_dir_operations;
2404 extern const struct inode_operations simple_dir_inode_operations;
2405 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2406 struct dentry *d_alloc_name(struct dentry *, const char *);
2407 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2408 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2409 extern void simple_release_fs(struct vfsmount **mount, int *count);
2411 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2412 loff_t *ppos, const void *from, size_t available);
2413 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2414 const void __user *from, size_t count);
2416 extern int generic_file_fsync(struct file *, int);
2418 #ifdef CONFIG_MIGRATION
2419 extern int buffer_migrate_page(struct address_space *,
2420 struct page *, struct page *);
2422 #define buffer_migrate_page NULL
2425 extern int inode_change_ok(const struct inode *, struct iattr *);
2426 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2427 extern int __must_check inode_setattr(struct inode *, const struct iattr *);
2428 extern void generic_setattr(struct inode *inode, const struct iattr *attr);
2430 extern void file_update_time(struct file *file);
2432 extern int generic_show_options(struct seq_file *m, struct vfsmount *mnt);
2433 extern void save_mount_options(struct super_block *sb, char *options);
2434 extern void replace_mount_options(struct super_block *sb, char *options);
2436 static inline ino_t parent_ino(struct dentry *dentry)
2440 spin_lock(&dentry->d_lock);
2441 res = dentry->d_parent->d_inode->i_ino;
2442 spin_unlock(&dentry->d_lock);
2446 /* Transaction based IO helpers */
2449 * An argresp is stored in an allocated page and holds the
2450 * size of the argument or response, along with its content
2452 struct simple_transaction_argresp {
2457 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2459 char *simple_transaction_get(struct file *file, const char __user *buf,
2461 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2462 size_t size, loff_t *pos);
2463 int simple_transaction_release(struct inode *inode, struct file *file);
2465 void simple_transaction_set(struct file *file, size_t n);
2468 * simple attribute files
2470 * These attributes behave similar to those in sysfs:
2472 * Writing to an attribute immediately sets a value, an open file can be
2473 * written to multiple times.
2475 * Reading from an attribute creates a buffer from the value that might get
2476 * read with multiple read calls. When the attribute has been read
2477 * completely, no further read calls are possible until the file is opened
2480 * All attributes contain a text representation of a numeric value
2481 * that are accessed with the get() and set() functions.
2483 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2484 static int __fops ## _open(struct inode *inode, struct file *file) \
2486 __simple_attr_check_format(__fmt, 0ull); \
2487 return simple_attr_open(inode, file, __get, __set, __fmt); \
2489 static const struct file_operations __fops = { \
2490 .owner = THIS_MODULE, \
2491 .open = __fops ## _open, \
2492 .release = simple_attr_release, \
2493 .read = simple_attr_read, \
2494 .write = simple_attr_write, \
2497 static inline void __attribute__((format(printf, 1, 2)))
2498 __simple_attr_check_format(const char *fmt, ...)
2500 /* don't do anything, just let the compiler check the arguments; */
2503 int simple_attr_open(struct inode *inode, struct file *file,
2504 int (*get)(void *, u64 *), int (*set)(void *, u64),
2506 int simple_attr_release(struct inode *inode, struct file *file);
2507 ssize_t simple_attr_read(struct file *file, char __user *buf,
2508 size_t len, loff_t *ppos);
2509 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2510 size_t len, loff_t *ppos);
2513 int proc_nr_files(struct ctl_table *table, int write,
2514 void __user *buffer, size_t *lenp, loff_t *ppos);
2516 int __init get_filesystem_list(char *buf);
2518 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2519 #define OPEN_FMODE(flag) ((__force fmode_t)((flag + 1) & O_ACCMODE))
2521 #endif /* __KERNEL__ */
2522 #endif /* _LINUX_FS_H */