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>
12 #include <linux/types.h>
15 * It's silly to have NR_OPEN bigger than NR_FILE, but you can change
16 * the file limit at runtime and only root can increase the per-process
17 * nr_file rlimit, so it's safe to set up a ridiculously high absolute
18 * upper limit on files-per-process.
20 * Some programs (notably those using select()) may have to be
21 * recompiled to take full advantage of the new limits..
24 /* Fixed constants first: */
26 #define INR_OPEN 1024 /* Initial setting for nfile rlimits */
28 #define BLOCK_SIZE_BITS 10
29 #define BLOCK_SIZE (1<<BLOCK_SIZE_BITS)
31 #define SEEK_SET 0 /* seek relative to beginning of file */
32 #define SEEK_CUR 1 /* seek relative to current file position */
33 #define SEEK_END 2 /* seek relative to end of file */
34 #define SEEK_MAX SEEK_END
42 /* And dynamically-tunable limits and defaults: */
43 struct files_stat_struct {
44 unsigned long nr_files; /* read only */
45 unsigned long nr_free_files; /* read only */
46 unsigned long max_files; /* tunable */
49 struct inodes_stat_t {
52 int dummy[5]; /* padding for sysctl ABI compatibility */
56 #define NR_FILE 8192 /* this can well be larger on a larger system */
67 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
68 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
71 /* file is open for reading */
72 #define FMODE_READ ((__force fmode_t)0x1)
73 /* file is open for writing */
74 #define FMODE_WRITE ((__force fmode_t)0x2)
75 /* file is seekable */
76 #define FMODE_LSEEK ((__force fmode_t)0x4)
77 /* file can be accessed using pread */
78 #define FMODE_PREAD ((__force fmode_t)0x8)
79 /* file can be accessed using pwrite */
80 #define FMODE_PWRITE ((__force fmode_t)0x10)
81 /* File is opened for execution with sys_execve / sys_uselib */
82 #define FMODE_EXEC ((__force fmode_t)0x20)
83 /* File is opened with O_NDELAY (only set for block devices) */
84 #define FMODE_NDELAY ((__force fmode_t)0x40)
85 /* File is opened with O_EXCL (only set for block devices) */
86 #define FMODE_EXCL ((__force fmode_t)0x80)
87 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
88 (specialy hack for floppy.c) */
89 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
92 * Don't update ctime and mtime.
94 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
95 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
97 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
99 /* Expect random access pattern */
100 #define FMODE_RANDOM ((__force fmode_t)0x1000)
102 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
103 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
105 /* File is opened with O_PATH; almost nothing can be done with it */
106 #define FMODE_PATH ((__force fmode_t)0x4000)
108 /* File was opened by fanotify and shouldn't generate fanotify events */
109 #define FMODE_NONOTIFY ((__force fmode_t)0x1000000)
112 * The below are the various read and write types that we support. Some of
113 * them include behavioral modifiers that send information down to the
114 * block layer and IO scheduler. Terminology:
116 * The block layer uses device plugging to defer IO a little bit, in
117 * the hope that we will see more IO very shortly. This increases
118 * coalescing of adjacent IO and thus reduces the number of IOs we
119 * have to send to the device. It also allows for better queuing,
120 * if the IO isn't mergeable. If the caller is going to be waiting
121 * for the IO, then he must ensure that the device is unplugged so
122 * that the IO is dispatched to the driver.
124 * All IO is handled async in Linux. This is fine for background
125 * writes, but for reads or writes that someone waits for completion
126 * on, we want to notify the block layer and IO scheduler so that they
127 * know about it. That allows them to make better scheduling
128 * decisions. So when the below references 'sync' and 'async', it
129 * is referencing this priority hint.
131 * With that in mind, the available types are:
133 * READ A normal read operation. Device will be plugged.
134 * READ_SYNC A synchronous read. Device is not plugged, caller can
135 * immediately wait on this read without caring about
137 * READA Used for read-ahead operations. Lower priority, and the
138 * block layer could (in theory) choose to ignore this
139 * request if it runs into resource problems.
140 * WRITE A normal async write. Device will be plugged.
141 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
142 * the hint that someone will be waiting on this IO
143 * shortly. The write equivalent of READ_SYNC.
144 * WRITE_ODIRECT Special case write for O_DIRECT only.
145 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
146 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
147 * non-volatile media on completion.
148 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
149 * by a cache flush and data is guaranteed to be on
150 * non-volatile media on completion.
153 #define RW_MASK REQ_WRITE
154 #define RWA_MASK REQ_RAHEAD
157 #define WRITE RW_MASK
158 #define READA RWA_MASK
160 #define READ_SYNC (READ | REQ_SYNC)
161 #define READ_META (READ | REQ_META)
162 #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
163 #define WRITE_ODIRECT (WRITE | REQ_SYNC)
164 #define WRITE_META (WRITE | REQ_META)
165 #define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
166 #define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
167 #define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
173 /* public flags for file_system_type */
174 #define FS_REQUIRES_DEV 1
175 #define FS_BINARY_MOUNTDATA 2
176 #define FS_HAS_SUBTYPE 4
177 #define FS_REVAL_DOT 16384 /* Check the paths ".", ".." for staleness */
178 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move()
179 * during rename() internally.
183 * These are the fs-independent mount-flags: up to 32 flags are supported
185 #define MS_RDONLY 1 /* Mount read-only */
186 #define MS_NOSUID 2 /* Ignore suid and sgid bits */
187 #define MS_NODEV 4 /* Disallow access to device special files */
188 #define MS_NOEXEC 8 /* Disallow program execution */
189 #define MS_SYNCHRONOUS 16 /* Writes are synced at once */
190 #define MS_REMOUNT 32 /* Alter flags of a mounted FS */
191 #define MS_MANDLOCK 64 /* Allow mandatory locks on an FS */
192 #define MS_DIRSYNC 128 /* Directory modifications are synchronous */
193 #define MS_NOATIME 1024 /* Do not update access times. */
194 #define MS_NODIRATIME 2048 /* Do not update directory access times */
198 #define MS_VERBOSE 32768 /* War is peace. Verbosity is silence.
199 MS_VERBOSE is deprecated. */
200 #define MS_SILENT 32768
201 #define MS_POSIXACL (1<<16) /* VFS does not apply the umask */
202 #define MS_UNBINDABLE (1<<17) /* change to unbindable */
203 #define MS_PRIVATE (1<<18) /* change to private */
204 #define MS_SLAVE (1<<19) /* change to slave */
205 #define MS_SHARED (1<<20) /* change to shared */
206 #define MS_RELATIME (1<<21) /* Update atime relative to mtime/ctime. */
207 #define MS_KERNMOUNT (1<<22) /* this is a kern_mount call */
208 #define MS_I_VERSION (1<<23) /* Update inode I_version field */
209 #define MS_STRICTATIME (1<<24) /* Always perform atime updates */
210 #define MS_BORN (1<<29)
211 #define MS_ACTIVE (1<<30)
212 #define MS_NOUSER (1<<31)
215 * Superblock flags that can be altered by MS_REMOUNT
217 #define MS_RMT_MASK (MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK|MS_I_VERSION)
220 * Old magic mount flag and mask
222 #define MS_MGC_VAL 0xC0ED0000
223 #define MS_MGC_MSK 0xffff0000
225 /* Inode flags - they have nothing to superblock flags now */
227 #define S_SYNC 1 /* Writes are synced at once */
228 #define S_NOATIME 2 /* Do not update access times */
229 #define S_APPEND 4 /* Append-only file */
230 #define S_IMMUTABLE 8 /* Immutable file */
231 #define S_DEAD 16 /* removed, but still open directory */
232 #define S_NOQUOTA 32 /* Inode is not counted to quota */
233 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
234 #define S_NOCMTIME 128 /* Do not update file c/mtime */
235 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
236 #define S_PRIVATE 512 /* Inode is fs-internal */
237 #define S_IMA 1024 /* Inode has an associated IMA struct */
238 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
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)
264 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
265 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
266 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
267 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
269 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
270 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
271 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
272 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
273 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
274 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
276 /* the read-only stuff doesn't really belong here, but any other place is
277 probably as bad and I don't want to create yet another include file. */
279 #define BLKROSET _IO(0x12,93) /* set device read-only (0 = read-write) */
280 #define BLKROGET _IO(0x12,94) /* get read-only status (0 = read_write) */
281 #define BLKRRPART _IO(0x12,95) /* re-read partition table */
282 #define BLKGETSIZE _IO(0x12,96) /* return device size /512 (long *arg) */
283 #define BLKFLSBUF _IO(0x12,97) /* flush buffer cache */
284 #define BLKRASET _IO(0x12,98) /* set read ahead for block device */
285 #define BLKRAGET _IO(0x12,99) /* get current read ahead setting */
286 #define BLKFRASET _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
287 #define BLKFRAGET _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
288 #define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
289 #define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
290 #define BLKSSZGET _IO(0x12,104)/* get block device sector size */
292 #define BLKPG _IO(0x12,105)/* See blkpg.h */
294 /* Some people are morons. Do not use sizeof! */
296 #define BLKELVGET _IOR(0x12,106,size_t)/* elevator get */
297 #define BLKELVSET _IOW(0x12,107,size_t)/* elevator set */
298 /* This was here just to show that the number is taken -
299 probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
301 /* A jump here: 108-111 have been used for various private purposes. */
302 #define BLKBSZGET _IOR(0x12,112,size_t)
303 #define BLKBSZSET _IOW(0x12,113,size_t)
304 #define BLKGETSIZE64 _IOR(0x12,114,size_t) /* return device size in bytes (u64 *arg) */
305 #define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup)
306 #define BLKTRACESTART _IO(0x12,116)
307 #define BLKTRACESTOP _IO(0x12,117)
308 #define BLKTRACETEARDOWN _IO(0x12,118)
309 #define BLKDISCARD _IO(0x12,119)
310 #define BLKIOMIN _IO(0x12,120)
311 #define BLKIOOPT _IO(0x12,121)
312 #define BLKALIGNOFF _IO(0x12,122)
313 #define BLKPBSZGET _IO(0x12,123)
314 #define BLKDISCARDZEROES _IO(0x12,124)
315 #define BLKSECDISCARD _IO(0x12,125)
317 #define BMAP_IOCTL 1 /* obsolete - kept for compatibility */
318 #define FIBMAP _IO(0x00,1) /* bmap access */
319 #define FIGETBSZ _IO(0x00,2) /* get the block size used for bmap */
320 #define FIFREEZE _IOWR('X', 119, int) /* Freeze */
321 #define FITHAW _IOWR('X', 120, int) /* Thaw */
322 #define FITRIM _IOWR('X', 121, struct fstrim_range) /* Trim */
324 #define FS_IOC_GETFLAGS _IOR('f', 1, long)
325 #define FS_IOC_SETFLAGS _IOW('f', 2, long)
326 #define FS_IOC_GETVERSION _IOR('v', 1, long)
327 #define FS_IOC_SETVERSION _IOW('v', 2, long)
328 #define FS_IOC_FIEMAP _IOWR('f', 11, struct fiemap)
329 #define FS_IOC32_GETFLAGS _IOR('f', 1, int)
330 #define FS_IOC32_SETFLAGS _IOW('f', 2, int)
331 #define FS_IOC32_GETVERSION _IOR('v', 1, int)
332 #define FS_IOC32_SETVERSION _IOW('v', 2, int)
335 * Inode flags (FS_IOC_GETFLAGS / FS_IOC_SETFLAGS)
337 #define FS_SECRM_FL 0x00000001 /* Secure deletion */
338 #define FS_UNRM_FL 0x00000002 /* Undelete */
339 #define FS_COMPR_FL 0x00000004 /* Compress file */
340 #define FS_SYNC_FL 0x00000008 /* Synchronous updates */
341 #define FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
342 #define FS_APPEND_FL 0x00000020 /* writes to file may only append */
343 #define FS_NODUMP_FL 0x00000040 /* do not dump file */
344 #define FS_NOATIME_FL 0x00000080 /* do not update atime */
345 /* Reserved for compression usage... */
346 #define FS_DIRTY_FL 0x00000100
347 #define FS_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
348 #define FS_NOCOMP_FL 0x00000400 /* Don't compress */
349 #define FS_ECOMPR_FL 0x00000800 /* Compression error */
350 /* End compression flags --- maybe not all used */
351 #define FS_BTREE_FL 0x00001000 /* btree format dir */
352 #define FS_INDEX_FL 0x00001000 /* hash-indexed directory */
353 #define FS_IMAGIC_FL 0x00002000 /* AFS directory */
354 #define FS_JOURNAL_DATA_FL 0x00004000 /* Reserved for ext3 */
355 #define FS_NOTAIL_FL 0x00008000 /* file tail should not be merged */
356 #define FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
357 #define FS_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
358 #define FS_EXTENT_FL 0x00080000 /* Extents */
359 #define FS_DIRECTIO_FL 0x00100000 /* Use direct i/o */
360 #define FS_RESERVED_FL 0x80000000 /* reserved for ext2 lib */
362 #define FS_FL_USER_VISIBLE 0x0003DFFF /* User visible flags */
363 #define FS_FL_USER_MODIFIABLE 0x000380FF /* User modifiable flags */
366 #define SYNC_FILE_RANGE_WAIT_BEFORE 1
367 #define SYNC_FILE_RANGE_WRITE 2
368 #define SYNC_FILE_RANGE_WAIT_AFTER 4
372 #include <linux/linkage.h>
373 #include <linux/wait.h>
374 #include <linux/types.h>
375 #include <linux/kdev_t.h>
376 #include <linux/dcache.h>
377 #include <linux/path.h>
378 #include <linux/stat.h>
379 #include <linux/cache.h>
380 #include <linux/list.h>
381 #include <linux/radix-tree.h>
382 #include <linux/prio_tree.h>
383 #include <linux/init.h>
384 #include <linux/pid.h>
385 #include <linux/mutex.h>
386 #include <linux/capability.h>
387 #include <linux/semaphore.h>
388 #include <linux/fiemap.h>
389 #include <linux/rculist_bl.h>
391 #include <asm/atomic.h>
392 #include <asm/byteorder.h>
394 struct export_operations;
400 struct pipe_inode_info;
401 struct poll_table_struct;
403 struct vm_area_struct;
407 extern void __init inode_init(void);
408 extern void __init inode_init_early(void);
409 extern void __init files_init(unsigned long);
411 extern struct files_stat_struct files_stat;
412 extern unsigned long get_max_files(void);
413 extern int sysctl_nr_open;
414 extern struct inodes_stat_t inodes_stat;
415 extern int leases_enable, lease_break_time;
418 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
419 struct buffer_head *bh_result, int create);
420 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
421 ssize_t bytes, void *private, int ret,
425 * Attribute flags. These should be or-ed together to figure out what
428 #define ATTR_MODE (1 << 0)
429 #define ATTR_UID (1 << 1)
430 #define ATTR_GID (1 << 2)
431 #define ATTR_SIZE (1 << 3)
432 #define ATTR_ATIME (1 << 4)
433 #define ATTR_MTIME (1 << 5)
434 #define ATTR_CTIME (1 << 6)
435 #define ATTR_ATIME_SET (1 << 7)
436 #define ATTR_MTIME_SET (1 << 8)
437 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
438 #define ATTR_ATTR_FLAG (1 << 10)
439 #define ATTR_KILL_SUID (1 << 11)
440 #define ATTR_KILL_SGID (1 << 12)
441 #define ATTR_FILE (1 << 13)
442 #define ATTR_KILL_PRIV (1 << 14)
443 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
444 #define ATTR_TIMES_SET (1 << 16)
447 * This is the Inode Attributes structure, used for notify_change(). It
448 * uses the above definitions as flags, to know which values have changed.
449 * Also, in this manner, a Filesystem can look at only the values it cares
450 * about. Basically, these are the attributes that the VFS layer can
451 * request to change from the FS layer.
453 * Derek Atkins <warlord@MIT.EDU> 94-10-20
456 unsigned int ia_valid;
461 struct timespec ia_atime;
462 struct timespec ia_mtime;
463 struct timespec ia_ctime;
466 * Not an attribute, but an auxilary info for filesystems wanting to
467 * implement an ftruncate() like method. NOTE: filesystem should
468 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
470 struct file *ia_file;
474 * Includes for diskquotas.
476 #include <linux/quota.h>
479 * enum positive_aop_returns - aop return codes with specific semantics
481 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
482 * completed, that the page is still locked, and
483 * should be considered active. The VM uses this hint
484 * to return the page to the active list -- it won't
485 * be a candidate for writeback again in the near
486 * future. Other callers must be careful to unlock
487 * the page if they get this return. Returned by
490 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
491 * unlocked it and the page might have been truncated.
492 * The caller should back up to acquiring a new page and
493 * trying again. The aop will be taking reasonable
494 * precautions not to livelock. If the caller held a page
495 * reference, it should drop it before retrying. Returned
498 * address_space_operation functions return these large constants to indicate
499 * special semantics to the caller. These are much larger than the bytes in a
500 * page to allow for functions that return the number of bytes operated on in a
504 enum positive_aop_returns {
505 AOP_WRITEPAGE_ACTIVATE = 0x80000,
506 AOP_TRUNCATED_PAGE = 0x80001,
509 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
510 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
511 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
512 * helper code (eg buffer layer)
513 * to clear GFP_FS from alloc */
516 * oh the beauties of C type declarations.
519 struct address_space;
520 struct writeback_control;
523 const struct iovec *iov;
524 unsigned long nr_segs;
529 size_t iov_iter_copy_from_user_atomic(struct page *page,
530 struct iov_iter *i, unsigned long offset, size_t bytes);
531 size_t iov_iter_copy_from_user(struct page *page,
532 struct iov_iter *i, unsigned long offset, size_t bytes);
533 void iov_iter_advance(struct iov_iter *i, size_t bytes);
534 int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes);
535 size_t iov_iter_single_seg_count(struct iov_iter *i);
537 static inline void iov_iter_init(struct iov_iter *i,
538 const struct iovec *iov, unsigned long nr_segs,
539 size_t count, size_t written)
542 i->nr_segs = nr_segs;
544 i->count = count + written;
546 iov_iter_advance(i, written);
549 static inline size_t iov_iter_count(struct iov_iter *i)
555 * "descriptor" for what we're up to with a read.
556 * This allows us to use the same read code yet
557 * have multiple different users of the data that
558 * we read from a file.
560 * The simplest case just copies the data to user
573 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
574 unsigned long, unsigned long);
576 struct address_space_operations {
577 int (*writepage)(struct page *page, struct writeback_control *wbc);
578 int (*readpage)(struct file *, 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 void (*freepage)(struct page *);
601 ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
602 loff_t offset, unsigned long nr_segs);
603 int (*get_xip_mem)(struct address_space *, pgoff_t, int,
604 void **, unsigned long *);
605 /* migrate the contents of a page to the specified target */
606 int (*migratepage) (struct address_space *,
607 struct page *, struct page *);
608 int (*launder_page) (struct page *);
609 int (*is_partially_uptodate) (struct page *, read_descriptor_t *,
611 int (*error_remove_page)(struct address_space *, struct page *);
615 * pagecache_write_begin/pagecache_write_end must be used by general code
616 * to write into the pagecache.
618 int pagecache_write_begin(struct file *, struct address_space *mapping,
619 loff_t pos, unsigned len, unsigned flags,
620 struct page **pagep, void **fsdata);
622 int pagecache_write_end(struct file *, struct address_space *mapping,
623 loff_t pos, unsigned len, unsigned copied,
624 struct page *page, void *fsdata);
626 struct backing_dev_info;
627 struct address_space {
628 struct inode *host; /* owner: inode, block_device */
629 struct radix_tree_root page_tree; /* radix tree of all pages */
630 spinlock_t tree_lock; /* and lock protecting it */
631 unsigned int i_mmap_writable;/* count VM_SHARED mappings */
632 struct prio_tree_root i_mmap; /* tree of private and shared mappings */
633 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
634 spinlock_t i_mmap_lock; /* protect tree, count, list */
635 unsigned int truncate_count; /* Cover race condition with truncate */
636 unsigned long nrpages; /* number of total pages */
637 pgoff_t writeback_index;/* writeback starts here */
638 const struct address_space_operations *a_ops; /* methods */
639 unsigned long flags; /* error bits/gfp mask */
640 struct backing_dev_info *backing_dev_info; /* device readahead, etc */
641 spinlock_t private_lock; /* for use by the address_space */
642 struct list_head private_list; /* ditto */
643 struct address_space *assoc_mapping; /* ditto */
644 struct mutex unmap_mutex; /* to protect unmapping */
645 } __attribute__((aligned(sizeof(long))));
647 * On most architectures that alignment is already the case; but
648 * must be enforced here for CRIS, to let the least signficant bit
649 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
652 struct block_device {
653 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;
657 struct mutex bd_mutex; /* open/close mutex */
658 struct list_head bd_inodes;
662 bool bd_write_holder;
664 struct list_head bd_holder_disks;
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 /* RCU path lookup touches following: */
737 const struct inode_operations *i_op;
738 struct super_block *i_sb;
740 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
741 unsigned int i_flags;
742 struct mutex i_mutex;
744 unsigned long i_state;
745 unsigned long dirtied_when; /* jiffies of first dirtying */
747 struct hlist_node i_hash;
748 struct list_head i_wb_list; /* backing dev IO list */
749 struct list_head i_lru; /* inode LRU list */
750 struct list_head i_sb_list;
752 struct list_head i_dentry;
753 struct rcu_head i_rcu;
757 unsigned int i_nlink;
759 unsigned int i_blkbits;
762 #ifdef __NEED_I_SIZE_ORDERED
763 seqcount_t i_size_seqcount;
765 struct timespec i_atime;
766 struct timespec i_mtime;
767 struct timespec i_ctime;
769 unsigned short i_bytes;
770 struct rw_semaphore i_alloc_sem;
771 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
772 struct file_lock *i_flock;
773 struct address_space *i_mapping;
774 struct address_space i_data;
776 struct dquot *i_dquot[MAXQUOTAS];
778 struct list_head i_devices;
780 struct pipe_inode_info *i_pipe;
781 struct block_device *i_bdev;
787 #ifdef CONFIG_FSNOTIFY
788 __u32 i_fsnotify_mask; /* all events this inode cares about */
789 struct hlist_head i_fsnotify_marks;
793 atomic_t i_readcount; /* struct files open RO */
795 atomic_t i_writecount;
796 #ifdef CONFIG_SECURITY
799 #ifdef CONFIG_FS_POSIX_ACL
800 struct posix_acl *i_acl;
801 struct posix_acl *i_default_acl;
803 void *i_private; /* fs or device private pointer */
806 static inline int inode_unhashed(struct inode *inode)
808 return hlist_unhashed(&inode->i_hash);
812 * inode->i_mutex nesting subclasses for the lock validator:
814 * 0: the object of the current VFS operation
819 * The locking order between these classes is
820 * parent -> child -> normal -> xattr -> quota
822 enum inode_i_mutex_lock_class
832 * NOTE: in a 32bit arch with a preemptable kernel and
833 * an UP compile the i_size_read/write must be atomic
834 * with respect to the local cpu (unlike with preempt disabled),
835 * but they don't need to be atomic with respect to other cpus like in
836 * true SMP (so they need either to either locally disable irq around
837 * the read or for example on x86 they can be still implemented as a
838 * cmpxchg8b without the need of the lock prefix). For SMP compiles
839 * and 64bit archs it makes no difference if preempt is enabled or not.
841 static inline loff_t i_size_read(const struct inode *inode)
843 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
848 seq = read_seqcount_begin(&inode->i_size_seqcount);
849 i_size = inode->i_size;
850 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
852 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
856 i_size = inode->i_size;
860 return inode->i_size;
865 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
866 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
867 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
869 static inline void i_size_write(struct inode *inode, loff_t i_size)
871 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
872 write_seqcount_begin(&inode->i_size_seqcount);
873 inode->i_size = i_size;
874 write_seqcount_end(&inode->i_size_seqcount);
875 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
877 inode->i_size = i_size;
880 inode->i_size = i_size;
884 static inline unsigned iminor(const struct inode *inode)
886 return MINOR(inode->i_rdev);
889 static inline unsigned imajor(const struct inode *inode)
891 return MAJOR(inode->i_rdev);
894 extern struct block_device *I_BDEV(struct inode *inode);
897 rwlock_t lock; /* protects pid, uid, euid fields */
898 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
899 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
900 uid_t uid, euid; /* uid/euid of process setting the owner */
901 int signum; /* posix.1b rt signal to be delivered on IO */
905 * Track a single file's readahead state
907 struct file_ra_state {
908 pgoff_t start; /* where readahead started */
909 unsigned int size; /* # of readahead pages */
910 unsigned int async_size; /* do asynchronous readahead when
911 there are only # of pages ahead */
913 unsigned int ra_pages; /* Maximum readahead window */
914 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
915 loff_t prev_pos; /* Cache last read() position */
919 * Check if @index falls in the readahead windows.
921 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
923 return (index >= ra->start &&
924 index < ra->start + ra->size);
927 #define FILE_MNT_WRITE_TAKEN 1
928 #define FILE_MNT_WRITE_RELEASED 2
932 * fu_list becomes invalid after file_free is called and queued via
933 * fu_rcuhead for RCU freeing
936 struct list_head fu_list;
937 struct rcu_head fu_rcuhead;
940 #define f_dentry f_path.dentry
941 #define f_vfsmnt f_path.mnt
942 const struct file_operations *f_op;
943 spinlock_t f_lock; /* f_ep_links, f_flags, no IRQ */
947 atomic_long_t f_count;
948 unsigned int f_flags;
951 struct fown_struct f_owner;
952 const struct cred *f_cred;
953 struct file_ra_state f_ra;
956 #ifdef CONFIG_SECURITY
959 /* needed for tty driver, and maybe others */
963 /* Used by fs/eventpoll.c to link all the hooks to this file */
964 struct list_head f_ep_links;
965 #endif /* #ifdef CONFIG_EPOLL */
966 struct address_space *f_mapping;
967 #ifdef CONFIG_DEBUG_WRITECOUNT
968 unsigned long f_mnt_write_state;
975 /* file identifier */
976 unsigned char f_handle[0];
979 #define get_file(x) atomic_long_inc(&(x)->f_count)
980 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
981 #define file_count(x) atomic_long_read(&(x)->f_count)
983 #ifdef CONFIG_DEBUG_WRITECOUNT
984 static inline void file_take_write(struct file *f)
986 WARN_ON(f->f_mnt_write_state != 0);
987 f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN;
989 static inline void file_release_write(struct file *f)
991 f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED;
993 static inline void file_reset_write(struct file *f)
995 f->f_mnt_write_state = 0;
997 static inline void file_check_state(struct file *f)
1000 * At this point, either both or neither of these bits
1003 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN);
1004 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED);
1006 static inline int file_check_writeable(struct file *f)
1008 if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN)
1010 printk(KERN_WARNING "writeable file with no "
1011 "mnt_want_write()\n");
1015 #else /* !CONFIG_DEBUG_WRITECOUNT */
1016 static inline void file_take_write(struct file *filp) {}
1017 static inline void file_release_write(struct file *filp) {}
1018 static inline void file_reset_write(struct file *filp) {}
1019 static inline void file_check_state(struct file *filp) {}
1020 static inline int file_check_writeable(struct file *filp)
1024 #endif /* CONFIG_DEBUG_WRITECOUNT */
1026 #define MAX_NON_LFS ((1UL<<31) - 1)
1028 /* Page cache limit. The filesystems should put that into their s_maxbytes
1029 limits, otherwise bad things can happen in VM. */
1030 #if BITS_PER_LONG==32
1031 #define MAX_LFS_FILESIZE (((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
1032 #elif BITS_PER_LONG==64
1033 #define MAX_LFS_FILESIZE 0x7fffffffffffffffUL
1038 #define FL_ACCESS 8 /* not trying to lock, just looking */
1039 #define FL_EXISTS 16 /* when unlocking, test for existence */
1040 #define FL_LEASE 32 /* lease held on this file */
1041 #define FL_CLOSE 64 /* unlock on close */
1042 #define FL_SLEEP 128 /* A blocking lock */
1045 * Special return value from posix_lock_file() and vfs_lock_file() for
1046 * asynchronous locking.
1048 #define FILE_LOCK_DEFERRED 1
1051 * The POSIX file lock owner is determined by
1052 * the "struct files_struct" in the thread group
1053 * (or NULL for no owner - BSD locks).
1055 * Lockd stuffs a "host" pointer into this.
1057 typedef struct files_struct *fl_owner_t;
1059 struct file_lock_operations {
1060 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1061 void (*fl_release_private)(struct file_lock *);
1064 struct lock_manager_operations {
1065 int (*fl_compare_owner)(struct file_lock *, struct file_lock *);
1066 void (*fl_notify)(struct file_lock *); /* unblock callback */
1067 int (*fl_grant)(struct file_lock *, struct file_lock *, int);
1068 void (*fl_release_private)(struct file_lock *);
1069 void (*fl_break)(struct file_lock *);
1070 int (*fl_change)(struct file_lock **, int);
1073 struct lock_manager {
1074 struct list_head list;
1077 void locks_start_grace(struct lock_manager *);
1078 void locks_end_grace(struct lock_manager *);
1079 int locks_in_grace(void);
1081 /* that will die - we need it for nfs_lock_info */
1082 #include <linux/nfs_fs_i.h>
1085 struct file_lock *fl_next; /* singly linked list for this inode */
1086 struct list_head fl_link; /* doubly linked list of all locks */
1087 struct list_head fl_block; /* circular list of blocked processes */
1088 fl_owner_t fl_owner;
1089 unsigned char fl_flags;
1090 unsigned char fl_type;
1091 unsigned int fl_pid;
1092 struct pid *fl_nspid;
1093 wait_queue_head_t fl_wait;
1094 struct file *fl_file;
1098 struct fasync_struct * fl_fasync; /* for lease break notifications */
1099 unsigned long fl_break_time; /* for nonblocking lease breaks */
1101 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1102 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1104 struct nfs_lock_info nfs_fl;
1105 struct nfs4_lock_info nfs4_fl;
1107 struct list_head link; /* link in AFS vnode's pending_locks list */
1108 int state; /* state of grant or error if -ve */
1113 /* The following constant reflects the upper bound of the file/locking space */
1115 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1116 #define OFFSET_MAX INT_LIMIT(loff_t)
1117 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1120 #include <linux/fcntl.h>
1122 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1124 #ifdef CONFIG_FILE_LOCKING
1125 extern int fcntl_getlk(struct file *, struct flock __user *);
1126 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1127 struct flock __user *);
1129 #if BITS_PER_LONG == 32
1130 extern int fcntl_getlk64(struct file *, struct flock64 __user *);
1131 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1132 struct flock64 __user *);
1135 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1136 extern int fcntl_getlease(struct file *filp);
1139 void locks_free_lock(struct file_lock *fl);
1140 extern void locks_init_lock(struct file_lock *);
1141 extern struct file_lock * locks_alloc_lock(void);
1142 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1143 extern void __locks_copy_lock(struct file_lock *, const struct file_lock *);
1144 extern void locks_remove_posix(struct file *, fl_owner_t);
1145 extern void locks_remove_flock(struct file *);
1146 extern void locks_release_private(struct file_lock *);
1147 extern void posix_test_lock(struct file *, struct file_lock *);
1148 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1149 extern int posix_lock_file_wait(struct file *, struct file_lock *);
1150 extern int posix_unblock_lock(struct file *, struct file_lock *);
1151 extern int vfs_test_lock(struct file *, struct file_lock *);
1152 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1153 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1154 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
1155 extern int __break_lease(struct inode *inode, unsigned int flags);
1156 extern void lease_get_mtime(struct inode *, struct timespec *time);
1157 extern int generic_setlease(struct file *, long, struct file_lock **);
1158 extern int vfs_setlease(struct file *, long, struct file_lock **);
1159 extern int lease_modify(struct file_lock **, int);
1160 extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
1161 extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
1162 extern void lock_flocks(void);
1163 extern void unlock_flocks(void);
1164 #else /* !CONFIG_FILE_LOCKING */
1165 static inline int fcntl_getlk(struct file *file, struct flock __user *user)
1170 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1171 unsigned int cmd, struct flock __user *user)
1176 #if BITS_PER_LONG == 32
1177 static inline int fcntl_getlk64(struct file *file, struct flock64 __user *user)
1182 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1183 unsigned int cmd, struct flock64 __user *user)
1188 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1193 static inline int fcntl_getlease(struct file *filp)
1198 static inline void locks_init_lock(struct file_lock *fl)
1203 static inline void __locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1208 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1213 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1218 static inline void locks_remove_flock(struct file *filp)
1223 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1228 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1229 struct file_lock *conflock)
1234 static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1239 static inline int posix_unblock_lock(struct file *filp,
1240 struct file_lock *waiter)
1245 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1250 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1251 struct file_lock *fl, struct file_lock *conf)
1256 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1261 static inline int flock_lock_file_wait(struct file *filp,
1262 struct file_lock *request)
1267 static inline int __break_lease(struct inode *inode, unsigned int mode)
1272 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1277 static inline int generic_setlease(struct file *filp, long arg,
1278 struct file_lock **flp)
1283 static inline int vfs_setlease(struct file *filp, long arg,
1284 struct file_lock **lease)
1289 static inline int lease_modify(struct file_lock **before, int arg)
1294 static inline int lock_may_read(struct inode *inode, loff_t start,
1300 static inline int lock_may_write(struct inode *inode, loff_t start,
1306 static inline void lock_flocks(void)
1310 static inline void unlock_flocks(void)
1314 #endif /* !CONFIG_FILE_LOCKING */
1317 struct fasync_struct {
1321 struct fasync_struct *fa_next; /* singly linked list */
1322 struct file *fa_file;
1323 struct rcu_head fa_rcu;
1326 #define FASYNC_MAGIC 0x4601
1328 /* SMP safe fasync helpers: */
1329 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1330 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1331 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1332 extern struct fasync_struct *fasync_alloc(void);
1333 extern void fasync_free(struct fasync_struct *);
1335 /* can be called from interrupts */
1336 extern void kill_fasync(struct fasync_struct **, int, int);
1338 extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1339 extern int f_setown(struct file *filp, unsigned long arg, int force);
1340 extern void f_delown(struct file *filp);
1341 extern pid_t f_getown(struct file *filp);
1342 extern int send_sigurg(struct fown_struct *fown);
1348 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1349 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1350 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1351 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1352 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1354 extern struct list_head super_blocks;
1355 extern spinlock_t sb_lock;
1357 struct super_block {
1358 struct list_head s_list; /* Keep this first */
1359 dev_t s_dev; /* search index; _not_ kdev_t */
1360 unsigned char s_dirt;
1361 unsigned char s_blocksize_bits;
1362 unsigned long s_blocksize;
1363 loff_t s_maxbytes; /* Max file size */
1364 struct file_system_type *s_type;
1365 const struct super_operations *s_op;
1366 const struct dquot_operations *dq_op;
1367 const struct quotactl_ops *s_qcop;
1368 const struct export_operations *s_export_op;
1369 unsigned long s_flags;
1370 unsigned long s_magic;
1371 struct dentry *s_root;
1372 struct rw_semaphore s_umount;
1373 struct mutex s_lock;
1376 #ifdef CONFIG_SECURITY
1379 const struct xattr_handler **s_xattr;
1381 struct list_head s_inodes; /* all inodes */
1382 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1384 struct list_head __percpu *s_files;
1386 struct list_head s_files;
1388 /* s_dentry_lru, s_nr_dentry_unused protected by dcache.c lru locks */
1389 struct list_head s_dentry_lru; /* unused dentry lru */
1390 int s_nr_dentry_unused; /* # of dentry on lru */
1392 struct block_device *s_bdev;
1393 struct backing_dev_info *s_bdi;
1394 struct mtd_info *s_mtd;
1395 struct list_head s_instances;
1396 struct quota_info s_dquot; /* Diskquota specific options */
1399 wait_queue_head_t s_wait_unfrozen;
1401 char s_id[32]; /* Informational name */
1402 u8 s_uuid[16]; /* UUID */
1404 void *s_fs_info; /* Filesystem private info */
1407 /* Granularity of c/m/atime in ns.
1408 Cannot be worse than a second */
1412 * The next field is for VFS *only*. No filesystems have any business
1413 * even looking at it. You had been warned.
1415 struct mutex s_vfs_rename_mutex; /* Kludge */
1418 * Filesystem subtype. If non-empty the filesystem type field
1419 * in /proc/mounts will be "type.subtype"
1424 * Saved mount options for lazy filesystems using
1425 * generic_show_options()
1427 char __rcu *s_options;
1428 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1431 extern struct timespec current_fs_time(struct super_block *sb);
1434 * Snapshotting support.
1438 SB_FREEZE_WRITE = 1,
1439 SB_FREEZE_TRANS = 2,
1442 #define vfs_check_frozen(sb, level) \
1443 wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
1445 #define get_fs_excl() atomic_inc(¤t->fs_excl)
1446 #define put_fs_excl() atomic_dec(¤t->fs_excl)
1447 #define has_fs_excl() atomic_read(¤t->fs_excl)
1450 * until VFS tracks user namespaces for inodes, just make all files
1451 * belong to init_user_ns
1453 extern struct user_namespace init_user_ns;
1454 #define inode_userns(inode) (&init_user_ns)
1455 extern bool inode_owner_or_capable(const struct inode *inode);
1457 /* not quite ready to be deprecated, but... */
1458 extern void lock_super(struct super_block *);
1459 extern void unlock_super(struct super_block *);
1462 * VFS helper functions..
1464 extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *);
1465 extern int vfs_mkdir(struct inode *, struct dentry *, int);
1466 extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t);
1467 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1468 extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
1469 extern int vfs_rmdir(struct inode *, struct dentry *);
1470 extern int vfs_unlink(struct inode *, struct dentry *);
1471 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1474 * VFS dentry helper functions.
1476 extern void dentry_unhash(struct dentry *dentry);
1479 * VFS file helper functions.
1481 extern int file_permission(struct file *, int);
1482 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1485 * VFS FS_IOC_FIEMAP helper definitions.
1487 struct fiemap_extent_info {
1488 unsigned int fi_flags; /* Flags as passed from user */
1489 unsigned int fi_extents_mapped; /* Number of mapped extents */
1490 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1491 struct fiemap_extent __user *fi_extents_start; /* Start of
1492 fiemap_extent array */
1494 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1495 u64 phys, u64 len, u32 flags);
1496 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1501 * NOTE! These match bits 12..15 of stat.st_mode
1502 * (ie "(i_mode >> 12) & 15").
1504 #define DT_UNKNOWN 0
1515 * This is the "filldir" function type, used by readdir() to let
1516 * the kernel specify what kind of dirent layout it wants to have.
1517 * This allows the kernel to read directories into kernel space or
1518 * to have different dirent layouts depending on the binary type.
1520 typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
1521 struct block_device_operations;
1523 /* These macros are for out of kernel modules to test that
1524 * the kernel supports the unlocked_ioctl and compat_ioctl
1525 * fields in struct file_operations. */
1526 #define HAVE_COMPAT_IOCTL 1
1527 #define HAVE_UNLOCKED_IOCTL 1
1531 * all file operations except setlease can be called without
1532 * the big kernel lock held in all filesystems.
1534 struct file_operations {
1535 struct module *owner;
1536 loff_t (*llseek) (struct file *, loff_t, int);
1537 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1538 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1539 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1540 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1541 int (*readdir) (struct file *, void *, filldir_t);
1542 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1543 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1544 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1545 int (*mmap) (struct file *, struct vm_area_struct *);
1546 int (*open) (struct inode *, struct file *);
1547 int (*flush) (struct file *, fl_owner_t id);
1548 int (*release) (struct inode *, struct file *);
1549 int (*fsync) (struct file *, int datasync);
1550 int (*aio_fsync) (struct kiocb *, int datasync);
1551 int (*fasync) (int, struct file *, int);
1552 int (*lock) (struct file *, int, struct file_lock *);
1553 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1554 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1555 int (*check_flags)(int);
1556 int (*flock) (struct file *, int, struct file_lock *);
1557 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1558 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1559 int (*setlease)(struct file *, long, struct file_lock **);
1560 long (*fallocate)(struct file *file, int mode, loff_t offset,
1564 #define IPERM_FLAG_RCU 0x0001
1566 struct inode_operations {
1567 struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
1568 void * (*follow_link) (struct dentry *, struct nameidata *);
1569 int (*permission) (struct inode *, int, unsigned int);
1570 int (*check_acl)(struct inode *, int, unsigned int);
1572 int (*readlink) (struct dentry *, char __user *,int);
1573 void (*put_link) (struct dentry *, struct nameidata *, void *);
1575 int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
1576 int (*link) (struct dentry *,struct inode *,struct dentry *);
1577 int (*unlink) (struct inode *,struct dentry *);
1578 int (*symlink) (struct inode *,struct dentry *,const char *);
1579 int (*mkdir) (struct inode *,struct dentry *,int);
1580 int (*rmdir) (struct inode *,struct dentry *);
1581 int (*mknod) (struct inode *,struct dentry *,int,dev_t);
1582 int (*rename) (struct inode *, struct dentry *,
1583 struct inode *, struct dentry *);
1584 void (*truncate) (struct inode *);
1585 int (*setattr) (struct dentry *, struct iattr *);
1586 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1587 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1588 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1589 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1590 int (*removexattr) (struct dentry *, const char *);
1591 void (*truncate_range)(struct inode *, loff_t, loff_t);
1592 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1594 } ____cacheline_aligned;
1598 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1599 unsigned long nr_segs, unsigned long fast_segs,
1600 struct iovec *fast_pointer,
1601 struct iovec **ret_pointer);
1603 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1604 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1605 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1606 unsigned long, loff_t *);
1607 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1608 unsigned long, loff_t *);
1610 struct super_operations {
1611 struct inode *(*alloc_inode)(struct super_block *sb);
1612 void (*destroy_inode)(struct inode *);
1614 void (*dirty_inode) (struct inode *);
1615 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1616 int (*drop_inode) (struct inode *);
1617 void (*evict_inode) (struct inode *);
1618 void (*put_super) (struct super_block *);
1619 void (*write_super) (struct super_block *);
1620 int (*sync_fs)(struct super_block *sb, int wait);
1621 int (*freeze_fs) (struct super_block *);
1622 int (*unfreeze_fs) (struct super_block *);
1623 int (*statfs) (struct dentry *, struct kstatfs *);
1624 int (*remount_fs) (struct super_block *, int *, char *);
1625 void (*umount_begin) (struct super_block *);
1627 int (*show_options)(struct seq_file *, struct vfsmount *);
1628 int (*show_devname)(struct seq_file *, struct vfsmount *);
1629 int (*show_path)(struct seq_file *, struct vfsmount *);
1630 int (*show_stats)(struct seq_file *, struct vfsmount *);
1632 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1633 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1635 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1639 * Inode state bits. Protected by inode->i_lock
1641 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1642 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1644 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1645 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1646 * various stages of removing an inode.
1648 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1650 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1651 * fdatasync(). i_atime is the usual cause.
1652 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1653 * these changes separately from I_DIRTY_SYNC so that we
1654 * don't have to write inode on fdatasync() when only
1655 * mtime has changed in it.
1656 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1657 * I_NEW Serves as both a mutex and completion notification.
1658 * New inodes set I_NEW. If two processes both create
1659 * the same inode, one of them will release its inode and
1660 * wait for I_NEW to be released before returning.
1661 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1662 * also cause waiting on I_NEW, without I_NEW actually
1663 * being set. find_inode() uses this to prevent returning
1664 * nearly-dead inodes.
1665 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1666 * is zero. I_FREEING must be set when I_WILL_FREE is
1668 * I_FREEING Set when inode is about to be freed but still has dirty
1669 * pages or buffers attached or the inode itself is still
1671 * I_CLEAR Added by end_writeback(). In this state the inode is clean
1672 * and can be destroyed. Inode keeps I_FREEING.
1674 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1675 * prohibited for many purposes. iget() must wait for
1676 * the inode to be completely released, then create it
1677 * anew. Other functions will just ignore such inodes,
1678 * if appropriate. I_NEW is used for waiting.
1680 * I_SYNC Synchonized write of dirty inode data. The bits is
1681 * set during data writeback, and cleared with a wakeup
1682 * on the bit address once it is done.
1684 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1686 #define I_DIRTY_SYNC (1 << 0)
1687 #define I_DIRTY_DATASYNC (1 << 1)
1688 #define I_DIRTY_PAGES (1 << 2)
1690 #define I_NEW (1 << __I_NEW)
1691 #define I_WILL_FREE (1 << 4)
1692 #define I_FREEING (1 << 5)
1693 #define I_CLEAR (1 << 6)
1695 #define I_SYNC (1 << __I_SYNC)
1696 #define I_REFERENCED (1 << 8)
1698 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1700 extern void __mark_inode_dirty(struct inode *, int);
1701 static inline void mark_inode_dirty(struct inode *inode)
1703 __mark_inode_dirty(inode, I_DIRTY);
1706 static inline void mark_inode_dirty_sync(struct inode *inode)
1708 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1712 * inc_nlink - directly increment an inode's link count
1715 * This is a low-level filesystem helper to replace any
1716 * direct filesystem manipulation of i_nlink. Currently,
1717 * it is only here for parity with dec_nlink().
1719 static inline void inc_nlink(struct inode *inode)
1724 static inline void inode_inc_link_count(struct inode *inode)
1727 mark_inode_dirty(inode);
1731 * drop_nlink - directly drop an inode's link count
1734 * This is a low-level filesystem helper to replace any
1735 * direct filesystem manipulation of i_nlink. In cases
1736 * where we are attempting to track writes to the
1737 * filesystem, a decrement to zero means an imminent
1738 * write when the file is truncated and actually unlinked
1739 * on the filesystem.
1741 static inline void drop_nlink(struct inode *inode)
1747 * clear_nlink - directly zero an inode's link count
1750 * This is a low-level filesystem helper to replace any
1751 * direct filesystem manipulation of i_nlink. See
1752 * drop_nlink() for why we care about i_nlink hitting zero.
1754 static inline void clear_nlink(struct inode *inode)
1759 static inline void inode_dec_link_count(struct inode *inode)
1762 mark_inode_dirty(inode);
1766 * inode_inc_iversion - increments i_version
1767 * @inode: inode that need to be updated
1769 * Every time the inode is modified, the i_version field will be incremented.
1770 * The filesystem has to be mounted with i_version flag
1773 static inline void inode_inc_iversion(struct inode *inode)
1775 spin_lock(&inode->i_lock);
1777 spin_unlock(&inode->i_lock);
1780 extern void touch_atime(struct vfsmount *mnt, struct dentry *dentry);
1781 static inline void file_accessed(struct file *file)
1783 if (!(file->f_flags & O_NOATIME))
1784 touch_atime(file->f_path.mnt, file->f_path.dentry);
1787 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1788 int sync_inode_metadata(struct inode *inode, int wait);
1790 struct file_system_type {
1793 struct dentry *(*mount) (struct file_system_type *, int,
1794 const char *, void *);
1795 void (*kill_sb) (struct super_block *);
1796 struct module *owner;
1797 struct file_system_type * next;
1798 struct list_head fs_supers;
1800 struct lock_class_key s_lock_key;
1801 struct lock_class_key s_umount_key;
1802 struct lock_class_key s_vfs_rename_key;
1804 struct lock_class_key i_lock_key;
1805 struct lock_class_key i_mutex_key;
1806 struct lock_class_key i_mutex_dir_key;
1807 struct lock_class_key i_alloc_sem_key;
1810 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1811 void *data, int (*fill_super)(struct super_block *, void *, int));
1812 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1813 int flags, const char *dev_name, void *data,
1814 int (*fill_super)(struct super_block *, void *, int));
1815 extern struct dentry *mount_single(struct file_system_type *fs_type,
1816 int flags, void *data,
1817 int (*fill_super)(struct super_block *, void *, int));
1818 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1819 int flags, void *data,
1820 int (*fill_super)(struct super_block *, void *, int));
1821 void generic_shutdown_super(struct super_block *sb);
1822 void kill_block_super(struct super_block *sb);
1823 void kill_anon_super(struct super_block *sb);
1824 void kill_litter_super(struct super_block *sb);
1825 void deactivate_super(struct super_block *sb);
1826 void deactivate_locked_super(struct super_block *sb);
1827 int set_anon_super(struct super_block *s, void *data);
1828 struct super_block *sget(struct file_system_type *type,
1829 int (*test)(struct super_block *,void *),
1830 int (*set)(struct super_block *,void *),
1832 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
1833 const struct super_operations *ops,
1834 const struct dentry_operations *dops,
1837 static inline void sb_mark_dirty(struct super_block *sb)
1841 static inline void sb_mark_clean(struct super_block *sb)
1845 static inline int sb_is_dirty(struct super_block *sb)
1850 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1851 #define fops_get(fops) \
1852 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1853 #define fops_put(fops) \
1854 do { if (fops) module_put((fops)->owner); } while(0)
1856 extern int register_filesystem(struct file_system_type *);
1857 extern int unregister_filesystem(struct file_system_type *);
1858 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1859 #define kern_mount(type) kern_mount_data(type, NULL)
1860 extern int may_umount_tree(struct vfsmount *);
1861 extern int may_umount(struct vfsmount *);
1862 extern long do_mount(char *, char *, char *, unsigned long, void *);
1863 extern struct vfsmount *collect_mounts(struct path *);
1864 extern void drop_collected_mounts(struct vfsmount *);
1865 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1867 extern int vfs_statfs(struct path *, struct kstatfs *);
1868 extern int user_statfs(const char __user *, struct kstatfs *);
1869 extern int fd_statfs(int, struct kstatfs *);
1870 extern int statfs_by_dentry(struct dentry *, struct kstatfs *);
1871 extern int freeze_super(struct super_block *super);
1872 extern int thaw_super(struct super_block *super);
1874 extern int current_umask(void);
1877 extern struct kobject *fs_kobj;
1879 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
1880 extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1882 #define FLOCK_VERIFY_READ 1
1883 #define FLOCK_VERIFY_WRITE 2
1885 #ifdef CONFIG_FILE_LOCKING
1886 extern int locks_mandatory_locked(struct inode *);
1887 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1890 * Candidates for mandatory locking have the setgid bit set
1891 * but no group execute bit - an otherwise meaningless combination.
1894 static inline int __mandatory_lock(struct inode *ino)
1896 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1900 * ... and these candidates should be on MS_MANDLOCK mounted fs,
1901 * otherwise these will be advisory locks
1904 static inline int mandatory_lock(struct inode *ino)
1906 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1909 static inline int locks_verify_locked(struct inode *inode)
1911 if (mandatory_lock(inode))
1912 return locks_mandatory_locked(inode);
1916 static inline int locks_verify_truncate(struct inode *inode,
1920 if (inode->i_flock && mandatory_lock(inode))
1921 return locks_mandatory_area(
1922 FLOCK_VERIFY_WRITE, inode, filp,
1923 size < inode->i_size ? size : inode->i_size,
1924 (size < inode->i_size ? inode->i_size - size
1925 : size - inode->i_size)
1930 static inline int break_lease(struct inode *inode, unsigned int mode)
1933 return __break_lease(inode, mode);
1936 #else /* !CONFIG_FILE_LOCKING */
1937 static inline int locks_mandatory_locked(struct inode *inode)
1942 static inline int locks_mandatory_area(int rw, struct inode *inode,
1943 struct file *filp, loff_t offset,
1949 static inline int __mandatory_lock(struct inode *inode)
1954 static inline int mandatory_lock(struct inode *inode)
1959 static inline int locks_verify_locked(struct inode *inode)
1964 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
1970 static inline int break_lease(struct inode *inode, unsigned int mode)
1975 #endif /* CONFIG_FILE_LOCKING */
1979 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
1981 extern int do_fallocate(struct file *file, int mode, loff_t offset,
1983 extern long do_sys_open(int dfd, const char __user *filename, int flags,
1985 extern struct file *filp_open(const char *, int, int);
1986 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
1988 extern struct file * dentry_open(struct dentry *, struct vfsmount *, int,
1989 const struct cred *);
1990 extern int filp_close(struct file *, fl_owner_t id);
1991 extern char * getname(const char __user *);
1995 extern int ioctl_preallocate(struct file *filp, void __user *argp);
1998 extern void __init vfs_caches_init_early(void);
1999 extern void __init vfs_caches_init(unsigned long);
2001 extern struct kmem_cache *names_cachep;
2003 #define __getname_gfp(gfp) kmem_cache_alloc(names_cachep, (gfp))
2004 #define __getname() __getname_gfp(GFP_KERNEL)
2005 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2006 #ifndef CONFIG_AUDITSYSCALL
2007 #define putname(name) __putname(name)
2009 extern void putname(const char *name);
2013 extern int register_blkdev(unsigned int, const char *);
2014 extern void unregister_blkdev(unsigned int, const char *);
2015 extern struct block_device *bdget(dev_t);
2016 extern struct block_device *bdgrab(struct block_device *bdev);
2017 extern void bd_set_size(struct block_device *, loff_t size);
2018 extern void bd_forget(struct inode *inode);
2019 extern void bdput(struct block_device *);
2020 extern void invalidate_bdev(struct block_device *);
2021 extern int sync_blockdev(struct block_device *bdev);
2022 extern struct super_block *freeze_bdev(struct block_device *);
2023 extern void emergency_thaw_all(void);
2024 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2025 extern int fsync_bdev(struct block_device *);
2027 static inline void bd_forget(struct inode *inode) {}
2028 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2029 static inline void invalidate_bdev(struct block_device *bdev) {}
2031 static inline struct super_block *freeze_bdev(struct block_device *sb)
2036 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2041 extern int sync_filesystem(struct super_block *);
2042 extern const struct file_operations def_blk_fops;
2043 extern const struct file_operations def_chr_fops;
2044 extern const struct file_operations bad_sock_fops;
2045 extern const struct file_operations def_fifo_fops;
2047 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2048 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2049 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2050 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2051 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2053 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2055 extern int blkdev_put(struct block_device *bdev, fmode_t mode);
2057 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2058 extern void bd_unlink_disk_holder(struct block_device *bdev,
2059 struct gendisk *disk);
2061 static inline int bd_link_disk_holder(struct block_device *bdev,
2062 struct gendisk *disk)
2066 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2067 struct gendisk *disk)
2074 #define CHRDEV_MAJOR_HASH_SIZE 255
2075 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2076 extern int register_chrdev_region(dev_t, unsigned, const char *);
2077 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2078 unsigned int count, const char *name,
2079 const struct file_operations *fops);
2080 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2081 unsigned int count, const char *name);
2082 extern void unregister_chrdev_region(dev_t, unsigned);
2083 extern void chrdev_show(struct seq_file *,off_t);
2085 static inline int register_chrdev(unsigned int major, const char *name,
2086 const struct file_operations *fops)
2088 return __register_chrdev(major, 0, 256, name, fops);
2091 static inline void unregister_chrdev(unsigned int major, const char *name)
2093 __unregister_chrdev(major, 0, 256, name);
2096 /* fs/block_dev.c */
2097 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2098 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2101 #define BLKDEV_MAJOR_HASH_SIZE 255
2102 extern const char *__bdevname(dev_t, char *buffer);
2103 extern const char *bdevname(struct block_device *bdev, char *buffer);
2104 extern struct block_device *lookup_bdev(const char *);
2105 extern void blkdev_show(struct seq_file *,off_t);
2108 #define BLKDEV_MAJOR_HASH_SIZE 0
2111 extern void init_special_inode(struct inode *, umode_t, dev_t);
2113 /* Invalid inode operations -- fs/bad_inode.c */
2114 extern void make_bad_inode(struct inode *);
2115 extern int is_bad_inode(struct inode *);
2117 extern const struct file_operations read_pipefifo_fops;
2118 extern const struct file_operations write_pipefifo_fops;
2119 extern const struct file_operations rdwr_pipefifo_fops;
2121 extern int fs_may_remount_ro(struct super_block *);
2125 * return READ, READA, or WRITE
2127 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2130 * return data direction, READ or WRITE
2132 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2134 extern void check_disk_size_change(struct gendisk *disk,
2135 struct block_device *bdev);
2136 extern int revalidate_disk(struct gendisk *);
2137 extern int check_disk_change(struct block_device *);
2138 extern int __invalidate_device(struct block_device *, bool);
2139 extern int invalidate_partition(struct gendisk *, int);
2141 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2142 pgoff_t start, pgoff_t end);
2144 static inline void invalidate_remote_inode(struct inode *inode)
2146 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2147 S_ISLNK(inode->i_mode))
2148 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2150 extern int invalidate_inode_pages2(struct address_space *mapping);
2151 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2152 pgoff_t start, pgoff_t end);
2153 extern int write_inode_now(struct inode *, int);
2154 extern int filemap_fdatawrite(struct address_space *);
2155 extern int filemap_flush(struct address_space *);
2156 extern int filemap_fdatawait(struct address_space *);
2157 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2159 extern int filemap_write_and_wait(struct address_space *mapping);
2160 extern int filemap_write_and_wait_range(struct address_space *mapping,
2161 loff_t lstart, loff_t lend);
2162 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2163 loff_t start, loff_t end, int sync_mode);
2164 extern int filemap_fdatawrite_range(struct address_space *mapping,
2165 loff_t start, loff_t end);
2167 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2169 extern int vfs_fsync(struct file *file, int datasync);
2170 extern int generic_write_sync(struct file *file, loff_t pos, loff_t count);
2171 extern void sync_supers(void);
2172 extern void emergency_sync(void);
2173 extern void emergency_remount(void);
2175 extern sector_t bmap(struct inode *, sector_t);
2177 extern int notify_change(struct dentry *, struct iattr *);
2178 extern int inode_permission(struct inode *, int);
2179 extern int generic_permission(struct inode *, int, unsigned int,
2180 int (*check_acl)(struct inode *, int, unsigned int));
2182 static inline bool execute_ok(struct inode *inode)
2184 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2187 extern int get_write_access(struct inode *);
2188 extern int deny_write_access(struct file *);
2189 static inline void put_write_access(struct inode * inode)
2191 atomic_dec(&inode->i_writecount);
2193 static inline void allow_write_access(struct file *file)
2196 atomic_inc(&file->f_path.dentry->d_inode->i_writecount);
2199 static inline void i_readcount_dec(struct inode *inode)
2201 BUG_ON(!atomic_read(&inode->i_readcount));
2202 atomic_dec(&inode->i_readcount);
2204 static inline void i_readcount_inc(struct inode *inode)
2206 atomic_inc(&inode->i_readcount);
2209 static inline void i_readcount_dec(struct inode *inode)
2213 static inline void i_readcount_inc(struct inode *inode)
2218 extern int do_pipe_flags(int *, int);
2219 extern struct file *create_read_pipe(struct file *f, int flags);
2220 extern struct file *create_write_pipe(int flags);
2221 extern void free_write_pipe(struct file *);
2223 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2224 extern struct file * open_exec(const char *);
2226 /* fs/dcache.c -- generic fs support functions */
2227 extern int is_subdir(struct dentry *, struct dentry *);
2228 extern int path_is_under(struct path *, struct path *);
2229 extern ino_t find_inode_number(struct dentry *, struct qstr *);
2231 #include <linux/err.h>
2233 /* needed for stackable file system support */
2234 extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
2236 extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
2238 extern int inode_init_always(struct super_block *, struct inode *);
2239 extern void inode_init_once(struct inode *);
2240 extern void address_space_init_once(struct address_space *mapping);
2241 extern void ihold(struct inode * inode);
2242 extern void iput(struct inode *);
2243 extern struct inode * igrab(struct inode *);
2244 extern ino_t iunique(struct super_block *, ino_t);
2245 extern int inode_needs_sync(struct inode *inode);
2246 extern int generic_delete_inode(struct inode *inode);
2247 extern int generic_drop_inode(struct inode *inode);
2249 extern struct inode *ilookup5_nowait(struct super_block *sb,
2250 unsigned long hashval, int (*test)(struct inode *, void *),
2252 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2253 int (*test)(struct inode *, void *), void *data);
2254 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2256 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2257 extern struct inode * iget_locked(struct super_block *, unsigned long);
2258 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2259 extern int insert_inode_locked(struct inode *);
2260 extern void unlock_new_inode(struct inode *);
2261 extern unsigned int get_next_ino(void);
2263 extern void __iget(struct inode * inode);
2264 extern void iget_failed(struct inode *);
2265 extern void end_writeback(struct inode *);
2266 extern void __destroy_inode(struct inode *);
2267 extern struct inode *new_inode(struct super_block *);
2268 extern void free_inode_nonrcu(struct inode *inode);
2269 extern int should_remove_suid(struct dentry *);
2270 extern int file_remove_suid(struct file *);
2272 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2273 extern void remove_inode_hash(struct inode *);
2274 static inline void insert_inode_hash(struct inode *inode)
2276 __insert_inode_hash(inode, inode->i_ino);
2278 extern void inode_sb_list_add(struct inode *inode);
2281 extern void submit_bio(int, struct bio *);
2282 extern int bdev_read_only(struct block_device *);
2284 extern int set_blocksize(struct block_device *, int);
2285 extern int sb_set_blocksize(struct super_block *, int);
2286 extern int sb_min_blocksize(struct super_block *, int);
2288 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2289 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2290 extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
2291 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
2292 extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2293 extern ssize_t __generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long,
2295 extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2296 extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
2297 unsigned long *, loff_t, loff_t *, size_t, size_t);
2298 extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
2299 unsigned long, loff_t, loff_t *, size_t, ssize_t);
2300 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2301 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2302 extern int generic_segment_checks(const struct iovec *iov,
2303 unsigned long *nr_segs, size_t *count, int access_flags);
2305 /* fs/block_dev.c */
2306 extern ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov,
2307 unsigned long nr_segs, loff_t pos);
2308 extern int blkdev_fsync(struct file *filp, int datasync);
2311 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2312 struct pipe_inode_info *, size_t, unsigned int);
2313 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2314 struct pipe_inode_info *, size_t, unsigned int);
2315 extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
2316 struct file *, loff_t *, size_t, unsigned int);
2317 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2318 struct file *out, loff_t *, size_t len, unsigned int flags);
2319 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2320 size_t len, unsigned int flags);
2323 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2324 extern loff_t noop_llseek(struct file *file, loff_t offset, int origin);
2325 extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
2326 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
2327 extern loff_t generic_file_llseek_unlocked(struct file *file, loff_t offset,
2329 extern int generic_file_open(struct inode * inode, struct file * filp);
2330 extern int nonseekable_open(struct inode * inode, struct file * filp);
2332 #ifdef CONFIG_FS_XIP
2333 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2335 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
2336 extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2337 size_t len, loff_t *ppos);
2338 extern int xip_truncate_page(struct address_space *mapping, loff_t from);
2340 static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2347 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2348 loff_t file_offset);
2351 /* need locking between buffered and direct access */
2354 /* filesystem does not support filling holes */
2355 DIO_SKIP_HOLES = 0x02,
2358 void dio_end_io(struct bio *bio, int error);
2360 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2361 struct block_device *bdev, const struct iovec *iov, loff_t offset,
2362 unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
2363 dio_submit_t submit_io, int flags);
2365 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
2366 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
2367 loff_t offset, unsigned long nr_segs, get_block_t get_block,
2368 dio_iodone_t end_io)
2370 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
2371 nr_segs, get_block, end_io, NULL,
2372 DIO_LOCKING | DIO_SKIP_HOLES);
2376 extern const struct file_operations generic_ro_fops;
2378 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2380 extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
2381 extern int vfs_follow_link(struct nameidata *, const char *);
2382 extern int page_readlink(struct dentry *, char __user *, int);
2383 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2384 extern void page_put_link(struct dentry *, struct nameidata *, void *);
2385 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2387 extern int page_symlink(struct inode *inode, const char *symname, int len);
2388 extern const struct inode_operations page_symlink_inode_operations;
2389 extern int generic_readlink(struct dentry *, char __user *, int);
2390 extern void generic_fillattr(struct inode *, struct kstat *);
2391 extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2392 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2393 void inode_add_bytes(struct inode *inode, loff_t bytes);
2394 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2395 loff_t inode_get_bytes(struct inode *inode);
2396 void inode_set_bytes(struct inode *inode, loff_t bytes);
2398 extern int vfs_readdir(struct file *, filldir_t, void *);
2400 extern int vfs_stat(const char __user *, struct kstat *);
2401 extern int vfs_lstat(const char __user *, struct kstat *);
2402 extern int vfs_fstat(unsigned int, struct kstat *);
2403 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2405 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2407 extern int __generic_block_fiemap(struct inode *inode,
2408 struct fiemap_extent_info *fieinfo,
2409 loff_t start, loff_t len,
2410 get_block_t *get_block);
2411 extern int generic_block_fiemap(struct inode *inode,
2412 struct fiemap_extent_info *fieinfo, u64 start,
2413 u64 len, get_block_t *get_block);
2415 extern void get_filesystem(struct file_system_type *fs);
2416 extern void put_filesystem(struct file_system_type *fs);
2417 extern struct file_system_type *get_fs_type(const char *name);
2418 extern struct super_block *get_super(struct block_device *);
2419 extern struct super_block *get_active_super(struct block_device *bdev);
2420 extern struct super_block *user_get_super(dev_t);
2421 extern void drop_super(struct super_block *sb);
2422 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2424 extern int dcache_dir_open(struct inode *, struct file *);
2425 extern int dcache_dir_close(struct inode *, struct file *);
2426 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2427 extern int dcache_readdir(struct file *, void *, filldir_t);
2428 extern int simple_setattr(struct dentry *, struct iattr *);
2429 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2430 extern int simple_statfs(struct dentry *, struct kstatfs *);
2431 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2432 extern int simple_unlink(struct inode *, struct dentry *);
2433 extern int simple_rmdir(struct inode *, struct dentry *);
2434 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2435 extern int noop_fsync(struct file *, int);
2436 extern int simple_empty(struct dentry *);
2437 extern int simple_readpage(struct file *file, struct page *page);
2438 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2439 loff_t pos, unsigned len, unsigned flags,
2440 struct page **pagep, void **fsdata);
2441 extern int simple_write_end(struct file *file, struct address_space *mapping,
2442 loff_t pos, unsigned len, unsigned copied,
2443 struct page *page, void *fsdata);
2445 extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
2446 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2447 extern const struct file_operations simple_dir_operations;
2448 extern const struct inode_operations simple_dir_inode_operations;
2449 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2450 struct dentry *d_alloc_name(struct dentry *, const char *);
2451 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2452 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2453 extern void simple_release_fs(struct vfsmount **mount, int *count);
2455 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2456 loff_t *ppos, const void *from, size_t available);
2457 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2458 const void __user *from, size_t count);
2460 extern int generic_file_fsync(struct file *, int);
2462 extern int generic_check_addressable(unsigned, u64);
2464 #ifdef CONFIG_MIGRATION
2465 extern int buffer_migrate_page(struct address_space *,
2466 struct page *, struct page *);
2468 #define buffer_migrate_page NULL
2471 extern int inode_change_ok(const struct inode *, struct iattr *);
2472 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2473 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2475 extern void file_update_time(struct file *file);
2477 extern int generic_show_options(struct seq_file *m, struct vfsmount *mnt);
2478 extern void save_mount_options(struct super_block *sb, char *options);
2479 extern void replace_mount_options(struct super_block *sb, char *options);
2481 static inline ino_t parent_ino(struct dentry *dentry)
2486 * Don't strictly need d_lock here? If the parent ino could change
2487 * then surely we'd have a deeper race in the caller?
2489 spin_lock(&dentry->d_lock);
2490 res = dentry->d_parent->d_inode->i_ino;
2491 spin_unlock(&dentry->d_lock);
2495 /* Transaction based IO helpers */
2498 * An argresp is stored in an allocated page and holds the
2499 * size of the argument or response, along with its content
2501 struct simple_transaction_argresp {
2506 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2508 char *simple_transaction_get(struct file *file, const char __user *buf,
2510 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2511 size_t size, loff_t *pos);
2512 int simple_transaction_release(struct inode *inode, struct file *file);
2514 void simple_transaction_set(struct file *file, size_t n);
2517 * simple attribute files
2519 * These attributes behave similar to those in sysfs:
2521 * Writing to an attribute immediately sets a value, an open file can be
2522 * written to multiple times.
2524 * Reading from an attribute creates a buffer from the value that might get
2525 * read with multiple read calls. When the attribute has been read
2526 * completely, no further read calls are possible until the file is opened
2529 * All attributes contain a text representation of a numeric value
2530 * that are accessed with the get() and set() functions.
2532 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2533 static int __fops ## _open(struct inode *inode, struct file *file) \
2535 __simple_attr_check_format(__fmt, 0ull); \
2536 return simple_attr_open(inode, file, __get, __set, __fmt); \
2538 static const struct file_operations __fops = { \
2539 .owner = THIS_MODULE, \
2540 .open = __fops ## _open, \
2541 .release = simple_attr_release, \
2542 .read = simple_attr_read, \
2543 .write = simple_attr_write, \
2544 .llseek = generic_file_llseek, \
2547 static inline void __attribute__((format(printf, 1, 2)))
2548 __simple_attr_check_format(const char *fmt, ...)
2550 /* don't do anything, just let the compiler check the arguments; */
2553 int simple_attr_open(struct inode *inode, struct file *file,
2554 int (*get)(void *, u64 *), int (*set)(void *, u64),
2556 int simple_attr_release(struct inode *inode, struct file *file);
2557 ssize_t simple_attr_read(struct file *file, char __user *buf,
2558 size_t len, loff_t *ppos);
2559 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2560 size_t len, loff_t *ppos);
2563 int proc_nr_files(struct ctl_table *table, int write,
2564 void __user *buffer, size_t *lenp, loff_t *ppos);
2565 int proc_nr_dentry(struct ctl_table *table, int write,
2566 void __user *buffer, size_t *lenp, loff_t *ppos);
2567 int proc_nr_inodes(struct ctl_table *table, int write,
2568 void __user *buffer, size_t *lenp, loff_t *ppos);
2569 int __init get_filesystem_list(char *buf);
2571 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
2572 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
2574 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2575 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2576 (flag & __FMODE_NONOTIFY)))
2578 #endif /* __KERNEL__ */
2579 #endif /* _LINUX_FS_H */