5 * This file has definitions for some important file table
9 #include <linux/limits.h>
10 #include <linux/ioctl.h>
11 #include <linux/blk_types.h>
14 * It's silly to have NR_OPEN bigger than NR_FILE, but you can change
15 * the file limit at runtime and only root can increase the per-process
16 * nr_file rlimit, so it's safe to set up a ridiculously high absolute
17 * upper limit on files-per-process.
19 * Some programs (notably those using select()) may have to be
20 * recompiled to take full advantage of the new limits..
23 /* Fixed constants first: */
25 #define INR_OPEN 1024 /* Initial setting for nfile rlimits */
27 #define BLOCK_SIZE_BITS 10
28 #define BLOCK_SIZE (1<<BLOCK_SIZE_BITS)
30 #define SEEK_SET 0 /* seek relative to beginning of file */
31 #define SEEK_CUR 1 /* seek relative to current file position */
32 #define SEEK_END 2 /* seek relative to end of file */
33 #define SEEK_MAX SEEK_END
41 /* And dynamically-tunable limits and defaults: */
42 struct files_stat_struct {
43 int nr_files; /* read only */
44 int nr_free_files; /* read only */
45 int max_files; /* tunable */
48 struct inodes_stat_t {
51 int dummy[5]; /* padding for sysctl ABI compatibility */
55 #define NR_FILE 8192 /* this can well be larger on a larger system */
66 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
67 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
70 /* file is open for reading */
71 #define FMODE_READ ((__force fmode_t)0x1)
72 /* file is open for writing */
73 #define FMODE_WRITE ((__force fmode_t)0x2)
74 /* file is seekable */
75 #define FMODE_LSEEK ((__force fmode_t)0x4)
76 /* file can be accessed using pread */
77 #define FMODE_PREAD ((__force fmode_t)0x8)
78 /* file can be accessed using pwrite */
79 #define FMODE_PWRITE ((__force fmode_t)0x10)
80 /* File is opened for execution with sys_execve / sys_uselib */
81 #define FMODE_EXEC ((__force fmode_t)0x20)
82 /* File is opened with O_NDELAY (only set for block devices) */
83 #define FMODE_NDELAY ((__force fmode_t)0x40)
84 /* File is opened with O_EXCL (only set for block devices) */
85 #define FMODE_EXCL ((__force fmode_t)0x80)
86 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
87 (specialy hack for floppy.c) */
88 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
91 * Don't update ctime and mtime.
93 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
94 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
96 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
98 /* Expect random access pattern */
99 #define FMODE_RANDOM ((__force fmode_t)0x1000)
101 /* File was opened by fanotify and shouldn't generate fanotify events */
102 #define FMODE_NONOTIFY ((__force fmode_t)0x1000000)
105 * The below are the various read and write types that we support. Some of
106 * them include behavioral modifiers that send information down to the
107 * block layer and IO scheduler. Terminology:
109 * The block layer uses device plugging to defer IO a little bit, in
110 * the hope that we will see more IO very shortly. This increases
111 * coalescing of adjacent IO and thus reduces the number of IOs we
112 * have to send to the device. It also allows for better queuing,
113 * if the IO isn't mergeable. If the caller is going to be waiting
114 * for the IO, then he must ensure that the device is unplugged so
115 * that the IO is dispatched to the driver.
117 * All IO is handled async in Linux. This is fine for background
118 * writes, but for reads or writes that someone waits for completion
119 * on, we want to notify the block layer and IO scheduler so that they
120 * know about it. That allows them to make better scheduling
121 * decisions. So when the below references 'sync' and 'async', it
122 * is referencing this priority hint.
124 * With that in mind, the available types are:
126 * READ A normal read operation. Device will be plugged.
127 * READ_SYNC A synchronous read. Device is not plugged, caller can
128 * immediately wait on this read without caring about
130 * READA Used for read-ahead operations. Lower priority, and the
131 * block layer could (in theory) choose to ignore this
132 * request if it runs into resource problems.
133 * WRITE A normal async write. Device will be plugged.
134 * WRITE_SYNC_PLUG Synchronous write. Identical to WRITE, but passes down
135 * the hint that someone will be waiting on this IO
136 * shortly. The device must still be unplugged explicitly,
137 * WRITE_SYNC_PLUG does not do this as we could be
138 * submitting more writes before we actually wait on any
140 * WRITE_SYNC Like WRITE_SYNC_PLUG, but also unplugs the device
141 * immediately after submission. The write equivalent
143 * WRITE_ODIRECT_PLUG Special case write for O_DIRECT only.
144 * WRITE_BARRIER Like WRITE_SYNC, but tells the block layer that all
145 * previously submitted writes must be safely on storage
146 * before this one is started. Also guarantees that when
147 * this write is complete, it itself is also safely on
148 * storage. Prevents reordering of writes on both sides
152 #define RW_MASK REQ_WRITE
153 #define RWA_MASK REQ_RAHEAD
156 #define WRITE RW_MASK
157 #define READA RWA_MASK
159 #define READ_SYNC (READ | REQ_SYNC | REQ_UNPLUG)
160 #define READ_META (READ | REQ_META)
161 #define WRITE_SYNC_PLUG (WRITE | REQ_SYNC | REQ_NOIDLE)
162 #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_UNPLUG)
163 #define WRITE_ODIRECT_PLUG (WRITE | REQ_SYNC)
164 #define WRITE_META (WRITE | REQ_META)
165 #define WRITE_BARRIER (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_UNPLUG | \
169 * These aren't really reads or writes, they pass down information about
170 * parts of device that are now unused by the file system.
172 #define DISCARD_NOBARRIER (WRITE | REQ_DISCARD)
173 #define DISCARD_BARRIER (WRITE | REQ_DISCARD | REQ_HARDBARRIER)
174 #define DISCARD_SECURE (DISCARD_NOBARRIER | REQ_SECURE)
180 /* public flags for file_system_type */
181 #define FS_REQUIRES_DEV 1
182 #define FS_BINARY_MOUNTDATA 2
183 #define FS_HAS_SUBTYPE 4
184 #define FS_REVAL_DOT 16384 /* Check the paths ".", ".." for staleness */
185 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move()
186 * during rename() internally.
190 * These are the fs-independent mount-flags: up to 32 flags are supported
192 #define MS_RDONLY 1 /* Mount read-only */
193 #define MS_NOSUID 2 /* Ignore suid and sgid bits */
194 #define MS_NODEV 4 /* Disallow access to device special files */
195 #define MS_NOEXEC 8 /* Disallow program execution */
196 #define MS_SYNCHRONOUS 16 /* Writes are synced at once */
197 #define MS_REMOUNT 32 /* Alter flags of a mounted FS */
198 #define MS_MANDLOCK 64 /* Allow mandatory locks on an FS */
199 #define MS_DIRSYNC 128 /* Directory modifications are synchronous */
200 #define MS_NOATIME 1024 /* Do not update access times. */
201 #define MS_NODIRATIME 2048 /* Do not update directory access times */
205 #define MS_VERBOSE 32768 /* War is peace. Verbosity is silence.
206 MS_VERBOSE is deprecated. */
207 #define MS_SILENT 32768
208 #define MS_POSIXACL (1<<16) /* VFS does not apply the umask */
209 #define MS_UNBINDABLE (1<<17) /* change to unbindable */
210 #define MS_PRIVATE (1<<18) /* change to private */
211 #define MS_SLAVE (1<<19) /* change to slave */
212 #define MS_SHARED (1<<20) /* change to shared */
213 #define MS_RELATIME (1<<21) /* Update atime relative to mtime/ctime. */
214 #define MS_KERNMOUNT (1<<22) /* this is a kern_mount call */
215 #define MS_I_VERSION (1<<23) /* Update inode I_version field */
216 #define MS_STRICTATIME (1<<24) /* Always perform atime updates */
217 #define MS_BORN (1<<29)
218 #define MS_ACTIVE (1<<30)
219 #define MS_NOUSER (1<<31)
222 * Superblock flags that can be altered by MS_REMOUNT
224 #define MS_RMT_MASK (MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK|MS_I_VERSION)
227 * Old magic mount flag and mask
229 #define MS_MGC_VAL 0xC0ED0000
230 #define MS_MGC_MSK 0xffff0000
232 /* Inode flags - they have nothing to superblock flags now */
234 #define S_SYNC 1 /* Writes are synced at once */
235 #define S_NOATIME 2 /* Do not update access times */
236 #define S_APPEND 4 /* Append-only file */
237 #define S_IMMUTABLE 8 /* Immutable file */
238 #define S_DEAD 16 /* removed, but still open directory */
239 #define S_NOQUOTA 32 /* Inode is not counted to quota */
240 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
241 #define S_NOCMTIME 128 /* Do not update file c/mtime */
242 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
243 #define S_PRIVATE 512 /* Inode is fs-internal */
246 * Note that nosuid etc flags are inode-specific: setting some file-system
247 * flags just means all the inodes inherit those flags by default. It might be
248 * possible to override it selectively if you really wanted to with some
249 * ioctl() that is not currently implemented.
251 * Exception: MS_RDONLY is always applied to the entire file system.
253 * Unfortunately, it is possible to change a filesystems flags with it mounted
254 * with files in use. This means that all of the inodes will not have their
255 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
256 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
258 #define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))
260 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
261 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
262 ((inode)->i_flags & S_SYNC))
263 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
264 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
265 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
266 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
267 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
269 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
270 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
271 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
272 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
274 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
275 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
276 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
277 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
279 /* the read-only stuff doesn't really belong here, but any other place is
280 probably as bad and I don't want to create yet another include file. */
282 #define BLKROSET _IO(0x12,93) /* set device read-only (0 = read-write) */
283 #define BLKROGET _IO(0x12,94) /* get read-only status (0 = read_write) */
284 #define BLKRRPART _IO(0x12,95) /* re-read partition table */
285 #define BLKGETSIZE _IO(0x12,96) /* return device size /512 (long *arg) */
286 #define BLKFLSBUF _IO(0x12,97) /* flush buffer cache */
287 #define BLKRASET _IO(0x12,98) /* set read ahead for block device */
288 #define BLKRAGET _IO(0x12,99) /* get current read ahead setting */
289 #define BLKFRASET _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
290 #define BLKFRAGET _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
291 #define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
292 #define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
293 #define BLKSSZGET _IO(0x12,104)/* get block device sector size */
295 #define BLKPG _IO(0x12,105)/* See blkpg.h */
297 /* Some people are morons. Do not use sizeof! */
299 #define BLKELVGET _IOR(0x12,106,size_t)/* elevator get */
300 #define BLKELVSET _IOW(0x12,107,size_t)/* elevator set */
301 /* This was here just to show that the number is taken -
302 probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
304 /* A jump here: 108-111 have been used for various private purposes. */
305 #define BLKBSZGET _IOR(0x12,112,size_t)
306 #define BLKBSZSET _IOW(0x12,113,size_t)
307 #define BLKGETSIZE64 _IOR(0x12,114,size_t) /* return device size in bytes (u64 *arg) */
308 #define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup)
309 #define BLKTRACESTART _IO(0x12,116)
310 #define BLKTRACESTOP _IO(0x12,117)
311 #define BLKTRACETEARDOWN _IO(0x12,118)
312 #define BLKDISCARD _IO(0x12,119)
313 #define BLKIOMIN _IO(0x12,120)
314 #define BLKIOOPT _IO(0x12,121)
315 #define BLKALIGNOFF _IO(0x12,122)
316 #define BLKPBSZGET _IO(0x12,123)
317 #define BLKDISCARDZEROES _IO(0x12,124)
318 #define BLKSECDISCARD _IO(0x12,125)
320 #define BMAP_IOCTL 1 /* obsolete - kept for compatibility */
321 #define FIBMAP _IO(0x00,1) /* bmap access */
322 #define FIGETBSZ _IO(0x00,2) /* get the block size used for bmap */
323 #define FIFREEZE _IOWR('X', 119, int) /* Freeze */
324 #define FITHAW _IOWR('X', 120, int) /* Thaw */
325 #define FITRIM _IOWR('X', 121, struct fstrim_range) /* Trim */
327 #define FS_IOC_GETFLAGS _IOR('f', 1, long)
328 #define FS_IOC_SETFLAGS _IOW('f', 2, long)
329 #define FS_IOC_GETVERSION _IOR('v', 1, long)
330 #define FS_IOC_SETVERSION _IOW('v', 2, long)
331 #define FS_IOC_FIEMAP _IOWR('f', 11, struct fiemap)
332 #define FS_IOC32_GETFLAGS _IOR('f', 1, int)
333 #define FS_IOC32_SETFLAGS _IOW('f', 2, int)
334 #define FS_IOC32_GETVERSION _IOR('v', 1, int)
335 #define FS_IOC32_SETVERSION _IOW('v', 2, int)
338 * Inode flags (FS_IOC_GETFLAGS / FS_IOC_SETFLAGS)
340 #define FS_SECRM_FL 0x00000001 /* Secure deletion */
341 #define FS_UNRM_FL 0x00000002 /* Undelete */
342 #define FS_COMPR_FL 0x00000004 /* Compress file */
343 #define FS_SYNC_FL 0x00000008 /* Synchronous updates */
344 #define FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
345 #define FS_APPEND_FL 0x00000020 /* writes to file may only append */
346 #define FS_NODUMP_FL 0x00000040 /* do not dump file */
347 #define FS_NOATIME_FL 0x00000080 /* do not update atime */
348 /* Reserved for compression usage... */
349 #define FS_DIRTY_FL 0x00000100
350 #define FS_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
351 #define FS_NOCOMP_FL 0x00000400 /* Don't compress */
352 #define FS_ECOMPR_FL 0x00000800 /* Compression error */
353 /* End compression flags --- maybe not all used */
354 #define FS_BTREE_FL 0x00001000 /* btree format dir */
355 #define FS_INDEX_FL 0x00001000 /* hash-indexed directory */
356 #define FS_IMAGIC_FL 0x00002000 /* AFS directory */
357 #define FS_JOURNAL_DATA_FL 0x00004000 /* Reserved for ext3 */
358 #define FS_NOTAIL_FL 0x00008000 /* file tail should not be merged */
359 #define FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
360 #define FS_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
361 #define FS_EXTENT_FL 0x00080000 /* Extents */
362 #define FS_DIRECTIO_FL 0x00100000 /* Use direct i/o */
363 #define FS_RESERVED_FL 0x80000000 /* reserved for ext2 lib */
365 #define FS_FL_USER_VISIBLE 0x0003DFFF /* User visible flags */
366 #define FS_FL_USER_MODIFIABLE 0x000380FF /* User modifiable flags */
369 #define SYNC_FILE_RANGE_WAIT_BEFORE 1
370 #define SYNC_FILE_RANGE_WRITE 2
371 #define SYNC_FILE_RANGE_WAIT_AFTER 4
375 #include <linux/linkage.h>
376 #include <linux/wait.h>
377 #include <linux/types.h>
378 #include <linux/kdev_t.h>
379 #include <linux/dcache.h>
380 #include <linux/path.h>
381 #include <linux/stat.h>
382 #include <linux/cache.h>
383 #include <linux/kobject.h>
384 #include <linux/list.h>
385 #include <linux/radix-tree.h>
386 #include <linux/prio_tree.h>
387 #include <linux/init.h>
388 #include <linux/pid.h>
389 #include <linux/mutex.h>
390 #include <linux/capability.h>
391 #include <linux/semaphore.h>
392 #include <linux/fiemap.h>
394 #include <asm/atomic.h>
395 #include <asm/byteorder.h>
397 struct export_operations;
402 struct pipe_inode_info;
403 struct poll_table_struct;
405 struct vm_area_struct;
409 extern void __init inode_init(void);
410 extern void __init inode_init_early(void);
411 extern void __init files_init(unsigned long);
413 extern struct files_stat_struct files_stat;
414 extern int get_max_files(void);
415 extern int sysctl_nr_open;
416 extern struct inodes_stat_t inodes_stat;
417 extern int leases_enable, lease_break_time;
420 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
421 struct buffer_head *bh_result, int create);
422 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
423 ssize_t bytes, void *private, int ret,
427 * Attribute flags. These should be or-ed together to figure out what
430 #define ATTR_MODE (1 << 0)
431 #define ATTR_UID (1 << 1)
432 #define ATTR_GID (1 << 2)
433 #define ATTR_SIZE (1 << 3)
434 #define ATTR_ATIME (1 << 4)
435 #define ATTR_MTIME (1 << 5)
436 #define ATTR_CTIME (1 << 6)
437 #define ATTR_ATIME_SET (1 << 7)
438 #define ATTR_MTIME_SET (1 << 8)
439 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
440 #define ATTR_ATTR_FLAG (1 << 10)
441 #define ATTR_KILL_SUID (1 << 11)
442 #define ATTR_KILL_SGID (1 << 12)
443 #define ATTR_FILE (1 << 13)
444 #define ATTR_KILL_PRIV (1 << 14)
445 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
446 #define ATTR_TIMES_SET (1 << 16)
449 * This is the Inode Attributes structure, used for notify_change(). It
450 * uses the above definitions as flags, to know which values have changed.
451 * Also, in this manner, a Filesystem can look at only the values it cares
452 * about. Basically, these are the attributes that the VFS layer can
453 * request to change from the FS layer.
455 * Derek Atkins <warlord@MIT.EDU> 94-10-20
458 unsigned int ia_valid;
463 struct timespec ia_atime;
464 struct timespec ia_mtime;
465 struct timespec ia_ctime;
468 * Not an attribute, but an auxilary info for filesystems wanting to
469 * implement an ftruncate() like method. NOTE: filesystem should
470 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
472 struct file *ia_file;
476 * Includes for diskquotas.
478 #include <linux/quota.h>
481 * enum positive_aop_returns - aop return codes with specific semantics
483 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
484 * completed, that the page is still locked, and
485 * should be considered active. The VM uses this hint
486 * to return the page to the active list -- it won't
487 * be a candidate for writeback again in the near
488 * future. Other callers must be careful to unlock
489 * the page if they get this return. Returned by
492 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
493 * unlocked it and the page might have been truncated.
494 * The caller should back up to acquiring a new page and
495 * trying again. The aop will be taking reasonable
496 * precautions not to livelock. If the caller held a page
497 * reference, it should drop it before retrying. Returned
500 * address_space_operation functions return these large constants to indicate
501 * special semantics to the caller. These are much larger than the bytes in a
502 * page to allow for functions that return the number of bytes operated on in a
506 enum positive_aop_returns {
507 AOP_WRITEPAGE_ACTIVATE = 0x80000,
508 AOP_TRUNCATED_PAGE = 0x80001,
511 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
512 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
513 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
514 * helper code (eg buffer layer)
515 * to clear GFP_FS from alloc */
518 * oh the beauties of C type declarations.
521 struct address_space;
522 struct writeback_control;
525 const struct iovec *iov;
526 unsigned long nr_segs;
531 size_t iov_iter_copy_from_user_atomic(struct page *page,
532 struct iov_iter *i, unsigned long offset, size_t bytes);
533 size_t iov_iter_copy_from_user(struct page *page,
534 struct iov_iter *i, unsigned long offset, size_t bytes);
535 void iov_iter_advance(struct iov_iter *i, size_t bytes);
536 int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes);
537 size_t iov_iter_single_seg_count(struct iov_iter *i);
539 static inline void iov_iter_init(struct iov_iter *i,
540 const struct iovec *iov, unsigned long nr_segs,
541 size_t count, size_t written)
544 i->nr_segs = nr_segs;
546 i->count = count + written;
548 iov_iter_advance(i, written);
551 static inline size_t iov_iter_count(struct iov_iter *i)
557 * "descriptor" for what we're up to with a read.
558 * This allows us to use the same read code yet
559 * have multiple different users of the data that
560 * we read from a file.
562 * The simplest case just copies the data to user
575 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
576 unsigned long, unsigned long);
578 struct address_space_operations {
579 int (*writepage)(struct page *page, struct writeback_control *wbc);
580 int (*readpage)(struct file *, struct page *);
581 void (*sync_page)(struct page *);
583 /* Write back some dirty pages from this mapping. */
584 int (*writepages)(struct address_space *, struct writeback_control *);
586 /* Set a page dirty. Return true if this dirtied it */
587 int (*set_page_dirty)(struct page *page);
589 int (*readpages)(struct file *filp, struct address_space *mapping,
590 struct list_head *pages, unsigned nr_pages);
592 int (*write_begin)(struct file *, struct address_space *mapping,
593 loff_t pos, unsigned len, unsigned flags,
594 struct page **pagep, void **fsdata);
595 int (*write_end)(struct file *, struct address_space *mapping,
596 loff_t pos, unsigned len, unsigned copied,
597 struct page *page, void *fsdata);
599 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
600 sector_t (*bmap)(struct address_space *, sector_t);
601 void (*invalidatepage) (struct page *, unsigned long);
602 int (*releasepage) (struct page *, gfp_t);
603 ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
604 loff_t offset, unsigned long nr_segs);
605 int (*get_xip_mem)(struct address_space *, pgoff_t, int,
606 void **, unsigned long *);
607 /* migrate the contents of a page to the specified target */
608 int (*migratepage) (struct address_space *,
609 struct page *, struct page *);
610 int (*launder_page) (struct page *);
611 int (*is_partially_uptodate) (struct page *, read_descriptor_t *,
613 int (*error_remove_page)(struct address_space *, struct page *);
617 * pagecache_write_begin/pagecache_write_end must be used by general code
618 * to write into the pagecache.
620 int pagecache_write_begin(struct file *, struct address_space *mapping,
621 loff_t pos, unsigned len, unsigned flags,
622 struct page **pagep, void **fsdata);
624 int pagecache_write_end(struct file *, struct address_space *mapping,
625 loff_t pos, unsigned len, unsigned copied,
626 struct page *page, void *fsdata);
628 struct backing_dev_info;
629 struct address_space {
630 struct inode *host; /* owner: inode, block_device */
631 struct radix_tree_root page_tree; /* radix tree of all pages */
632 spinlock_t tree_lock; /* and lock protecting it */
633 unsigned int i_mmap_writable;/* count VM_SHARED mappings */
634 struct prio_tree_root i_mmap; /* tree of private and shared mappings */
635 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
636 spinlock_t i_mmap_lock; /* protect tree, count, list */
637 unsigned int truncate_count; /* Cover race condition with truncate */
638 unsigned long nrpages; /* number of total pages */
639 pgoff_t writeback_index;/* writeback starts here */
640 const struct address_space_operations *a_ops; /* methods */
641 unsigned long flags; /* error bits/gfp mask */
642 struct backing_dev_info *backing_dev_info; /* device readahead, etc */
643 spinlock_t private_lock; /* for use by the address_space */
644 struct list_head private_list; /* ditto */
645 struct address_space *assoc_mapping; /* ditto */
646 } __attribute__((aligned(sizeof(long))));
648 * On most architectures that alignment is already the case; but
649 * must be enforced here for CRIS, to let the least signficant bit
650 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
653 struct block_device {
654 dev_t bd_dev; /* not a kdev_t - it's a search key */
655 struct inode * bd_inode; /* will die */
656 struct super_block * bd_super;
658 struct mutex bd_mutex; /* open/close mutex */
659 struct list_head bd_inodes;
664 struct list_head bd_holder_list;
666 struct block_device * bd_contains;
667 unsigned bd_block_size;
668 struct hd_struct * bd_part;
669 /* number of times partitions within this device have been opened. */
670 unsigned bd_part_count;
672 struct gendisk * bd_disk;
673 struct list_head bd_list;
675 * Private data. You must have bd_claim'ed the block_device
676 * to use this. NOTE: bd_claim allows an owner to claim
677 * the same device multiple times, the owner must take special
678 * care to not mess up bd_private for that case.
680 unsigned long bd_private;
682 /* The counter of freeze processes */
683 int bd_fsfreeze_count;
684 /* Mutex for freeze */
685 struct mutex bd_fsfreeze_mutex;
689 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
692 #define PAGECACHE_TAG_DIRTY 0
693 #define PAGECACHE_TAG_WRITEBACK 1
694 #define PAGECACHE_TAG_TOWRITE 2
696 int mapping_tagged(struct address_space *mapping, int tag);
699 * Might pages of this file be mapped into userspace?
701 static inline int mapping_mapped(struct address_space *mapping)
703 return !prio_tree_empty(&mapping->i_mmap) ||
704 !list_empty(&mapping->i_mmap_nonlinear);
708 * Might pages of this file have been modified in userspace?
709 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
710 * marks vma as VM_SHARED if it is shared, and the file was opened for
711 * writing i.e. vma may be mprotected writable even if now readonly.
713 static inline int mapping_writably_mapped(struct address_space *mapping)
715 return mapping->i_mmap_writable != 0;
719 * Use sequence counter to get consistent i_size on 32-bit processors.
721 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
722 #include <linux/seqlock.h>
723 #define __NEED_I_SIZE_ORDERED
724 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
726 #define i_size_ordered_init(inode) do { } while (0)
730 #define ACL_NOT_CACHED ((void *)(-1))
733 struct hlist_node i_hash;
734 struct list_head i_list; /* backing dev IO list */
735 struct list_head i_sb_list;
736 struct list_head i_dentry;
739 unsigned int i_nlink;
743 unsigned int i_blkbits;
746 #ifdef __NEED_I_SIZE_ORDERED
747 seqcount_t i_size_seqcount;
749 struct timespec i_atime;
750 struct timespec i_mtime;
751 struct timespec i_ctime;
753 unsigned short i_bytes;
755 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
756 struct mutex i_mutex;
757 struct rw_semaphore i_alloc_sem;
758 const struct inode_operations *i_op;
759 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
760 struct super_block *i_sb;
761 struct file_lock *i_flock;
762 struct address_space *i_mapping;
763 struct address_space i_data;
765 struct dquot *i_dquot[MAXQUOTAS];
767 struct list_head i_devices;
769 struct pipe_inode_info *i_pipe;
770 struct block_device *i_bdev;
776 #ifdef CONFIG_FSNOTIFY
777 __u32 i_fsnotify_mask; /* all events this inode cares about */
778 struct hlist_head i_fsnotify_marks;
781 unsigned long i_state;
782 unsigned long dirtied_when; /* jiffies of first dirtying */
784 unsigned int i_flags;
786 atomic_t i_writecount;
787 #ifdef CONFIG_SECURITY
790 #ifdef CONFIG_FS_POSIX_ACL
791 struct posix_acl *i_acl;
792 struct posix_acl *i_default_acl;
794 void *i_private; /* fs or device private pointer */
798 * inode->i_mutex nesting subclasses for the lock validator:
800 * 0: the object of the current VFS operation
805 * The locking order between these classes is
806 * parent -> child -> normal -> xattr -> quota
808 enum inode_i_mutex_lock_class
818 * NOTE: in a 32bit arch with a preemptable kernel and
819 * an UP compile the i_size_read/write must be atomic
820 * with respect to the local cpu (unlike with preempt disabled),
821 * but they don't need to be atomic with respect to other cpus like in
822 * true SMP (so they need either to either locally disable irq around
823 * the read or for example on x86 they can be still implemented as a
824 * cmpxchg8b without the need of the lock prefix). For SMP compiles
825 * and 64bit archs it makes no difference if preempt is enabled or not.
827 static inline loff_t i_size_read(const struct inode *inode)
829 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
834 seq = read_seqcount_begin(&inode->i_size_seqcount);
835 i_size = inode->i_size;
836 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
838 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
842 i_size = inode->i_size;
846 return inode->i_size;
851 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
852 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
853 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
855 static inline void i_size_write(struct inode *inode, loff_t i_size)
857 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
858 write_seqcount_begin(&inode->i_size_seqcount);
859 inode->i_size = i_size;
860 write_seqcount_end(&inode->i_size_seqcount);
861 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
863 inode->i_size = i_size;
866 inode->i_size = i_size;
870 static inline unsigned iminor(const struct inode *inode)
872 return MINOR(inode->i_rdev);
875 static inline unsigned imajor(const struct inode *inode)
877 return MAJOR(inode->i_rdev);
880 extern struct block_device *I_BDEV(struct inode *inode);
883 rwlock_t lock; /* protects pid, uid, euid fields */
884 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
885 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
886 uid_t uid, euid; /* uid/euid of process setting the owner */
887 int signum; /* posix.1b rt signal to be delivered on IO */
891 * Track a single file's readahead state
893 struct file_ra_state {
894 pgoff_t start; /* where readahead started */
895 unsigned int size; /* # of readahead pages */
896 unsigned int async_size; /* do asynchronous readahead when
897 there are only # of pages ahead */
899 unsigned int ra_pages; /* Maximum readahead window */
900 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
901 loff_t prev_pos; /* Cache last read() position */
905 * Check if @index falls in the readahead windows.
907 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
909 return (index >= ra->start &&
910 index < ra->start + ra->size);
913 #define FILE_MNT_WRITE_TAKEN 1
914 #define FILE_MNT_WRITE_RELEASED 2
918 * fu_list becomes invalid after file_free is called and queued via
919 * fu_rcuhead for RCU freeing
922 struct list_head fu_list;
923 struct rcu_head fu_rcuhead;
926 #define f_dentry f_path.dentry
927 #define f_vfsmnt f_path.mnt
928 const struct file_operations *f_op;
929 spinlock_t f_lock; /* f_ep_links, f_flags, no IRQ */
933 atomic_long_t f_count;
934 unsigned int f_flags;
937 struct fown_struct f_owner;
938 const struct cred *f_cred;
939 struct file_ra_state f_ra;
942 #ifdef CONFIG_SECURITY
945 /* needed for tty driver, and maybe others */
949 /* Used by fs/eventpoll.c to link all the hooks to this file */
950 struct list_head f_ep_links;
951 #endif /* #ifdef CONFIG_EPOLL */
952 struct address_space *f_mapping;
953 #ifdef CONFIG_DEBUG_WRITECOUNT
954 unsigned long f_mnt_write_state;
958 #define get_file(x) atomic_long_inc(&(x)->f_count)
959 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
960 #define file_count(x) atomic_long_read(&(x)->f_count)
962 #ifdef CONFIG_DEBUG_WRITECOUNT
963 static inline void file_take_write(struct file *f)
965 WARN_ON(f->f_mnt_write_state != 0);
966 f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN;
968 static inline void file_release_write(struct file *f)
970 f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED;
972 static inline void file_reset_write(struct file *f)
974 f->f_mnt_write_state = 0;
976 static inline void file_check_state(struct file *f)
979 * At this point, either both or neither of these bits
982 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN);
983 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED);
985 static inline int file_check_writeable(struct file *f)
987 if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN)
989 printk(KERN_WARNING "writeable file with no "
990 "mnt_want_write()\n");
994 #else /* !CONFIG_DEBUG_WRITECOUNT */
995 static inline void file_take_write(struct file *filp) {}
996 static inline void file_release_write(struct file *filp) {}
997 static inline void file_reset_write(struct file *filp) {}
998 static inline void file_check_state(struct file *filp) {}
999 static inline int file_check_writeable(struct file *filp)
1003 #endif /* CONFIG_DEBUG_WRITECOUNT */
1005 #define MAX_NON_LFS ((1UL<<31) - 1)
1007 /* Page cache limit. The filesystems should put that into their s_maxbytes
1008 limits, otherwise bad things can happen in VM. */
1009 #if BITS_PER_LONG==32
1010 #define MAX_LFS_FILESIZE (((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
1011 #elif BITS_PER_LONG==64
1012 #define MAX_LFS_FILESIZE 0x7fffffffffffffffUL
1017 #define FL_ACCESS 8 /* not trying to lock, just looking */
1018 #define FL_EXISTS 16 /* when unlocking, test for existence */
1019 #define FL_LEASE 32 /* lease held on this file */
1020 #define FL_CLOSE 64 /* unlock on close */
1021 #define FL_SLEEP 128 /* A blocking lock */
1024 * Special return value from posix_lock_file() and vfs_lock_file() for
1025 * asynchronous locking.
1027 #define FILE_LOCK_DEFERRED 1
1030 * The POSIX file lock owner is determined by
1031 * the "struct files_struct" in the thread group
1032 * (or NULL for no owner - BSD locks).
1034 * Lockd stuffs a "host" pointer into this.
1036 typedef struct files_struct *fl_owner_t;
1038 struct file_lock_operations {
1039 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1040 void (*fl_release_private)(struct file_lock *);
1043 struct lock_manager_operations {
1044 int (*fl_compare_owner)(struct file_lock *, struct file_lock *);
1045 void (*fl_notify)(struct file_lock *); /* unblock callback */
1046 int (*fl_grant)(struct file_lock *, struct file_lock *, int);
1047 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1048 void (*fl_release_private)(struct file_lock *);
1049 void (*fl_break)(struct file_lock *);
1050 int (*fl_mylease)(struct file_lock *, struct file_lock *);
1051 int (*fl_change)(struct file_lock **, int);
1054 struct lock_manager {
1055 struct list_head list;
1058 void locks_start_grace(struct lock_manager *);
1059 void locks_end_grace(struct lock_manager *);
1060 int locks_in_grace(void);
1062 /* that will die - we need it for nfs_lock_info */
1063 #include <linux/nfs_fs_i.h>
1066 struct file_lock *fl_next; /* singly linked list for this inode */
1067 struct list_head fl_link; /* doubly linked list of all locks */
1068 struct list_head fl_block; /* circular list of blocked processes */
1069 fl_owner_t fl_owner;
1070 unsigned char fl_flags;
1071 unsigned char fl_type;
1072 unsigned int fl_pid;
1073 struct pid *fl_nspid;
1074 wait_queue_head_t fl_wait;
1075 struct file *fl_file;
1079 struct fasync_struct * fl_fasync; /* for lease break notifications */
1080 unsigned long fl_break_time; /* for nonblocking lease breaks */
1082 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1083 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1085 struct nfs_lock_info nfs_fl;
1086 struct nfs4_lock_info nfs4_fl;
1088 struct list_head link; /* link in AFS vnode's pending_locks list */
1089 int state; /* state of grant or error if -ve */
1094 /* The following constant reflects the upper bound of the file/locking space */
1096 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1097 #define OFFSET_MAX INT_LIMIT(loff_t)
1098 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1101 #include <linux/fcntl.h>
1103 /* temporary stubs for BKL removal */
1104 #define lock_flocks() lock_kernel()
1105 #define unlock_flocks() unlock_kernel()
1107 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1109 #ifdef CONFIG_FILE_LOCKING
1110 extern int fcntl_getlk(struct file *, struct flock __user *);
1111 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1112 struct flock __user *);
1114 #if BITS_PER_LONG == 32
1115 extern int fcntl_getlk64(struct file *, struct flock64 __user *);
1116 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1117 struct flock64 __user *);
1120 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1121 extern int fcntl_getlease(struct file *filp);
1124 extern void locks_init_lock(struct file_lock *);
1125 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1126 extern void __locks_copy_lock(struct file_lock *, const struct file_lock *);
1127 extern void locks_remove_posix(struct file *, fl_owner_t);
1128 extern void locks_remove_flock(struct file *);
1129 extern void locks_release_private(struct file_lock *);
1130 extern void posix_test_lock(struct file *, struct file_lock *);
1131 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1132 extern int posix_lock_file_wait(struct file *, struct file_lock *);
1133 extern int posix_unblock_lock(struct file *, struct file_lock *);
1134 extern int vfs_test_lock(struct file *, struct file_lock *);
1135 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1136 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1137 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
1138 extern int __break_lease(struct inode *inode, unsigned int flags);
1139 extern void lease_get_mtime(struct inode *, struct timespec *time);
1140 extern int generic_setlease(struct file *, long, struct file_lock **);
1141 extern int vfs_setlease(struct file *, long, struct file_lock **);
1142 extern int lease_modify(struct file_lock **, int);
1143 extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
1144 extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
1145 #else /* !CONFIG_FILE_LOCKING */
1146 static inline int fcntl_getlk(struct file *file, struct flock __user *user)
1151 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1152 unsigned int cmd, struct flock __user *user)
1157 #if BITS_PER_LONG == 32
1158 static inline int fcntl_getlk64(struct file *file, struct flock64 __user *user)
1163 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1164 unsigned int cmd, struct flock64 __user *user)
1169 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1174 static inline int fcntl_getlease(struct file *filp)
1179 static inline void locks_init_lock(struct file_lock *fl)
1184 static inline void __locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1189 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1194 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1199 static inline void locks_remove_flock(struct file *filp)
1204 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1209 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1210 struct file_lock *conflock)
1215 static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1220 static inline int posix_unblock_lock(struct file *filp,
1221 struct file_lock *waiter)
1226 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1231 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1232 struct file_lock *fl, struct file_lock *conf)
1237 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1242 static inline int flock_lock_file_wait(struct file *filp,
1243 struct file_lock *request)
1248 static inline int __break_lease(struct inode *inode, unsigned int mode)
1253 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1258 static inline int generic_setlease(struct file *filp, long arg,
1259 struct file_lock **flp)
1264 static inline int vfs_setlease(struct file *filp, long arg,
1265 struct file_lock **lease)
1270 static inline int lease_modify(struct file_lock **before, int arg)
1275 static inline int lock_may_read(struct inode *inode, loff_t start,
1281 static inline int lock_may_write(struct inode *inode, loff_t start,
1287 #endif /* !CONFIG_FILE_LOCKING */
1290 struct fasync_struct {
1294 struct fasync_struct *fa_next; /* singly linked list */
1295 struct file *fa_file;
1296 struct rcu_head fa_rcu;
1299 #define FASYNC_MAGIC 0x4601
1301 /* SMP safe fasync helpers: */
1302 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1303 /* can be called from interrupts */
1304 extern void kill_fasync(struct fasync_struct **, int, int);
1306 extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1307 extern int f_setown(struct file *filp, unsigned long arg, int force);
1308 extern void f_delown(struct file *filp);
1309 extern pid_t f_getown(struct file *filp);
1310 extern int send_sigurg(struct fown_struct *fown);
1316 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1317 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1318 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1319 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1320 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1322 extern struct list_head super_blocks;
1323 extern spinlock_t sb_lock;
1325 struct super_block {
1326 struct list_head s_list; /* Keep this first */
1327 dev_t s_dev; /* search index; _not_ kdev_t */
1328 unsigned char s_dirt;
1329 unsigned char s_blocksize_bits;
1330 unsigned long s_blocksize;
1331 loff_t s_maxbytes; /* Max file size */
1332 struct file_system_type *s_type;
1333 const struct super_operations *s_op;
1334 const struct dquot_operations *dq_op;
1335 const struct quotactl_ops *s_qcop;
1336 const struct export_operations *s_export_op;
1337 unsigned long s_flags;
1338 unsigned long s_magic;
1339 struct dentry *s_root;
1340 struct rw_semaphore s_umount;
1341 struct mutex s_lock;
1344 #ifdef CONFIG_SECURITY
1347 const struct xattr_handler **s_xattr;
1349 struct list_head s_inodes; /* all inodes */
1350 struct hlist_head s_anon; /* anonymous dentries for (nfs) exporting */
1352 struct list_head __percpu *s_files;
1354 struct list_head s_files;
1356 /* s_dentry_lru and s_nr_dentry_unused are protected by dcache_lock */
1357 struct list_head s_dentry_lru; /* unused dentry lru */
1358 int s_nr_dentry_unused; /* # of dentry on lru */
1360 struct block_device *s_bdev;
1361 struct backing_dev_info *s_bdi;
1362 struct mtd_info *s_mtd;
1363 struct list_head s_instances;
1364 struct quota_info s_dquot; /* Diskquota specific options */
1367 wait_queue_head_t s_wait_unfrozen;
1369 char s_id[32]; /* Informational name */
1371 void *s_fs_info; /* Filesystem private info */
1374 /* Granularity of c/m/atime in ns.
1375 Cannot be worse than a second */
1379 * The next field is for VFS *only*. No filesystems have any business
1380 * even looking at it. You had been warned.
1382 struct mutex s_vfs_rename_mutex; /* Kludge */
1385 * Filesystem subtype. If non-empty the filesystem type field
1386 * in /proc/mounts will be "type.subtype"
1391 * Saved mount options for lazy filesystems using
1392 * generic_show_options()
1397 extern struct timespec current_fs_time(struct super_block *sb);
1400 * Snapshotting support.
1404 SB_FREEZE_WRITE = 1,
1405 SB_FREEZE_TRANS = 2,
1408 #define vfs_check_frozen(sb, level) \
1409 wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
1411 #define get_fs_excl() atomic_inc(¤t->fs_excl)
1412 #define put_fs_excl() atomic_dec(¤t->fs_excl)
1413 #define has_fs_excl() atomic_read(¤t->fs_excl)
1415 #define is_owner_or_cap(inode) \
1416 ((current_fsuid() == (inode)->i_uid) || capable(CAP_FOWNER))
1418 /* not quite ready to be deprecated, but... */
1419 extern void lock_super(struct super_block *);
1420 extern void unlock_super(struct super_block *);
1423 * VFS helper functions..
1425 extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *);
1426 extern int vfs_mkdir(struct inode *, struct dentry *, int);
1427 extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t);
1428 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1429 extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
1430 extern int vfs_rmdir(struct inode *, struct dentry *);
1431 extern int vfs_unlink(struct inode *, struct dentry *);
1432 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1435 * VFS dentry helper functions.
1437 extern void dentry_unhash(struct dentry *dentry);
1440 * VFS file helper functions.
1442 extern int file_permission(struct file *, int);
1443 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1446 * VFS FS_IOC_FIEMAP helper definitions.
1448 struct fiemap_extent_info {
1449 unsigned int fi_flags; /* Flags as passed from user */
1450 unsigned int fi_extents_mapped; /* Number of mapped extents */
1451 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1452 struct fiemap_extent *fi_extents_start; /* Start of fiemap_extent
1455 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1456 u64 phys, u64 len, u32 flags);
1457 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1462 * NOTE! These match bits 12..15 of stat.st_mode
1463 * (ie "(i_mode >> 12) & 15").
1465 #define DT_UNKNOWN 0
1476 * This is the "filldir" function type, used by readdir() to let
1477 * the kernel specify what kind of dirent layout it wants to have.
1478 * This allows the kernel to read directories into kernel space or
1479 * to have different dirent layouts depending on the binary type.
1481 typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
1482 struct block_device_operations;
1484 /* These macros are for out of kernel modules to test that
1485 * the kernel supports the unlocked_ioctl and compat_ioctl
1486 * fields in struct file_operations. */
1487 #define HAVE_COMPAT_IOCTL 1
1488 #define HAVE_UNLOCKED_IOCTL 1
1492 * all file operations except setlease can be called without
1493 * the big kernel lock held in all filesystems.
1495 struct file_operations {
1496 struct module *owner;
1497 loff_t (*llseek) (struct file *, loff_t, int);
1498 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1499 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1500 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1501 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1502 int (*readdir) (struct file *, void *, filldir_t);
1503 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1504 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1505 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1506 int (*mmap) (struct file *, struct vm_area_struct *);
1507 int (*open) (struct inode *, struct file *);
1508 int (*flush) (struct file *, fl_owner_t id);
1509 int (*release) (struct inode *, struct file *);
1510 int (*fsync) (struct file *, int datasync);
1511 int (*aio_fsync) (struct kiocb *, int datasync);
1512 int (*fasync) (int, struct file *, int);
1513 int (*lock) (struct file *, int, struct file_lock *);
1514 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1515 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1516 int (*check_flags)(int);
1517 int (*flock) (struct file *, int, struct file_lock *);
1518 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1519 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1520 int (*setlease)(struct file *, long, struct file_lock **);
1523 struct inode_operations {
1524 int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
1525 struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
1526 int (*link) (struct dentry *,struct inode *,struct dentry *);
1527 int (*unlink) (struct inode *,struct dentry *);
1528 int (*symlink) (struct inode *,struct dentry *,const char *);
1529 int (*mkdir) (struct inode *,struct dentry *,int);
1530 int (*rmdir) (struct inode *,struct dentry *);
1531 int (*mknod) (struct inode *,struct dentry *,int,dev_t);
1532 int (*rename) (struct inode *, struct dentry *,
1533 struct inode *, struct dentry *);
1534 int (*readlink) (struct dentry *, char __user *,int);
1535 void * (*follow_link) (struct dentry *, struct nameidata *);
1536 void (*put_link) (struct dentry *, struct nameidata *, void *);
1537 void (*truncate) (struct inode *);
1538 int (*permission) (struct inode *, int);
1539 int (*check_acl)(struct inode *, int);
1540 int (*setattr) (struct dentry *, struct iattr *);
1541 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1542 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1543 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1544 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1545 int (*removexattr) (struct dentry *, const char *);
1546 void (*truncate_range)(struct inode *, loff_t, loff_t);
1547 long (*fallocate)(struct inode *inode, int mode, loff_t offset,
1549 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1555 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1556 unsigned long nr_segs, unsigned long fast_segs,
1557 struct iovec *fast_pointer,
1558 struct iovec **ret_pointer);
1560 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1561 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1562 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1563 unsigned long, loff_t *);
1564 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1565 unsigned long, loff_t *);
1567 struct super_operations {
1568 struct inode *(*alloc_inode)(struct super_block *sb);
1569 void (*destroy_inode)(struct inode *);
1571 void (*dirty_inode) (struct inode *);
1572 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1573 int (*drop_inode) (struct inode *);
1574 void (*evict_inode) (struct inode *);
1575 void (*put_super) (struct super_block *);
1576 void (*write_super) (struct super_block *);
1577 int (*sync_fs)(struct super_block *sb, int wait);
1578 int (*freeze_fs) (struct super_block *);
1579 int (*unfreeze_fs) (struct super_block *);
1580 int (*statfs) (struct dentry *, struct kstatfs *);
1581 int (*remount_fs) (struct super_block *, int *, char *);
1582 void (*umount_begin) (struct super_block *);
1584 int (*show_options)(struct seq_file *, struct vfsmount *);
1585 int (*show_stats)(struct seq_file *, struct vfsmount *);
1587 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1588 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1590 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1591 int (*trim_fs) (struct super_block *, struct fstrim_range *);
1595 * Inode state bits. Protected by inode_lock.
1597 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1598 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1600 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1601 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1602 * various stages of removing an inode.
1604 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1606 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1607 * fdatasync(). i_atime is the usual cause.
1608 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1609 * these changes separately from I_DIRTY_SYNC so that we
1610 * don't have to write inode on fdatasync() when only
1611 * mtime has changed in it.
1612 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1613 * I_NEW Serves as both a mutex and completion notification.
1614 * New inodes set I_NEW. If two processes both create
1615 * the same inode, one of them will release its inode and
1616 * wait for I_NEW to be released before returning.
1617 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1618 * also cause waiting on I_NEW, without I_NEW actually
1619 * being set. find_inode() uses this to prevent returning
1620 * nearly-dead inodes.
1621 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1622 * is zero. I_FREEING must be set when I_WILL_FREE is
1624 * I_FREEING Set when inode is about to be freed but still has dirty
1625 * pages or buffers attached or the inode itself is still
1627 * I_CLEAR Added by end_writeback(). In this state the inode is clean
1628 * and can be destroyed. Inode keeps I_FREEING.
1630 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1631 * prohibited for many purposes. iget() must wait for
1632 * the inode to be completely released, then create it
1633 * anew. Other functions will just ignore such inodes,
1634 * if appropriate. I_NEW is used for waiting.
1636 * I_SYNC Synchonized write of dirty inode data. The bits is
1637 * set during data writeback, and cleared with a wakeup
1638 * on the bit address once it is done.
1640 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1642 #define I_DIRTY_SYNC 1
1643 #define I_DIRTY_DATASYNC 2
1644 #define I_DIRTY_PAGES 4
1646 #define I_NEW (1 << __I_NEW)
1647 #define I_WILL_FREE 16
1648 #define I_FREEING 32
1651 #define I_SYNC (1 << __I_SYNC)
1653 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1655 extern void __mark_inode_dirty(struct inode *, int);
1656 static inline void mark_inode_dirty(struct inode *inode)
1658 __mark_inode_dirty(inode, I_DIRTY);
1661 static inline void mark_inode_dirty_sync(struct inode *inode)
1663 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1667 * inc_nlink - directly increment an inode's link count
1670 * This is a low-level filesystem helper to replace any
1671 * direct filesystem manipulation of i_nlink. Currently,
1672 * it is only here for parity with dec_nlink().
1674 static inline void inc_nlink(struct inode *inode)
1679 static inline void inode_inc_link_count(struct inode *inode)
1682 mark_inode_dirty(inode);
1686 * drop_nlink - directly drop an inode's link count
1689 * This is a low-level filesystem helper to replace any
1690 * direct filesystem manipulation of i_nlink. In cases
1691 * where we are attempting to track writes to the
1692 * filesystem, a decrement to zero means an imminent
1693 * write when the file is truncated and actually unlinked
1694 * on the filesystem.
1696 static inline void drop_nlink(struct inode *inode)
1702 * clear_nlink - directly zero an inode's link count
1705 * This is a low-level filesystem helper to replace any
1706 * direct filesystem manipulation of i_nlink. See
1707 * drop_nlink() for why we care about i_nlink hitting zero.
1709 static inline void clear_nlink(struct inode *inode)
1714 static inline void inode_dec_link_count(struct inode *inode)
1717 mark_inode_dirty(inode);
1721 * inode_inc_iversion - increments i_version
1722 * @inode: inode that need to be updated
1724 * Every time the inode is modified, the i_version field will be incremented.
1725 * The filesystem has to be mounted with i_version flag
1728 static inline void inode_inc_iversion(struct inode *inode)
1730 spin_lock(&inode->i_lock);
1732 spin_unlock(&inode->i_lock);
1735 extern void touch_atime(struct vfsmount *mnt, struct dentry *dentry);
1736 static inline void file_accessed(struct file *file)
1738 if (!(file->f_flags & O_NOATIME))
1739 touch_atime(file->f_path.mnt, file->f_path.dentry);
1742 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1744 struct file_system_type {
1747 int (*get_sb) (struct file_system_type *, int,
1748 const char *, void *, struct vfsmount *);
1749 void (*kill_sb) (struct super_block *);
1750 struct module *owner;
1751 struct file_system_type * next;
1752 struct list_head fs_supers;
1754 struct lock_class_key s_lock_key;
1755 struct lock_class_key s_umount_key;
1756 struct lock_class_key s_vfs_rename_key;
1758 struct lock_class_key i_lock_key;
1759 struct lock_class_key i_mutex_key;
1760 struct lock_class_key i_mutex_dir_key;
1761 struct lock_class_key i_alloc_sem_key;
1764 extern int get_sb_ns(struct file_system_type *fs_type, int flags, void *data,
1765 int (*fill_super)(struct super_block *, void *, int),
1766 struct vfsmount *mnt);
1767 extern int get_sb_bdev(struct file_system_type *fs_type,
1768 int flags, const char *dev_name, void *data,
1769 int (*fill_super)(struct super_block *, void *, int),
1770 struct vfsmount *mnt);
1771 extern int get_sb_single(struct file_system_type *fs_type,
1772 int flags, void *data,
1773 int (*fill_super)(struct super_block *, void *, int),
1774 struct vfsmount *mnt);
1775 extern int get_sb_nodev(struct file_system_type *fs_type,
1776 int flags, void *data,
1777 int (*fill_super)(struct super_block *, void *, int),
1778 struct vfsmount *mnt);
1779 void generic_shutdown_super(struct super_block *sb);
1780 void kill_block_super(struct super_block *sb);
1781 void kill_anon_super(struct super_block *sb);
1782 void kill_litter_super(struct super_block *sb);
1783 void deactivate_super(struct super_block *sb);
1784 void deactivate_locked_super(struct super_block *sb);
1785 int set_anon_super(struct super_block *s, void *data);
1786 struct super_block *sget(struct file_system_type *type,
1787 int (*test)(struct super_block *,void *),
1788 int (*set)(struct super_block *,void *),
1790 extern int get_sb_pseudo(struct file_system_type *, char *,
1791 const struct super_operations *ops, unsigned long,
1792 struct vfsmount *mnt);
1793 extern void simple_set_mnt(struct vfsmount *mnt, struct super_block *sb);
1795 static inline void sb_mark_dirty(struct super_block *sb)
1799 static inline void sb_mark_clean(struct super_block *sb)
1803 static inline int sb_is_dirty(struct super_block *sb)
1808 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1809 #define fops_get(fops) \
1810 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1811 #define fops_put(fops) \
1812 do { if (fops) module_put((fops)->owner); } while(0)
1814 extern int register_filesystem(struct file_system_type *);
1815 extern int unregister_filesystem(struct file_system_type *);
1816 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1817 #define kern_mount(type) kern_mount_data(type, NULL)
1818 extern int may_umount_tree(struct vfsmount *);
1819 extern int may_umount(struct vfsmount *);
1820 extern long do_mount(char *, char *, char *, unsigned long, void *);
1821 extern struct vfsmount *collect_mounts(struct path *);
1822 extern void drop_collected_mounts(struct vfsmount *);
1823 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1825 extern int vfs_statfs(struct path *, struct kstatfs *);
1826 extern int statfs_by_dentry(struct dentry *, struct kstatfs *);
1827 extern int freeze_super(struct super_block *super);
1828 extern int thaw_super(struct super_block *super);
1830 extern int current_umask(void);
1833 extern struct kobject *fs_kobj;
1835 extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1837 #define FLOCK_VERIFY_READ 1
1838 #define FLOCK_VERIFY_WRITE 2
1840 #ifdef CONFIG_FILE_LOCKING
1841 extern int locks_mandatory_locked(struct inode *);
1842 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1845 * Candidates for mandatory locking have the setgid bit set
1846 * but no group execute bit - an otherwise meaningless combination.
1849 static inline int __mandatory_lock(struct inode *ino)
1851 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1855 * ... and these candidates should be on MS_MANDLOCK mounted fs,
1856 * otherwise these will be advisory locks
1859 static inline int mandatory_lock(struct inode *ino)
1861 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1864 static inline int locks_verify_locked(struct inode *inode)
1866 if (mandatory_lock(inode))
1867 return locks_mandatory_locked(inode);
1871 static inline int locks_verify_truncate(struct inode *inode,
1875 if (inode->i_flock && mandatory_lock(inode))
1876 return locks_mandatory_area(
1877 FLOCK_VERIFY_WRITE, inode, filp,
1878 size < inode->i_size ? size : inode->i_size,
1879 (size < inode->i_size ? inode->i_size - size
1880 : size - inode->i_size)
1885 static inline int break_lease(struct inode *inode, unsigned int mode)
1888 return __break_lease(inode, mode);
1891 #else /* !CONFIG_FILE_LOCKING */
1892 static inline int locks_mandatory_locked(struct inode *inode)
1897 static inline int locks_mandatory_area(int rw, struct inode *inode,
1898 struct file *filp, loff_t offset,
1904 static inline int __mandatory_lock(struct inode *inode)
1909 static inline int mandatory_lock(struct inode *inode)
1914 static inline int locks_verify_locked(struct inode *inode)
1919 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
1925 static inline int break_lease(struct inode *inode, unsigned int mode)
1930 #endif /* CONFIG_FILE_LOCKING */
1934 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
1936 extern int do_fallocate(struct file *file, int mode, loff_t offset,
1938 extern long do_sys_open(int dfd, const char __user *filename, int flags,
1940 extern struct file *filp_open(const char *, int, int);
1941 extern struct file * dentry_open(struct dentry *, struct vfsmount *, int,
1942 const struct cred *);
1943 extern int filp_close(struct file *, fl_owner_t id);
1944 extern char * getname(const char __user *);
1948 extern int ioctl_preallocate(struct file *filp, void __user *argp);
1951 extern void __init vfs_caches_init_early(void);
1952 extern void __init vfs_caches_init(unsigned long);
1954 extern struct kmem_cache *names_cachep;
1956 #define __getname_gfp(gfp) kmem_cache_alloc(names_cachep, (gfp))
1957 #define __getname() __getname_gfp(GFP_KERNEL)
1958 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
1959 #ifndef CONFIG_AUDITSYSCALL
1960 #define putname(name) __putname(name)
1962 extern void putname(const char *name);
1966 extern int register_blkdev(unsigned int, const char *);
1967 extern void unregister_blkdev(unsigned int, const char *);
1968 extern struct block_device *bdget(dev_t);
1969 extern struct block_device *bdgrab(struct block_device *bdev);
1970 extern void bd_set_size(struct block_device *, loff_t size);
1971 extern void bd_forget(struct inode *inode);
1972 extern void bdput(struct block_device *);
1973 extern struct block_device *open_by_devnum(dev_t, fmode_t);
1974 extern void invalidate_bdev(struct block_device *);
1975 extern int sync_blockdev(struct block_device *bdev);
1976 extern struct super_block *freeze_bdev(struct block_device *);
1977 extern void emergency_thaw_all(void);
1978 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
1979 extern int fsync_bdev(struct block_device *);
1981 static inline void bd_forget(struct inode *inode) {}
1982 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
1983 static inline void invalidate_bdev(struct block_device *bdev) {}
1985 static inline struct super_block *freeze_bdev(struct block_device *sb)
1990 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
1995 extern int sync_filesystem(struct super_block *);
1996 extern const struct file_operations def_blk_fops;
1997 extern const struct file_operations def_chr_fops;
1998 extern const struct file_operations bad_sock_fops;
1999 extern const struct file_operations def_fifo_fops;
2001 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2002 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2003 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2004 extern int blkdev_get(struct block_device *, fmode_t);
2005 extern int blkdev_put(struct block_device *, fmode_t);
2006 extern int bd_claim(struct block_device *, void *);
2007 extern void bd_release(struct block_device *);
2009 extern int bd_claim_by_disk(struct block_device *, void *, struct gendisk *);
2010 extern void bd_release_from_disk(struct block_device *, struct gendisk *);
2012 #define bd_claim_by_disk(bdev, holder, disk) bd_claim(bdev, holder)
2013 #define bd_release_from_disk(bdev, disk) bd_release(bdev)
2018 #define CHRDEV_MAJOR_HASH_SIZE 255
2019 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2020 extern int register_chrdev_region(dev_t, unsigned, const char *);
2021 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2022 unsigned int count, const char *name,
2023 const struct file_operations *fops);
2024 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2025 unsigned int count, const char *name);
2026 extern void unregister_chrdev_region(dev_t, unsigned);
2027 extern void chrdev_show(struct seq_file *,off_t);
2029 static inline int register_chrdev(unsigned int major, const char *name,
2030 const struct file_operations *fops)
2032 return __register_chrdev(major, 0, 256, name, fops);
2035 static inline void unregister_chrdev(unsigned int major, const char *name)
2037 __unregister_chrdev(major, 0, 256, name);
2040 /* fs/block_dev.c */
2041 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2042 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2045 #define BLKDEV_MAJOR_HASH_SIZE 255
2046 extern const char *__bdevname(dev_t, char *buffer);
2047 extern const char *bdevname(struct block_device *bdev, char *buffer);
2048 extern struct block_device *lookup_bdev(const char *);
2049 extern struct block_device *open_bdev_exclusive(const char *, fmode_t, void *);
2050 extern void close_bdev_exclusive(struct block_device *, fmode_t);
2051 extern void blkdev_show(struct seq_file *,off_t);
2054 #define BLKDEV_MAJOR_HASH_SIZE 0
2057 extern void init_special_inode(struct inode *, umode_t, dev_t);
2059 /* Invalid inode operations -- fs/bad_inode.c */
2060 extern void make_bad_inode(struct inode *);
2061 extern int is_bad_inode(struct inode *);
2063 extern const struct file_operations read_pipefifo_fops;
2064 extern const struct file_operations write_pipefifo_fops;
2065 extern const struct file_operations rdwr_pipefifo_fops;
2067 extern int fs_may_remount_ro(struct super_block *);
2071 * return READ, READA, or WRITE
2073 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2076 * return data direction, READ or WRITE
2078 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2080 extern void check_disk_size_change(struct gendisk *disk,
2081 struct block_device *bdev);
2082 extern int revalidate_disk(struct gendisk *);
2083 extern int check_disk_change(struct block_device *);
2084 extern int __invalidate_device(struct block_device *);
2085 extern int invalidate_partition(struct gendisk *, int);
2087 extern int invalidate_inodes(struct super_block *);
2088 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2089 pgoff_t start, pgoff_t end);
2091 static inline void invalidate_remote_inode(struct inode *inode)
2093 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2094 S_ISLNK(inode->i_mode))
2095 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2097 extern int invalidate_inode_pages2(struct address_space *mapping);
2098 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2099 pgoff_t start, pgoff_t end);
2100 extern int write_inode_now(struct inode *, int);
2101 extern int filemap_fdatawrite(struct address_space *);
2102 extern int filemap_flush(struct address_space *);
2103 extern int filemap_fdatawait(struct address_space *);
2104 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2106 extern int filemap_write_and_wait(struct address_space *mapping);
2107 extern int filemap_write_and_wait_range(struct address_space *mapping,
2108 loff_t lstart, loff_t lend);
2109 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2110 loff_t start, loff_t end, int sync_mode);
2111 extern int filemap_fdatawrite_range(struct address_space *mapping,
2112 loff_t start, loff_t end);
2114 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2116 extern int vfs_fsync(struct file *file, int datasync);
2117 extern int generic_write_sync(struct file *file, loff_t pos, loff_t count);
2118 extern void sync_supers(void);
2119 extern void emergency_sync(void);
2120 extern void emergency_remount(void);
2122 extern sector_t bmap(struct inode *, sector_t);
2124 extern int notify_change(struct dentry *, struct iattr *);
2125 extern int inode_permission(struct inode *, int);
2126 extern int generic_permission(struct inode *, int,
2127 int (*check_acl)(struct inode *, int));
2129 static inline bool execute_ok(struct inode *inode)
2131 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2134 extern int get_write_access(struct inode *);
2135 extern int deny_write_access(struct file *);
2136 static inline void put_write_access(struct inode * inode)
2138 atomic_dec(&inode->i_writecount);
2140 static inline void allow_write_access(struct file *file)
2143 atomic_inc(&file->f_path.dentry->d_inode->i_writecount);
2145 extern int do_pipe_flags(int *, int);
2146 extern struct file *create_read_pipe(struct file *f, int flags);
2147 extern struct file *create_write_pipe(int flags);
2148 extern void free_write_pipe(struct file *);
2150 extern struct file *do_filp_open(int dfd, const char *pathname,
2151 int open_flag, int mode, int acc_mode);
2152 extern int may_open(struct path *, int, int);
2154 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2155 extern struct file * open_exec(const char *);
2157 /* fs/dcache.c -- generic fs support functions */
2158 extern int is_subdir(struct dentry *, struct dentry *);
2159 extern int path_is_under(struct path *, struct path *);
2160 extern ino_t find_inode_number(struct dentry *, struct qstr *);
2162 #include <linux/err.h>
2164 /* needed for stackable file system support */
2165 extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
2167 extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
2169 extern int inode_init_always(struct super_block *, struct inode *);
2170 extern void inode_init_once(struct inode *);
2171 extern void inode_add_to_lists(struct super_block *, struct inode *);
2172 extern void iput(struct inode *);
2173 extern struct inode * igrab(struct inode *);
2174 extern ino_t iunique(struct super_block *, ino_t);
2175 extern int inode_needs_sync(struct inode *inode);
2176 extern int generic_delete_inode(struct inode *inode);
2177 extern int generic_drop_inode(struct inode *inode);
2179 extern struct inode *ilookup5_nowait(struct super_block *sb,
2180 unsigned long hashval, int (*test)(struct inode *, void *),
2182 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2183 int (*test)(struct inode *, void *), void *data);
2184 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2186 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2187 extern struct inode * iget_locked(struct super_block *, unsigned long);
2188 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2189 extern int insert_inode_locked(struct inode *);
2190 extern void unlock_new_inode(struct inode *);
2192 extern void __iget(struct inode * inode);
2193 extern void iget_failed(struct inode *);
2194 extern void end_writeback(struct inode *);
2195 extern void destroy_inode(struct inode *);
2196 extern void __destroy_inode(struct inode *);
2197 extern struct inode *new_inode(struct super_block *);
2198 extern int should_remove_suid(struct dentry *);
2199 extern int file_remove_suid(struct file *);
2201 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2202 extern void remove_inode_hash(struct inode *);
2203 static inline void insert_inode_hash(struct inode *inode) {
2204 __insert_inode_hash(inode, inode->i_ino);
2208 extern void submit_bio(int, struct bio *);
2209 extern int bdev_read_only(struct block_device *);
2211 extern int set_blocksize(struct block_device *, int);
2212 extern int sb_set_blocksize(struct super_block *, int);
2213 extern int sb_min_blocksize(struct super_block *, int);
2215 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2216 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2217 extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
2218 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
2219 extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2220 extern ssize_t __generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long,
2222 extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2223 extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
2224 unsigned long *, loff_t, loff_t *, size_t, size_t);
2225 extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
2226 unsigned long, loff_t, loff_t *, size_t, ssize_t);
2227 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2228 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2229 extern int generic_segment_checks(const struct iovec *iov,
2230 unsigned long *nr_segs, size_t *count, int access_flags);
2232 /* fs/block_dev.c */
2233 extern ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov,
2234 unsigned long nr_segs, loff_t pos);
2235 extern int blkdev_fsync(struct file *filp, int datasync);
2238 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2239 struct pipe_inode_info *, size_t, unsigned int);
2240 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2241 struct pipe_inode_info *, size_t, unsigned int);
2242 extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
2243 struct file *, loff_t *, size_t, unsigned int);
2244 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2245 struct file *out, loff_t *, size_t len, unsigned int flags);
2246 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2247 size_t len, unsigned int flags);
2250 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2251 extern loff_t noop_llseek(struct file *file, loff_t offset, int origin);
2252 extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
2253 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
2254 extern loff_t generic_file_llseek_unlocked(struct file *file, loff_t offset,
2256 extern int generic_file_open(struct inode * inode, struct file * filp);
2257 extern int nonseekable_open(struct inode * inode, struct file * filp);
2259 #ifdef CONFIG_FS_XIP
2260 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2262 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
2263 extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2264 size_t len, loff_t *ppos);
2265 extern int xip_truncate_page(struct address_space *mapping, loff_t from);
2267 static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2274 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2275 loff_t file_offset);
2278 /* need locking between buffered and direct access */
2281 /* filesystem does not support filling holes */
2282 DIO_SKIP_HOLES = 0x02,
2285 void dio_end_io(struct bio *bio, int error);
2287 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2288 struct block_device *bdev, const struct iovec *iov, loff_t offset,
2289 unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
2290 dio_submit_t submit_io, int flags);
2292 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
2293 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
2294 loff_t offset, unsigned long nr_segs, get_block_t get_block,
2295 dio_iodone_t end_io)
2297 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
2298 nr_segs, get_block, end_io, NULL,
2299 DIO_LOCKING | DIO_SKIP_HOLES);
2303 extern const struct file_operations generic_ro_fops;
2305 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2307 extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
2308 extern int vfs_follow_link(struct nameidata *, const char *);
2309 extern int page_readlink(struct dentry *, char __user *, int);
2310 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2311 extern void page_put_link(struct dentry *, struct nameidata *, void *);
2312 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2314 extern int page_symlink(struct inode *inode, const char *symname, int len);
2315 extern const struct inode_operations page_symlink_inode_operations;
2316 extern int generic_readlink(struct dentry *, char __user *, int);
2317 extern void generic_fillattr(struct inode *, struct kstat *);
2318 extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2319 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2320 void inode_add_bytes(struct inode *inode, loff_t bytes);
2321 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2322 loff_t inode_get_bytes(struct inode *inode);
2323 void inode_set_bytes(struct inode *inode, loff_t bytes);
2325 extern int vfs_readdir(struct file *, filldir_t, void *);
2327 extern int vfs_stat(const char __user *, struct kstat *);
2328 extern int vfs_lstat(const char __user *, struct kstat *);
2329 extern int vfs_fstat(unsigned int, struct kstat *);
2330 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2332 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2334 extern int __generic_block_fiemap(struct inode *inode,
2335 struct fiemap_extent_info *fieinfo,
2336 loff_t start, loff_t len,
2337 get_block_t *get_block);
2338 extern int generic_block_fiemap(struct inode *inode,
2339 struct fiemap_extent_info *fieinfo, u64 start,
2340 u64 len, get_block_t *get_block);
2342 extern void get_filesystem(struct file_system_type *fs);
2343 extern void put_filesystem(struct file_system_type *fs);
2344 extern struct file_system_type *get_fs_type(const char *name);
2345 extern struct super_block *get_super(struct block_device *);
2346 extern struct super_block *get_active_super(struct block_device *bdev);
2347 extern struct super_block *user_get_super(dev_t);
2348 extern void drop_super(struct super_block *sb);
2349 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2351 extern int dcache_dir_open(struct inode *, struct file *);
2352 extern int dcache_dir_close(struct inode *, struct file *);
2353 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2354 extern int dcache_readdir(struct file *, void *, filldir_t);
2355 extern int simple_setattr(struct dentry *, struct iattr *);
2356 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2357 extern int simple_statfs(struct dentry *, struct kstatfs *);
2358 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2359 extern int simple_unlink(struct inode *, struct dentry *);
2360 extern int simple_rmdir(struct inode *, struct dentry *);
2361 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2362 extern int noop_fsync(struct file *, int);
2363 extern int simple_empty(struct dentry *);
2364 extern int simple_readpage(struct file *file, struct page *page);
2365 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2366 loff_t pos, unsigned len, unsigned flags,
2367 struct page **pagep, void **fsdata);
2368 extern int simple_write_end(struct file *file, struct address_space *mapping,
2369 loff_t pos, unsigned len, unsigned copied,
2370 struct page *page, void *fsdata);
2372 extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
2373 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2374 extern const struct file_operations simple_dir_operations;
2375 extern const struct inode_operations simple_dir_inode_operations;
2376 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2377 struct dentry *d_alloc_name(struct dentry *, const char *);
2378 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2379 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2380 extern void simple_release_fs(struct vfsmount **mount, int *count);
2382 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2383 loff_t *ppos, const void *from, size_t available);
2384 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2385 const void __user *from, size_t count);
2387 extern int generic_file_fsync(struct file *, int);
2389 #ifdef CONFIG_MIGRATION
2390 extern int buffer_migrate_page(struct address_space *,
2391 struct page *, struct page *);
2393 #define buffer_migrate_page NULL
2396 extern int inode_change_ok(const struct inode *, struct iattr *);
2397 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2398 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2400 extern void file_update_time(struct file *file);
2402 extern int generic_show_options(struct seq_file *m, struct vfsmount *mnt);
2403 extern void save_mount_options(struct super_block *sb, char *options);
2404 extern void replace_mount_options(struct super_block *sb, char *options);
2406 static inline ino_t parent_ino(struct dentry *dentry)
2410 spin_lock(&dentry->d_lock);
2411 res = dentry->d_parent->d_inode->i_ino;
2412 spin_unlock(&dentry->d_lock);
2416 /* Transaction based IO helpers */
2419 * An argresp is stored in an allocated page and holds the
2420 * size of the argument or response, along with its content
2422 struct simple_transaction_argresp {
2427 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2429 char *simple_transaction_get(struct file *file, const char __user *buf,
2431 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2432 size_t size, loff_t *pos);
2433 int simple_transaction_release(struct inode *inode, struct file *file);
2435 void simple_transaction_set(struct file *file, size_t n);
2438 * simple attribute files
2440 * These attributes behave similar to those in sysfs:
2442 * Writing to an attribute immediately sets a value, an open file can be
2443 * written to multiple times.
2445 * Reading from an attribute creates a buffer from the value that might get
2446 * read with multiple read calls. When the attribute has been read
2447 * completely, no further read calls are possible until the file is opened
2450 * All attributes contain a text representation of a numeric value
2451 * that are accessed with the get() and set() functions.
2453 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2454 static int __fops ## _open(struct inode *inode, struct file *file) \
2456 __simple_attr_check_format(__fmt, 0ull); \
2457 return simple_attr_open(inode, file, __get, __set, __fmt); \
2459 static const struct file_operations __fops = { \
2460 .owner = THIS_MODULE, \
2461 .open = __fops ## _open, \
2462 .release = simple_attr_release, \
2463 .read = simple_attr_read, \
2464 .write = simple_attr_write, \
2467 static inline void __attribute__((format(printf, 1, 2)))
2468 __simple_attr_check_format(const char *fmt, ...)
2470 /* don't do anything, just let the compiler check the arguments; */
2473 int simple_attr_open(struct inode *inode, struct file *file,
2474 int (*get)(void *, u64 *), int (*set)(void *, u64),
2476 int simple_attr_release(struct inode *inode, struct file *file);
2477 ssize_t simple_attr_read(struct file *file, char __user *buf,
2478 size_t len, loff_t *ppos);
2479 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2480 size_t len, loff_t *ppos);
2483 int proc_nr_files(struct ctl_table *table, int write,
2484 void __user *buffer, size_t *lenp, loff_t *ppos);
2486 int __init get_filesystem_list(char *buf);
2488 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2489 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2490 (flag & FMODE_NONOTIFY)))
2492 #endif /* __KERNEL__ */
2493 #endif /* _LINUX_FS_H */