]> git.proxmox.com Git - mirror_ubuntu-jammy-kernel.git/blob - include/linux/fs.h
move exec_permission() up to the rest of permission-related functions
[mirror_ubuntu-jammy-kernel.git] / include / linux / fs.h
1 #ifndef _LINUX_FS_H
2 #define _LINUX_FS_H
3
4 /*
5 * This file has definitions for some important file table
6 * structures etc.
7 */
8
9 #include <linux/limits.h>
10 #include <linux/ioctl.h>
11 #include <linux/blk_types.h>
12 #include <linux/types.h>
13
14 /*
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.
19 *
20 * Some programs (notably those using select()) may have to be
21 * recompiled to take full advantage of the new limits..
22 */
23
24 /* Fixed constants first: */
25 #undef NR_OPEN
26 #define INR_OPEN_CUR 1024 /* Initial setting for nfile rlimits */
27 #define INR_OPEN_MAX 4096 /* Hard limit for nfile rlimits */
28
29 #define BLOCK_SIZE_BITS 10
30 #define BLOCK_SIZE (1<<BLOCK_SIZE_BITS)
31
32 #define SEEK_SET 0 /* seek relative to beginning of file */
33 #define SEEK_CUR 1 /* seek relative to current file position */
34 #define SEEK_END 2 /* seek relative to end of file */
35 #define SEEK_MAX SEEK_END
36
37 struct fstrim_range {
38 __u64 start;
39 __u64 len;
40 __u64 minlen;
41 };
42
43 /* And dynamically-tunable limits and defaults: */
44 struct files_stat_struct {
45 unsigned long nr_files; /* read only */
46 unsigned long nr_free_files; /* read only */
47 unsigned long max_files; /* tunable */
48 };
49
50 struct inodes_stat_t {
51 int nr_inodes;
52 int nr_unused;
53 int dummy[5]; /* padding for sysctl ABI compatibility */
54 };
55
56
57 #define NR_FILE 8192 /* this can well be larger on a larger system */
58
59 #define MAY_EXEC 1
60 #define MAY_WRITE 2
61 #define MAY_READ 4
62 #define MAY_APPEND 8
63 #define MAY_ACCESS 16
64 #define MAY_OPEN 32
65 #define MAY_CHDIR 64
66
67 /*
68 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
69 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
70 */
71
72 /* file is open for reading */
73 #define FMODE_READ ((__force fmode_t)0x1)
74 /* file is open for writing */
75 #define FMODE_WRITE ((__force fmode_t)0x2)
76 /* file is seekable */
77 #define FMODE_LSEEK ((__force fmode_t)0x4)
78 /* file can be accessed using pread */
79 #define FMODE_PREAD ((__force fmode_t)0x8)
80 /* file can be accessed using pwrite */
81 #define FMODE_PWRITE ((__force fmode_t)0x10)
82 /* File is opened for execution with sys_execve / sys_uselib */
83 #define FMODE_EXEC ((__force fmode_t)0x20)
84 /* File is opened with O_NDELAY (only set for block devices) */
85 #define FMODE_NDELAY ((__force fmode_t)0x40)
86 /* File is opened with O_EXCL (only set for block devices) */
87 #define FMODE_EXCL ((__force fmode_t)0x80)
88 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
89 (specialy hack for floppy.c) */
90 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
91
92 /*
93 * Don't update ctime and mtime.
94 *
95 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
96 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
97 */
98 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
99
100 /* Expect random access pattern */
101 #define FMODE_RANDOM ((__force fmode_t)0x1000)
102
103 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
104 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
105
106 /* File is opened with O_PATH; almost nothing can be done with it */
107 #define FMODE_PATH ((__force fmode_t)0x4000)
108
109 /* File was opened by fanotify and shouldn't generate fanotify events */
110 #define FMODE_NONOTIFY ((__force fmode_t)0x1000000)
111
112 /*
113 * The below are the various read and write types that we support. Some of
114 * them include behavioral modifiers that send information down to the
115 * block layer and IO scheduler. Terminology:
116 *
117 * The block layer uses device plugging to defer IO a little bit, in
118 * the hope that we will see more IO very shortly. This increases
119 * coalescing of adjacent IO and thus reduces the number of IOs we
120 * have to send to the device. It also allows for better queuing,
121 * if the IO isn't mergeable. If the caller is going to be waiting
122 * for the IO, then he must ensure that the device is unplugged so
123 * that the IO is dispatched to the driver.
124 *
125 * All IO is handled async in Linux. This is fine for background
126 * writes, but for reads or writes that someone waits for completion
127 * on, we want to notify the block layer and IO scheduler so that they
128 * know about it. That allows them to make better scheduling
129 * decisions. So when the below references 'sync' and 'async', it
130 * is referencing this priority hint.
131 *
132 * With that in mind, the available types are:
133 *
134 * READ A normal read operation. Device will be plugged.
135 * READ_SYNC A synchronous read. Device is not plugged, caller can
136 * immediately wait on this read without caring about
137 * unplugging.
138 * READA Used for read-ahead operations. Lower priority, and the
139 * block layer could (in theory) choose to ignore this
140 * request if it runs into resource problems.
141 * WRITE A normal async write. Device will be plugged.
142 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
143 * the hint that someone will be waiting on this IO
144 * shortly. The write equivalent of READ_SYNC.
145 * WRITE_ODIRECT Special case write for O_DIRECT only.
146 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
147 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
148 * non-volatile media on completion.
149 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
150 * by a cache flush and data is guaranteed to be on
151 * non-volatile media on completion.
152 *
153 */
154 #define RW_MASK REQ_WRITE
155 #define RWA_MASK REQ_RAHEAD
156
157 #define READ 0
158 #define WRITE RW_MASK
159 #define READA RWA_MASK
160
161 #define READ_SYNC (READ | REQ_SYNC)
162 #define READ_META (READ | REQ_META)
163 #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
164 #define WRITE_ODIRECT (WRITE | REQ_SYNC)
165 #define WRITE_META (WRITE | REQ_META)
166 #define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
167 #define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
168 #define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
169
170 #define SEL_IN 1
171 #define SEL_OUT 2
172 #define SEL_EX 4
173
174 /* public flags for file_system_type */
175 #define FS_REQUIRES_DEV 1
176 #define FS_BINARY_MOUNTDATA 2
177 #define FS_HAS_SUBTYPE 4
178 #define FS_REVAL_DOT 16384 /* Check the paths ".", ".." for staleness */
179 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move()
180 * during rename() internally.
181 */
182
183 /*
184 * These are the fs-independent mount-flags: up to 32 flags are supported
185 */
186 #define MS_RDONLY 1 /* Mount read-only */
187 #define MS_NOSUID 2 /* Ignore suid and sgid bits */
188 #define MS_NODEV 4 /* Disallow access to device special files */
189 #define MS_NOEXEC 8 /* Disallow program execution */
190 #define MS_SYNCHRONOUS 16 /* Writes are synced at once */
191 #define MS_REMOUNT 32 /* Alter flags of a mounted FS */
192 #define MS_MANDLOCK 64 /* Allow mandatory locks on an FS */
193 #define MS_DIRSYNC 128 /* Directory modifications are synchronous */
194 #define MS_NOATIME 1024 /* Do not update access times. */
195 #define MS_NODIRATIME 2048 /* Do not update directory access times */
196 #define MS_BIND 4096
197 #define MS_MOVE 8192
198 #define MS_REC 16384
199 #define MS_VERBOSE 32768 /* War is peace. Verbosity is silence.
200 MS_VERBOSE is deprecated. */
201 #define MS_SILENT 32768
202 #define MS_POSIXACL (1<<16) /* VFS does not apply the umask */
203 #define MS_UNBINDABLE (1<<17) /* change to unbindable */
204 #define MS_PRIVATE (1<<18) /* change to private */
205 #define MS_SLAVE (1<<19) /* change to slave */
206 #define MS_SHARED (1<<20) /* change to shared */
207 #define MS_RELATIME (1<<21) /* Update atime relative to mtime/ctime. */
208 #define MS_KERNMOUNT (1<<22) /* this is a kern_mount call */
209 #define MS_I_VERSION (1<<23) /* Update inode I_version field */
210 #define MS_STRICTATIME (1<<24) /* Always perform atime updates */
211 #define MS_NOSEC (1<<28)
212 #define MS_BORN (1<<29)
213 #define MS_ACTIVE (1<<30)
214 #define MS_NOUSER (1<<31)
215
216 /*
217 * Superblock flags that can be altered by MS_REMOUNT
218 */
219 #define MS_RMT_MASK (MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK|MS_I_VERSION)
220
221 /*
222 * Old magic mount flag and mask
223 */
224 #define MS_MGC_VAL 0xC0ED0000
225 #define MS_MGC_MSK 0xffff0000
226
227 /* Inode flags - they have nothing to superblock flags now */
228
229 #define S_SYNC 1 /* Writes are synced at once */
230 #define S_NOATIME 2 /* Do not update access times */
231 #define S_APPEND 4 /* Append-only file */
232 #define S_IMMUTABLE 8 /* Immutable file */
233 #define S_DEAD 16 /* removed, but still open directory */
234 #define S_NOQUOTA 32 /* Inode is not counted to quota */
235 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
236 #define S_NOCMTIME 128 /* Do not update file c/mtime */
237 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
238 #define S_PRIVATE 512 /* Inode is fs-internal */
239 #define S_IMA 1024 /* Inode has an associated IMA struct */
240 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
241 #define S_NOSEC 4096 /* no suid or xattr security attributes */
242
243 /*
244 * Note that nosuid etc flags are inode-specific: setting some file-system
245 * flags just means all the inodes inherit those flags by default. It might be
246 * possible to override it selectively if you really wanted to with some
247 * ioctl() that is not currently implemented.
248 *
249 * Exception: MS_RDONLY is always applied to the entire file system.
250 *
251 * Unfortunately, it is possible to change a filesystems flags with it mounted
252 * with files in use. This means that all of the inodes will not have their
253 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
254 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
255 */
256 #define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))
257
258 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
259 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
260 ((inode)->i_flags & S_SYNC))
261 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
262 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
263 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
264 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
265 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
266
267 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
268 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
269 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
270 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
271
272 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
273 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
274 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
275 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
276 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
277 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
278 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
279
280 /* the read-only stuff doesn't really belong here, but any other place is
281 probably as bad and I don't want to create yet another include file. */
282
283 #define BLKROSET _IO(0x12,93) /* set device read-only (0 = read-write) */
284 #define BLKROGET _IO(0x12,94) /* get read-only status (0 = read_write) */
285 #define BLKRRPART _IO(0x12,95) /* re-read partition table */
286 #define BLKGETSIZE _IO(0x12,96) /* return device size /512 (long *arg) */
287 #define BLKFLSBUF _IO(0x12,97) /* flush buffer cache */
288 #define BLKRASET _IO(0x12,98) /* set read ahead for block device */
289 #define BLKRAGET _IO(0x12,99) /* get current read ahead setting */
290 #define BLKFRASET _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
291 #define BLKFRAGET _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
292 #define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
293 #define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
294 #define BLKSSZGET _IO(0x12,104)/* get block device sector size */
295 #if 0
296 #define BLKPG _IO(0x12,105)/* See blkpg.h */
297
298 /* Some people are morons. Do not use sizeof! */
299
300 #define BLKELVGET _IOR(0x12,106,size_t)/* elevator get */
301 #define BLKELVSET _IOW(0x12,107,size_t)/* elevator set */
302 /* This was here just to show that the number is taken -
303 probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
304 #endif
305 /* A jump here: 108-111 have been used for various private purposes. */
306 #define BLKBSZGET _IOR(0x12,112,size_t)
307 #define BLKBSZSET _IOW(0x12,113,size_t)
308 #define BLKGETSIZE64 _IOR(0x12,114,size_t) /* return device size in bytes (u64 *arg) */
309 #define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup)
310 #define BLKTRACESTART _IO(0x12,116)
311 #define BLKTRACESTOP _IO(0x12,117)
312 #define BLKTRACETEARDOWN _IO(0x12,118)
313 #define BLKDISCARD _IO(0x12,119)
314 #define BLKIOMIN _IO(0x12,120)
315 #define BLKIOOPT _IO(0x12,121)
316 #define BLKALIGNOFF _IO(0x12,122)
317 #define BLKPBSZGET _IO(0x12,123)
318 #define BLKDISCARDZEROES _IO(0x12,124)
319 #define BLKSECDISCARD _IO(0x12,125)
320
321 #define BMAP_IOCTL 1 /* obsolete - kept for compatibility */
322 #define FIBMAP _IO(0x00,1) /* bmap access */
323 #define FIGETBSZ _IO(0x00,2) /* get the block size used for bmap */
324 #define FIFREEZE _IOWR('X', 119, int) /* Freeze */
325 #define FITHAW _IOWR('X', 120, int) /* Thaw */
326 #define FITRIM _IOWR('X', 121, struct fstrim_range) /* Trim */
327
328 #define FS_IOC_GETFLAGS _IOR('f', 1, long)
329 #define FS_IOC_SETFLAGS _IOW('f', 2, long)
330 #define FS_IOC_GETVERSION _IOR('v', 1, long)
331 #define FS_IOC_SETVERSION _IOW('v', 2, long)
332 #define FS_IOC_FIEMAP _IOWR('f', 11, struct fiemap)
333 #define FS_IOC32_GETFLAGS _IOR('f', 1, int)
334 #define FS_IOC32_SETFLAGS _IOW('f', 2, int)
335 #define FS_IOC32_GETVERSION _IOR('v', 1, int)
336 #define FS_IOC32_SETVERSION _IOW('v', 2, int)
337
338 /*
339 * Inode flags (FS_IOC_GETFLAGS / FS_IOC_SETFLAGS)
340 */
341 #define FS_SECRM_FL 0x00000001 /* Secure deletion */
342 #define FS_UNRM_FL 0x00000002 /* Undelete */
343 #define FS_COMPR_FL 0x00000004 /* Compress file */
344 #define FS_SYNC_FL 0x00000008 /* Synchronous updates */
345 #define FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
346 #define FS_APPEND_FL 0x00000020 /* writes to file may only append */
347 #define FS_NODUMP_FL 0x00000040 /* do not dump file */
348 #define FS_NOATIME_FL 0x00000080 /* do not update atime */
349 /* Reserved for compression usage... */
350 #define FS_DIRTY_FL 0x00000100
351 #define FS_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
352 #define FS_NOCOMP_FL 0x00000400 /* Don't compress */
353 #define FS_ECOMPR_FL 0x00000800 /* Compression error */
354 /* End compression flags --- maybe not all used */
355 #define FS_BTREE_FL 0x00001000 /* btree format dir */
356 #define FS_INDEX_FL 0x00001000 /* hash-indexed directory */
357 #define FS_IMAGIC_FL 0x00002000 /* AFS directory */
358 #define FS_JOURNAL_DATA_FL 0x00004000 /* Reserved for ext3 */
359 #define FS_NOTAIL_FL 0x00008000 /* file tail should not be merged */
360 #define FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
361 #define FS_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
362 #define FS_EXTENT_FL 0x00080000 /* Extents */
363 #define FS_DIRECTIO_FL 0x00100000 /* Use direct i/o */
364 #define FS_NOCOW_FL 0x00800000 /* Do not cow file */
365 #define FS_RESERVED_FL 0x80000000 /* reserved for ext2 lib */
366
367 #define FS_FL_USER_VISIBLE 0x0003DFFF /* User visible flags */
368 #define FS_FL_USER_MODIFIABLE 0x000380FF /* User modifiable flags */
369
370
371 #define SYNC_FILE_RANGE_WAIT_BEFORE 1
372 #define SYNC_FILE_RANGE_WRITE 2
373 #define SYNC_FILE_RANGE_WAIT_AFTER 4
374
375 #ifdef __KERNEL__
376
377 #include <linux/linkage.h>
378 #include <linux/wait.h>
379 #include <linux/types.h>
380 #include <linux/kdev_t.h>
381 #include <linux/dcache.h>
382 #include <linux/path.h>
383 #include <linux/stat.h>
384 #include <linux/cache.h>
385 #include <linux/list.h>
386 #include <linux/radix-tree.h>
387 #include <linux/prio_tree.h>
388 #include <linux/init.h>
389 #include <linux/pid.h>
390 #include <linux/mutex.h>
391 #include <linux/capability.h>
392 #include <linux/semaphore.h>
393 #include <linux/fiemap.h>
394 #include <linux/rculist_bl.h>
395
396 #include <asm/atomic.h>
397 #include <asm/byteorder.h>
398
399 struct export_operations;
400 struct hd_geometry;
401 struct iovec;
402 struct nameidata;
403 struct kiocb;
404 struct kobject;
405 struct pipe_inode_info;
406 struct poll_table_struct;
407 struct kstatfs;
408 struct vm_area_struct;
409 struct vfsmount;
410 struct cred;
411
412 extern void __init inode_init(void);
413 extern void __init inode_init_early(void);
414 extern void __init files_init(unsigned long);
415
416 extern struct files_stat_struct files_stat;
417 extern unsigned long get_max_files(void);
418 extern int sysctl_nr_open;
419 extern struct inodes_stat_t inodes_stat;
420 extern int leases_enable, lease_break_time;
421
422 struct buffer_head;
423 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
424 struct buffer_head *bh_result, int create);
425 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
426 ssize_t bytes, void *private, int ret,
427 bool is_async);
428
429 /*
430 * Attribute flags. These should be or-ed together to figure out what
431 * has been changed!
432 */
433 #define ATTR_MODE (1 << 0)
434 #define ATTR_UID (1 << 1)
435 #define ATTR_GID (1 << 2)
436 #define ATTR_SIZE (1 << 3)
437 #define ATTR_ATIME (1 << 4)
438 #define ATTR_MTIME (1 << 5)
439 #define ATTR_CTIME (1 << 6)
440 #define ATTR_ATIME_SET (1 << 7)
441 #define ATTR_MTIME_SET (1 << 8)
442 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
443 #define ATTR_ATTR_FLAG (1 << 10)
444 #define ATTR_KILL_SUID (1 << 11)
445 #define ATTR_KILL_SGID (1 << 12)
446 #define ATTR_FILE (1 << 13)
447 #define ATTR_KILL_PRIV (1 << 14)
448 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
449 #define ATTR_TIMES_SET (1 << 16)
450
451 /*
452 * This is the Inode Attributes structure, used for notify_change(). It
453 * uses the above definitions as flags, to know which values have changed.
454 * Also, in this manner, a Filesystem can look at only the values it cares
455 * about. Basically, these are the attributes that the VFS layer can
456 * request to change from the FS layer.
457 *
458 * Derek Atkins <warlord@MIT.EDU> 94-10-20
459 */
460 struct iattr {
461 unsigned int ia_valid;
462 umode_t ia_mode;
463 uid_t ia_uid;
464 gid_t ia_gid;
465 loff_t ia_size;
466 struct timespec ia_atime;
467 struct timespec ia_mtime;
468 struct timespec ia_ctime;
469
470 /*
471 * Not an attribute, but an auxiliary info for filesystems wanting to
472 * implement an ftruncate() like method. NOTE: filesystem should
473 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
474 */
475 struct file *ia_file;
476 };
477
478 /*
479 * Includes for diskquotas.
480 */
481 #include <linux/quota.h>
482
483 /**
484 * enum positive_aop_returns - aop return codes with specific semantics
485 *
486 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
487 * completed, that the page is still locked, and
488 * should be considered active. The VM uses this hint
489 * to return the page to the active list -- it won't
490 * be a candidate for writeback again in the near
491 * future. Other callers must be careful to unlock
492 * the page if they get this return. Returned by
493 * writepage();
494 *
495 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
496 * unlocked it and the page might have been truncated.
497 * The caller should back up to acquiring a new page and
498 * trying again. The aop will be taking reasonable
499 * precautions not to livelock. If the caller held a page
500 * reference, it should drop it before retrying. Returned
501 * by readpage().
502 *
503 * address_space_operation functions return these large constants to indicate
504 * special semantics to the caller. These are much larger than the bytes in a
505 * page to allow for functions that return the number of bytes operated on in a
506 * given page.
507 */
508
509 enum positive_aop_returns {
510 AOP_WRITEPAGE_ACTIVATE = 0x80000,
511 AOP_TRUNCATED_PAGE = 0x80001,
512 };
513
514 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
515 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
516 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
517 * helper code (eg buffer layer)
518 * to clear GFP_FS from alloc */
519
520 /*
521 * oh the beauties of C type declarations.
522 */
523 struct page;
524 struct address_space;
525 struct writeback_control;
526
527 struct iov_iter {
528 const struct iovec *iov;
529 unsigned long nr_segs;
530 size_t iov_offset;
531 size_t count;
532 };
533
534 size_t iov_iter_copy_from_user_atomic(struct page *page,
535 struct iov_iter *i, unsigned long offset, size_t bytes);
536 size_t iov_iter_copy_from_user(struct page *page,
537 struct iov_iter *i, unsigned long offset, size_t bytes);
538 void iov_iter_advance(struct iov_iter *i, size_t bytes);
539 int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes);
540 size_t iov_iter_single_seg_count(struct iov_iter *i);
541
542 static inline void iov_iter_init(struct iov_iter *i,
543 const struct iovec *iov, unsigned long nr_segs,
544 size_t count, size_t written)
545 {
546 i->iov = iov;
547 i->nr_segs = nr_segs;
548 i->iov_offset = 0;
549 i->count = count + written;
550
551 iov_iter_advance(i, written);
552 }
553
554 static inline size_t iov_iter_count(struct iov_iter *i)
555 {
556 return i->count;
557 }
558
559 /*
560 * "descriptor" for what we're up to with a read.
561 * This allows us to use the same read code yet
562 * have multiple different users of the data that
563 * we read from a file.
564 *
565 * The simplest case just copies the data to user
566 * mode.
567 */
568 typedef struct {
569 size_t written;
570 size_t count;
571 union {
572 char __user *buf;
573 void *data;
574 } arg;
575 int error;
576 } read_descriptor_t;
577
578 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
579 unsigned long, unsigned long);
580
581 struct address_space_operations {
582 int (*writepage)(struct page *page, struct writeback_control *wbc);
583 int (*readpage)(struct file *, struct page *);
584
585 /* Write back some dirty pages from this mapping. */
586 int (*writepages)(struct address_space *, struct writeback_control *);
587
588 /* Set a page dirty. Return true if this dirtied it */
589 int (*set_page_dirty)(struct page *page);
590
591 int (*readpages)(struct file *filp, struct address_space *mapping,
592 struct list_head *pages, unsigned nr_pages);
593
594 int (*write_begin)(struct file *, struct address_space *mapping,
595 loff_t pos, unsigned len, unsigned flags,
596 struct page **pagep, void **fsdata);
597 int (*write_end)(struct file *, struct address_space *mapping,
598 loff_t pos, unsigned len, unsigned copied,
599 struct page *page, void *fsdata);
600
601 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
602 sector_t (*bmap)(struct address_space *, sector_t);
603 void (*invalidatepage) (struct page *, unsigned long);
604 int (*releasepage) (struct page *, gfp_t);
605 void (*freepage)(struct page *);
606 ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
607 loff_t offset, unsigned long nr_segs);
608 int (*get_xip_mem)(struct address_space *, pgoff_t, int,
609 void **, unsigned long *);
610 /* migrate the contents of a page to the specified target */
611 int (*migratepage) (struct address_space *,
612 struct page *, struct page *);
613 int (*launder_page) (struct page *);
614 int (*is_partially_uptodate) (struct page *, read_descriptor_t *,
615 unsigned long);
616 int (*error_remove_page)(struct address_space *, struct page *);
617 };
618
619 extern const struct address_space_operations empty_aops;
620
621 /*
622 * pagecache_write_begin/pagecache_write_end must be used by general code
623 * to write into the pagecache.
624 */
625 int pagecache_write_begin(struct file *, struct address_space *mapping,
626 loff_t pos, unsigned len, unsigned flags,
627 struct page **pagep, void **fsdata);
628
629 int pagecache_write_end(struct file *, struct address_space *mapping,
630 loff_t pos, unsigned len, unsigned copied,
631 struct page *page, void *fsdata);
632
633 struct backing_dev_info;
634 struct address_space {
635 struct inode *host; /* owner: inode, block_device */
636 struct radix_tree_root page_tree; /* radix tree of all pages */
637 spinlock_t tree_lock; /* and lock protecting it */
638 unsigned int i_mmap_writable;/* count VM_SHARED mappings */
639 struct prio_tree_root i_mmap; /* tree of private and shared mappings */
640 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
641 struct mutex i_mmap_mutex; /* protect tree, count, list */
642 /* Protected by tree_lock together with the radix tree */
643 unsigned long nrpages; /* number of total pages */
644 pgoff_t writeback_index;/* writeback starts here */
645 const struct address_space_operations *a_ops; /* methods */
646 unsigned long flags; /* error bits/gfp mask */
647 struct backing_dev_info *backing_dev_info; /* device readahead, etc */
648 spinlock_t private_lock; /* for use by the address_space */
649 struct list_head private_list; /* ditto */
650 struct address_space *assoc_mapping; /* ditto */
651 } __attribute__((aligned(sizeof(long))));
652 /*
653 * On most architectures that alignment is already the case; but
654 * must be enforced here for CRIS, to let the least significant bit
655 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
656 */
657
658 struct block_device {
659 dev_t bd_dev; /* not a kdev_t - it's a search key */
660 int bd_openers;
661 struct inode * bd_inode; /* will die */
662 struct super_block * bd_super;
663 struct mutex bd_mutex; /* open/close mutex */
664 struct list_head bd_inodes;
665 void * bd_claiming;
666 void * bd_holder;
667 int bd_holders;
668 bool bd_write_holder;
669 #ifdef CONFIG_SYSFS
670 struct list_head bd_holder_disks;
671 #endif
672 struct block_device * bd_contains;
673 unsigned bd_block_size;
674 struct hd_struct * bd_part;
675 /* number of times partitions within this device have been opened. */
676 unsigned bd_part_count;
677 int bd_invalidated;
678 struct gendisk * bd_disk;
679 struct list_head bd_list;
680 /*
681 * Private data. You must have bd_claim'ed the block_device
682 * to use this. NOTE: bd_claim allows an owner to claim
683 * the same device multiple times, the owner must take special
684 * care to not mess up bd_private for that case.
685 */
686 unsigned long bd_private;
687
688 /* The counter of freeze processes */
689 int bd_fsfreeze_count;
690 /* Mutex for freeze */
691 struct mutex bd_fsfreeze_mutex;
692 };
693
694 /*
695 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
696 * radix trees
697 */
698 #define PAGECACHE_TAG_DIRTY 0
699 #define PAGECACHE_TAG_WRITEBACK 1
700 #define PAGECACHE_TAG_TOWRITE 2
701
702 int mapping_tagged(struct address_space *mapping, int tag);
703
704 /*
705 * Might pages of this file be mapped into userspace?
706 */
707 static inline int mapping_mapped(struct address_space *mapping)
708 {
709 return !prio_tree_empty(&mapping->i_mmap) ||
710 !list_empty(&mapping->i_mmap_nonlinear);
711 }
712
713 /*
714 * Might pages of this file have been modified in userspace?
715 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
716 * marks vma as VM_SHARED if it is shared, and the file was opened for
717 * writing i.e. vma may be mprotected writable even if now readonly.
718 */
719 static inline int mapping_writably_mapped(struct address_space *mapping)
720 {
721 return mapping->i_mmap_writable != 0;
722 }
723
724 /*
725 * Use sequence counter to get consistent i_size on 32-bit processors.
726 */
727 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
728 #include <linux/seqlock.h>
729 #define __NEED_I_SIZE_ORDERED
730 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
731 #else
732 #define i_size_ordered_init(inode) do { } while (0)
733 #endif
734
735 struct posix_acl;
736 #define ACL_NOT_CACHED ((void *)(-1))
737
738 struct inode {
739 /* RCU path lookup touches following: */
740 umode_t i_mode;
741 uid_t i_uid;
742 gid_t i_gid;
743 const struct inode_operations *i_op;
744 struct super_block *i_sb;
745
746 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
747 unsigned int i_flags;
748 unsigned long i_state;
749 #ifdef CONFIG_SECURITY
750 void *i_security;
751 #endif
752 struct mutex i_mutex;
753
754
755 unsigned long dirtied_when; /* jiffies of first dirtying */
756
757 struct hlist_node i_hash;
758 struct list_head i_wb_list; /* backing dev IO list */
759 struct list_head i_lru; /* inode LRU list */
760 struct list_head i_sb_list;
761 union {
762 struct list_head i_dentry;
763 struct rcu_head i_rcu;
764 };
765 unsigned long i_ino;
766 atomic_t i_count;
767 unsigned int i_nlink;
768 dev_t i_rdev;
769 unsigned int i_blkbits;
770 u64 i_version;
771 loff_t i_size;
772 #ifdef __NEED_I_SIZE_ORDERED
773 seqcount_t i_size_seqcount;
774 #endif
775 struct timespec i_atime;
776 struct timespec i_mtime;
777 struct timespec i_ctime;
778 blkcnt_t i_blocks;
779 unsigned short i_bytes;
780 struct rw_semaphore i_alloc_sem;
781 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
782 struct file_lock *i_flock;
783 struct address_space *i_mapping;
784 struct address_space i_data;
785 #ifdef CONFIG_QUOTA
786 struct dquot *i_dquot[MAXQUOTAS];
787 #endif
788 struct list_head i_devices;
789 union {
790 struct pipe_inode_info *i_pipe;
791 struct block_device *i_bdev;
792 struct cdev *i_cdev;
793 };
794
795 __u32 i_generation;
796
797 #ifdef CONFIG_FSNOTIFY
798 __u32 i_fsnotify_mask; /* all events this inode cares about */
799 struct hlist_head i_fsnotify_marks;
800 #endif
801
802 #ifdef CONFIG_IMA
803 atomic_t i_readcount; /* struct files open RO */
804 #endif
805 atomic_t i_writecount;
806 #ifdef CONFIG_FS_POSIX_ACL
807 struct posix_acl *i_acl;
808 struct posix_acl *i_default_acl;
809 #endif
810 void *i_private; /* fs or device private pointer */
811 };
812
813 static inline int inode_unhashed(struct inode *inode)
814 {
815 return hlist_unhashed(&inode->i_hash);
816 }
817
818 /*
819 * inode->i_mutex nesting subclasses for the lock validator:
820 *
821 * 0: the object of the current VFS operation
822 * 1: parent
823 * 2: child/target
824 * 3: quota file
825 *
826 * The locking order between these classes is
827 * parent -> child -> normal -> xattr -> quota
828 */
829 enum inode_i_mutex_lock_class
830 {
831 I_MUTEX_NORMAL,
832 I_MUTEX_PARENT,
833 I_MUTEX_CHILD,
834 I_MUTEX_XATTR,
835 I_MUTEX_QUOTA
836 };
837
838 /*
839 * NOTE: in a 32bit arch with a preemptable kernel and
840 * an UP compile the i_size_read/write must be atomic
841 * with respect to the local cpu (unlike with preempt disabled),
842 * but they don't need to be atomic with respect to other cpus like in
843 * true SMP (so they need either to either locally disable irq around
844 * the read or for example on x86 they can be still implemented as a
845 * cmpxchg8b without the need of the lock prefix). For SMP compiles
846 * and 64bit archs it makes no difference if preempt is enabled or not.
847 */
848 static inline loff_t i_size_read(const struct inode *inode)
849 {
850 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
851 loff_t i_size;
852 unsigned int seq;
853
854 do {
855 seq = read_seqcount_begin(&inode->i_size_seqcount);
856 i_size = inode->i_size;
857 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
858 return i_size;
859 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
860 loff_t i_size;
861
862 preempt_disable();
863 i_size = inode->i_size;
864 preempt_enable();
865 return i_size;
866 #else
867 return inode->i_size;
868 #endif
869 }
870
871 /*
872 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
873 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
874 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
875 */
876 static inline void i_size_write(struct inode *inode, loff_t i_size)
877 {
878 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
879 write_seqcount_begin(&inode->i_size_seqcount);
880 inode->i_size = i_size;
881 write_seqcount_end(&inode->i_size_seqcount);
882 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
883 preempt_disable();
884 inode->i_size = i_size;
885 preempt_enable();
886 #else
887 inode->i_size = i_size;
888 #endif
889 }
890
891 static inline unsigned iminor(const struct inode *inode)
892 {
893 return MINOR(inode->i_rdev);
894 }
895
896 static inline unsigned imajor(const struct inode *inode)
897 {
898 return MAJOR(inode->i_rdev);
899 }
900
901 extern struct block_device *I_BDEV(struct inode *inode);
902
903 struct fown_struct {
904 rwlock_t lock; /* protects pid, uid, euid fields */
905 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
906 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
907 uid_t uid, euid; /* uid/euid of process setting the owner */
908 int signum; /* posix.1b rt signal to be delivered on IO */
909 };
910
911 /*
912 * Track a single file's readahead state
913 */
914 struct file_ra_state {
915 pgoff_t start; /* where readahead started */
916 unsigned int size; /* # of readahead pages */
917 unsigned int async_size; /* do asynchronous readahead when
918 there are only # of pages ahead */
919
920 unsigned int ra_pages; /* Maximum readahead window */
921 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
922 loff_t prev_pos; /* Cache last read() position */
923 };
924
925 /*
926 * Check if @index falls in the readahead windows.
927 */
928 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
929 {
930 return (index >= ra->start &&
931 index < ra->start + ra->size);
932 }
933
934 #define FILE_MNT_WRITE_TAKEN 1
935 #define FILE_MNT_WRITE_RELEASED 2
936
937 struct file {
938 /*
939 * fu_list becomes invalid after file_free is called and queued via
940 * fu_rcuhead for RCU freeing
941 */
942 union {
943 struct list_head fu_list;
944 struct rcu_head fu_rcuhead;
945 } f_u;
946 struct path f_path;
947 #define f_dentry f_path.dentry
948 #define f_vfsmnt f_path.mnt
949 const struct file_operations *f_op;
950 spinlock_t f_lock; /* f_ep_links, f_flags, no IRQ */
951 #ifdef CONFIG_SMP
952 int f_sb_list_cpu;
953 #endif
954 atomic_long_t f_count;
955 unsigned int f_flags;
956 fmode_t f_mode;
957 loff_t f_pos;
958 struct fown_struct f_owner;
959 const struct cred *f_cred;
960 struct file_ra_state f_ra;
961
962 u64 f_version;
963 #ifdef CONFIG_SECURITY
964 void *f_security;
965 #endif
966 /* needed for tty driver, and maybe others */
967 void *private_data;
968
969 #ifdef CONFIG_EPOLL
970 /* Used by fs/eventpoll.c to link all the hooks to this file */
971 struct list_head f_ep_links;
972 #endif /* #ifdef CONFIG_EPOLL */
973 struct address_space *f_mapping;
974 #ifdef CONFIG_DEBUG_WRITECOUNT
975 unsigned long f_mnt_write_state;
976 #endif
977 };
978
979 struct file_handle {
980 __u32 handle_bytes;
981 int handle_type;
982 /* file identifier */
983 unsigned char f_handle[0];
984 };
985
986 #define get_file(x) atomic_long_inc(&(x)->f_count)
987 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
988 #define file_count(x) atomic_long_read(&(x)->f_count)
989
990 #ifdef CONFIG_DEBUG_WRITECOUNT
991 static inline void file_take_write(struct file *f)
992 {
993 WARN_ON(f->f_mnt_write_state != 0);
994 f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN;
995 }
996 static inline void file_release_write(struct file *f)
997 {
998 f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED;
999 }
1000 static inline void file_reset_write(struct file *f)
1001 {
1002 f->f_mnt_write_state = 0;
1003 }
1004 static inline void file_check_state(struct file *f)
1005 {
1006 /*
1007 * At this point, either both or neither of these bits
1008 * should be set.
1009 */
1010 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN);
1011 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED);
1012 }
1013 static inline int file_check_writeable(struct file *f)
1014 {
1015 if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN)
1016 return 0;
1017 printk(KERN_WARNING "writeable file with no "
1018 "mnt_want_write()\n");
1019 WARN_ON(1);
1020 return -EINVAL;
1021 }
1022 #else /* !CONFIG_DEBUG_WRITECOUNT */
1023 static inline void file_take_write(struct file *filp) {}
1024 static inline void file_release_write(struct file *filp) {}
1025 static inline void file_reset_write(struct file *filp) {}
1026 static inline void file_check_state(struct file *filp) {}
1027 static inline int file_check_writeable(struct file *filp)
1028 {
1029 return 0;
1030 }
1031 #endif /* CONFIG_DEBUG_WRITECOUNT */
1032
1033 #define MAX_NON_LFS ((1UL<<31) - 1)
1034
1035 /* Page cache limit. The filesystems should put that into their s_maxbytes
1036 limits, otherwise bad things can happen in VM. */
1037 #if BITS_PER_LONG==32
1038 #define MAX_LFS_FILESIZE (((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
1039 #elif BITS_PER_LONG==64
1040 #define MAX_LFS_FILESIZE 0x7fffffffffffffffUL
1041 #endif
1042
1043 #define FL_POSIX 1
1044 #define FL_FLOCK 2
1045 #define FL_ACCESS 8 /* not trying to lock, just looking */
1046 #define FL_EXISTS 16 /* when unlocking, test for existence */
1047 #define FL_LEASE 32 /* lease held on this file */
1048 #define FL_CLOSE 64 /* unlock on close */
1049 #define FL_SLEEP 128 /* A blocking lock */
1050
1051 /*
1052 * Special return value from posix_lock_file() and vfs_lock_file() for
1053 * asynchronous locking.
1054 */
1055 #define FILE_LOCK_DEFERRED 1
1056
1057 /*
1058 * The POSIX file lock owner is determined by
1059 * the "struct files_struct" in the thread group
1060 * (or NULL for no owner - BSD locks).
1061 *
1062 * Lockd stuffs a "host" pointer into this.
1063 */
1064 typedef struct files_struct *fl_owner_t;
1065
1066 struct file_lock_operations {
1067 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1068 void (*fl_release_private)(struct file_lock *);
1069 };
1070
1071 struct lock_manager_operations {
1072 int (*fl_compare_owner)(struct file_lock *, struct file_lock *);
1073 void (*fl_notify)(struct file_lock *); /* unblock callback */
1074 int (*fl_grant)(struct file_lock *, struct file_lock *, int);
1075 void (*fl_release_private)(struct file_lock *);
1076 void (*fl_break)(struct file_lock *);
1077 int (*fl_change)(struct file_lock **, int);
1078 };
1079
1080 struct lock_manager {
1081 struct list_head list;
1082 };
1083
1084 void locks_start_grace(struct lock_manager *);
1085 void locks_end_grace(struct lock_manager *);
1086 int locks_in_grace(void);
1087
1088 /* that will die - we need it for nfs_lock_info */
1089 #include <linux/nfs_fs_i.h>
1090
1091 struct file_lock {
1092 struct file_lock *fl_next; /* singly linked list for this inode */
1093 struct list_head fl_link; /* doubly linked list of all locks */
1094 struct list_head fl_block; /* circular list of blocked processes */
1095 fl_owner_t fl_owner;
1096 unsigned char fl_flags;
1097 unsigned char fl_type;
1098 unsigned int fl_pid;
1099 struct pid *fl_nspid;
1100 wait_queue_head_t fl_wait;
1101 struct file *fl_file;
1102 loff_t fl_start;
1103 loff_t fl_end;
1104
1105 struct fasync_struct * fl_fasync; /* for lease break notifications */
1106 unsigned long fl_break_time; /* for nonblocking lease breaks */
1107
1108 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1109 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1110 union {
1111 struct nfs_lock_info nfs_fl;
1112 struct nfs4_lock_info nfs4_fl;
1113 struct {
1114 struct list_head link; /* link in AFS vnode's pending_locks list */
1115 int state; /* state of grant or error if -ve */
1116 } afs;
1117 } fl_u;
1118 };
1119
1120 /* The following constant reflects the upper bound of the file/locking space */
1121 #ifndef OFFSET_MAX
1122 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1123 #define OFFSET_MAX INT_LIMIT(loff_t)
1124 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1125 #endif
1126
1127 #include <linux/fcntl.h>
1128
1129 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1130
1131 #ifdef CONFIG_FILE_LOCKING
1132 extern int fcntl_getlk(struct file *, struct flock __user *);
1133 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1134 struct flock __user *);
1135
1136 #if BITS_PER_LONG == 32
1137 extern int fcntl_getlk64(struct file *, struct flock64 __user *);
1138 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1139 struct flock64 __user *);
1140 #endif
1141
1142 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1143 extern int fcntl_getlease(struct file *filp);
1144
1145 /* fs/locks.c */
1146 void locks_free_lock(struct file_lock *fl);
1147 extern void locks_init_lock(struct file_lock *);
1148 extern struct file_lock * locks_alloc_lock(void);
1149 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1150 extern void __locks_copy_lock(struct file_lock *, const struct file_lock *);
1151 extern void locks_remove_posix(struct file *, fl_owner_t);
1152 extern void locks_remove_flock(struct file *);
1153 extern void locks_release_private(struct file_lock *);
1154 extern void posix_test_lock(struct file *, struct file_lock *);
1155 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1156 extern int posix_lock_file_wait(struct file *, struct file_lock *);
1157 extern int posix_unblock_lock(struct file *, struct file_lock *);
1158 extern int vfs_test_lock(struct file *, struct file_lock *);
1159 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1160 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1161 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
1162 extern int __break_lease(struct inode *inode, unsigned int flags);
1163 extern void lease_get_mtime(struct inode *, struct timespec *time);
1164 extern int generic_setlease(struct file *, long, struct file_lock **);
1165 extern int vfs_setlease(struct file *, long, struct file_lock **);
1166 extern int lease_modify(struct file_lock **, int);
1167 extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
1168 extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
1169 extern void lock_flocks(void);
1170 extern void unlock_flocks(void);
1171 #else /* !CONFIG_FILE_LOCKING */
1172 static inline int fcntl_getlk(struct file *file, struct flock __user *user)
1173 {
1174 return -EINVAL;
1175 }
1176
1177 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1178 unsigned int cmd, struct flock __user *user)
1179 {
1180 return -EACCES;
1181 }
1182
1183 #if BITS_PER_LONG == 32
1184 static inline int fcntl_getlk64(struct file *file, struct flock64 __user *user)
1185 {
1186 return -EINVAL;
1187 }
1188
1189 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1190 unsigned int cmd, struct flock64 __user *user)
1191 {
1192 return -EACCES;
1193 }
1194 #endif
1195 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1196 {
1197 return 0;
1198 }
1199
1200 static inline int fcntl_getlease(struct file *filp)
1201 {
1202 return 0;
1203 }
1204
1205 static inline void locks_init_lock(struct file_lock *fl)
1206 {
1207 return;
1208 }
1209
1210 static inline void __locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1211 {
1212 return;
1213 }
1214
1215 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1216 {
1217 return;
1218 }
1219
1220 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1221 {
1222 return;
1223 }
1224
1225 static inline void locks_remove_flock(struct file *filp)
1226 {
1227 return;
1228 }
1229
1230 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1231 {
1232 return;
1233 }
1234
1235 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1236 struct file_lock *conflock)
1237 {
1238 return -ENOLCK;
1239 }
1240
1241 static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1242 {
1243 return -ENOLCK;
1244 }
1245
1246 static inline int posix_unblock_lock(struct file *filp,
1247 struct file_lock *waiter)
1248 {
1249 return -ENOENT;
1250 }
1251
1252 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1253 {
1254 return 0;
1255 }
1256
1257 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1258 struct file_lock *fl, struct file_lock *conf)
1259 {
1260 return -ENOLCK;
1261 }
1262
1263 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1264 {
1265 return 0;
1266 }
1267
1268 static inline int flock_lock_file_wait(struct file *filp,
1269 struct file_lock *request)
1270 {
1271 return -ENOLCK;
1272 }
1273
1274 static inline int __break_lease(struct inode *inode, unsigned int mode)
1275 {
1276 return 0;
1277 }
1278
1279 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1280 {
1281 return;
1282 }
1283
1284 static inline int generic_setlease(struct file *filp, long arg,
1285 struct file_lock **flp)
1286 {
1287 return -EINVAL;
1288 }
1289
1290 static inline int vfs_setlease(struct file *filp, long arg,
1291 struct file_lock **lease)
1292 {
1293 return -EINVAL;
1294 }
1295
1296 static inline int lease_modify(struct file_lock **before, int arg)
1297 {
1298 return -EINVAL;
1299 }
1300
1301 static inline int lock_may_read(struct inode *inode, loff_t start,
1302 unsigned long len)
1303 {
1304 return 1;
1305 }
1306
1307 static inline int lock_may_write(struct inode *inode, loff_t start,
1308 unsigned long len)
1309 {
1310 return 1;
1311 }
1312
1313 static inline void lock_flocks(void)
1314 {
1315 }
1316
1317 static inline void unlock_flocks(void)
1318 {
1319 }
1320
1321 #endif /* !CONFIG_FILE_LOCKING */
1322
1323
1324 struct fasync_struct {
1325 spinlock_t fa_lock;
1326 int magic;
1327 int fa_fd;
1328 struct fasync_struct *fa_next; /* singly linked list */
1329 struct file *fa_file;
1330 struct rcu_head fa_rcu;
1331 };
1332
1333 #define FASYNC_MAGIC 0x4601
1334
1335 /* SMP safe fasync helpers: */
1336 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1337 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1338 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1339 extern struct fasync_struct *fasync_alloc(void);
1340 extern void fasync_free(struct fasync_struct *);
1341
1342 /* can be called from interrupts */
1343 extern void kill_fasync(struct fasync_struct **, int, int);
1344
1345 extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1346 extern int f_setown(struct file *filp, unsigned long arg, int force);
1347 extern void f_delown(struct file *filp);
1348 extern pid_t f_getown(struct file *filp);
1349 extern int send_sigurg(struct fown_struct *fown);
1350
1351 /*
1352 * Umount options
1353 */
1354
1355 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1356 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1357 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1358 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1359 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1360
1361 extern struct list_head super_blocks;
1362 extern spinlock_t sb_lock;
1363
1364 struct super_block {
1365 struct list_head s_list; /* Keep this first */
1366 dev_t s_dev; /* search index; _not_ kdev_t */
1367 unsigned char s_dirt;
1368 unsigned char s_blocksize_bits;
1369 unsigned long s_blocksize;
1370 loff_t s_maxbytes; /* Max file size */
1371 struct file_system_type *s_type;
1372 const struct super_operations *s_op;
1373 const struct dquot_operations *dq_op;
1374 const struct quotactl_ops *s_qcop;
1375 const struct export_operations *s_export_op;
1376 unsigned long s_flags;
1377 unsigned long s_magic;
1378 struct dentry *s_root;
1379 struct rw_semaphore s_umount;
1380 struct mutex s_lock;
1381 int s_count;
1382 atomic_t s_active;
1383 #ifdef CONFIG_SECURITY
1384 void *s_security;
1385 #endif
1386 const struct xattr_handler **s_xattr;
1387
1388 struct list_head s_inodes; /* all inodes */
1389 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1390 #ifdef CONFIG_SMP
1391 struct list_head __percpu *s_files;
1392 #else
1393 struct list_head s_files;
1394 #endif
1395 /* s_dentry_lru, s_nr_dentry_unused protected by dcache.c lru locks */
1396 struct list_head s_dentry_lru; /* unused dentry lru */
1397 int s_nr_dentry_unused; /* # of dentry on lru */
1398
1399 struct block_device *s_bdev;
1400 struct backing_dev_info *s_bdi;
1401 struct mtd_info *s_mtd;
1402 struct list_head s_instances;
1403 struct quota_info s_dquot; /* Diskquota specific options */
1404
1405 int s_frozen;
1406 wait_queue_head_t s_wait_unfrozen;
1407
1408 char s_id[32]; /* Informational name */
1409 u8 s_uuid[16]; /* UUID */
1410
1411 void *s_fs_info; /* Filesystem private info */
1412 fmode_t s_mode;
1413
1414 /* Granularity of c/m/atime in ns.
1415 Cannot be worse than a second */
1416 u32 s_time_gran;
1417
1418 /*
1419 * The next field is for VFS *only*. No filesystems have any business
1420 * even looking at it. You had been warned.
1421 */
1422 struct mutex s_vfs_rename_mutex; /* Kludge */
1423
1424 /*
1425 * Filesystem subtype. If non-empty the filesystem type field
1426 * in /proc/mounts will be "type.subtype"
1427 */
1428 char *s_subtype;
1429
1430 /*
1431 * Saved mount options for lazy filesystems using
1432 * generic_show_options()
1433 */
1434 char __rcu *s_options;
1435 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1436
1437 /*
1438 * Saved pool identifier for cleancache (-1 means none)
1439 */
1440 int cleancache_poolid;
1441 };
1442
1443 extern struct timespec current_fs_time(struct super_block *sb);
1444
1445 /*
1446 * Snapshotting support.
1447 */
1448 enum {
1449 SB_UNFROZEN = 0,
1450 SB_FREEZE_WRITE = 1,
1451 SB_FREEZE_TRANS = 2,
1452 };
1453
1454 #define vfs_check_frozen(sb, level) \
1455 wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
1456
1457 #define get_fs_excl() atomic_inc(&current->fs_excl)
1458 #define put_fs_excl() atomic_dec(&current->fs_excl)
1459 #define has_fs_excl() atomic_read(&current->fs_excl)
1460
1461 /*
1462 * until VFS tracks user namespaces for inodes, just make all files
1463 * belong to init_user_ns
1464 */
1465 extern struct user_namespace init_user_ns;
1466 #define inode_userns(inode) (&init_user_ns)
1467 extern bool inode_owner_or_capable(const struct inode *inode);
1468
1469 /* not quite ready to be deprecated, but... */
1470 extern void lock_super(struct super_block *);
1471 extern void unlock_super(struct super_block *);
1472
1473 /*
1474 * VFS helper functions..
1475 */
1476 extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *);
1477 extern int vfs_mkdir(struct inode *, struct dentry *, int);
1478 extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t);
1479 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1480 extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
1481 extern int vfs_rmdir(struct inode *, struct dentry *);
1482 extern int vfs_unlink(struct inode *, struct dentry *);
1483 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1484
1485 /*
1486 * VFS dentry helper functions.
1487 */
1488 extern void dentry_unhash(struct dentry *dentry);
1489
1490 /*
1491 * VFS file helper functions.
1492 */
1493 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1494 mode_t mode);
1495 /*
1496 * VFS FS_IOC_FIEMAP helper definitions.
1497 */
1498 struct fiemap_extent_info {
1499 unsigned int fi_flags; /* Flags as passed from user */
1500 unsigned int fi_extents_mapped; /* Number of mapped extents */
1501 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1502 struct fiemap_extent __user *fi_extents_start; /* Start of
1503 fiemap_extent array */
1504 };
1505 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1506 u64 phys, u64 len, u32 flags);
1507 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1508
1509 /*
1510 * File types
1511 *
1512 * NOTE! These match bits 12..15 of stat.st_mode
1513 * (ie "(i_mode >> 12) & 15").
1514 */
1515 #define DT_UNKNOWN 0
1516 #define DT_FIFO 1
1517 #define DT_CHR 2
1518 #define DT_DIR 4
1519 #define DT_BLK 6
1520 #define DT_REG 8
1521 #define DT_LNK 10
1522 #define DT_SOCK 12
1523 #define DT_WHT 14
1524
1525 /*
1526 * This is the "filldir" function type, used by readdir() to let
1527 * the kernel specify what kind of dirent layout it wants to have.
1528 * This allows the kernel to read directories into kernel space or
1529 * to have different dirent layouts depending on the binary type.
1530 */
1531 typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
1532 struct block_device_operations;
1533
1534 /* These macros are for out of kernel modules to test that
1535 * the kernel supports the unlocked_ioctl and compat_ioctl
1536 * fields in struct file_operations. */
1537 #define HAVE_COMPAT_IOCTL 1
1538 #define HAVE_UNLOCKED_IOCTL 1
1539
1540 /*
1541 * NOTE:
1542 * all file operations except setlease can be called without
1543 * the big kernel lock held in all filesystems.
1544 */
1545 struct file_operations {
1546 struct module *owner;
1547 loff_t (*llseek) (struct file *, loff_t, int);
1548 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1549 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1550 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1551 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1552 int (*readdir) (struct file *, void *, filldir_t);
1553 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1554 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1555 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1556 int (*mmap) (struct file *, struct vm_area_struct *);
1557 int (*open) (struct inode *, struct file *);
1558 int (*flush) (struct file *, fl_owner_t id);
1559 int (*release) (struct inode *, struct file *);
1560 int (*fsync) (struct file *, int datasync);
1561 int (*aio_fsync) (struct kiocb *, int datasync);
1562 int (*fasync) (int, struct file *, int);
1563 int (*lock) (struct file *, int, struct file_lock *);
1564 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1565 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1566 int (*check_flags)(int);
1567 int (*flock) (struct file *, int, struct file_lock *);
1568 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1569 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1570 int (*setlease)(struct file *, long, struct file_lock **);
1571 long (*fallocate)(struct file *file, int mode, loff_t offset,
1572 loff_t len);
1573 };
1574
1575 #define IPERM_FLAG_RCU 0x0001
1576
1577 struct inode_operations {
1578 struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
1579 void * (*follow_link) (struct dentry *, struct nameidata *);
1580 int (*permission) (struct inode *, int, unsigned int);
1581 int (*check_acl)(struct inode *, int, unsigned int);
1582
1583 int (*readlink) (struct dentry *, char __user *,int);
1584 void (*put_link) (struct dentry *, struct nameidata *, void *);
1585
1586 int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
1587 int (*link) (struct dentry *,struct inode *,struct dentry *);
1588 int (*unlink) (struct inode *,struct dentry *);
1589 int (*symlink) (struct inode *,struct dentry *,const char *);
1590 int (*mkdir) (struct inode *,struct dentry *,int);
1591 int (*rmdir) (struct inode *,struct dentry *);
1592 int (*mknod) (struct inode *,struct dentry *,int,dev_t);
1593 int (*rename) (struct inode *, struct dentry *,
1594 struct inode *, struct dentry *);
1595 void (*truncate) (struct inode *);
1596 int (*setattr) (struct dentry *, struct iattr *);
1597 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1598 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1599 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1600 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1601 int (*removexattr) (struct dentry *, const char *);
1602 void (*truncate_range)(struct inode *, loff_t, loff_t);
1603 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1604 u64 len);
1605 } ____cacheline_aligned;
1606
1607 struct seq_file;
1608
1609 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1610 unsigned long nr_segs, unsigned long fast_segs,
1611 struct iovec *fast_pointer,
1612 struct iovec **ret_pointer);
1613
1614 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1615 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1616 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1617 unsigned long, loff_t *);
1618 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1619 unsigned long, loff_t *);
1620
1621 struct super_operations {
1622 struct inode *(*alloc_inode)(struct super_block *sb);
1623 void (*destroy_inode)(struct inode *);
1624
1625 void (*dirty_inode) (struct inode *, int flags);
1626 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1627 int (*drop_inode) (struct inode *);
1628 void (*evict_inode) (struct inode *);
1629 void (*put_super) (struct super_block *);
1630 void (*write_super) (struct super_block *);
1631 int (*sync_fs)(struct super_block *sb, int wait);
1632 int (*freeze_fs) (struct super_block *);
1633 int (*unfreeze_fs) (struct super_block *);
1634 int (*statfs) (struct dentry *, struct kstatfs *);
1635 int (*remount_fs) (struct super_block *, int *, char *);
1636 void (*umount_begin) (struct super_block *);
1637
1638 int (*show_options)(struct seq_file *, struct vfsmount *);
1639 int (*show_devname)(struct seq_file *, struct vfsmount *);
1640 int (*show_path)(struct seq_file *, struct vfsmount *);
1641 int (*show_stats)(struct seq_file *, struct vfsmount *);
1642 #ifdef CONFIG_QUOTA
1643 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1644 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1645 #endif
1646 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1647 };
1648
1649 /*
1650 * Inode state bits. Protected by inode->i_lock
1651 *
1652 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1653 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1654 *
1655 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1656 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1657 * various stages of removing an inode.
1658 *
1659 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1660 *
1661 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1662 * fdatasync(). i_atime is the usual cause.
1663 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1664 * these changes separately from I_DIRTY_SYNC so that we
1665 * don't have to write inode on fdatasync() when only
1666 * mtime has changed in it.
1667 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1668 * I_NEW Serves as both a mutex and completion notification.
1669 * New inodes set I_NEW. If two processes both create
1670 * the same inode, one of them will release its inode and
1671 * wait for I_NEW to be released before returning.
1672 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1673 * also cause waiting on I_NEW, without I_NEW actually
1674 * being set. find_inode() uses this to prevent returning
1675 * nearly-dead inodes.
1676 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1677 * is zero. I_FREEING must be set when I_WILL_FREE is
1678 * cleared.
1679 * I_FREEING Set when inode is about to be freed but still has dirty
1680 * pages or buffers attached or the inode itself is still
1681 * dirty.
1682 * I_CLEAR Added by end_writeback(). In this state the inode is clean
1683 * and can be destroyed. Inode keeps I_FREEING.
1684 *
1685 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1686 * prohibited for many purposes. iget() must wait for
1687 * the inode to be completely released, then create it
1688 * anew. Other functions will just ignore such inodes,
1689 * if appropriate. I_NEW is used for waiting.
1690 *
1691 * I_SYNC Synchonized write of dirty inode data. The bits is
1692 * set during data writeback, and cleared with a wakeup
1693 * on the bit address once it is done.
1694 *
1695 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1696 */
1697 #define I_DIRTY_SYNC (1 << 0)
1698 #define I_DIRTY_DATASYNC (1 << 1)
1699 #define I_DIRTY_PAGES (1 << 2)
1700 #define __I_NEW 3
1701 #define I_NEW (1 << __I_NEW)
1702 #define I_WILL_FREE (1 << 4)
1703 #define I_FREEING (1 << 5)
1704 #define I_CLEAR (1 << 6)
1705 #define __I_SYNC 7
1706 #define I_SYNC (1 << __I_SYNC)
1707 #define I_REFERENCED (1 << 8)
1708
1709 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1710
1711 extern void __mark_inode_dirty(struct inode *, int);
1712 static inline void mark_inode_dirty(struct inode *inode)
1713 {
1714 __mark_inode_dirty(inode, I_DIRTY);
1715 }
1716
1717 static inline void mark_inode_dirty_sync(struct inode *inode)
1718 {
1719 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1720 }
1721
1722 /**
1723 * inc_nlink - directly increment an inode's link count
1724 * @inode: inode
1725 *
1726 * This is a low-level filesystem helper to replace any
1727 * direct filesystem manipulation of i_nlink. Currently,
1728 * it is only here for parity with dec_nlink().
1729 */
1730 static inline void inc_nlink(struct inode *inode)
1731 {
1732 inode->i_nlink++;
1733 }
1734
1735 static inline void inode_inc_link_count(struct inode *inode)
1736 {
1737 inc_nlink(inode);
1738 mark_inode_dirty(inode);
1739 }
1740
1741 /**
1742 * drop_nlink - directly drop an inode's link count
1743 * @inode: inode
1744 *
1745 * This is a low-level filesystem helper to replace any
1746 * direct filesystem manipulation of i_nlink. In cases
1747 * where we are attempting to track writes to the
1748 * filesystem, a decrement to zero means an imminent
1749 * write when the file is truncated and actually unlinked
1750 * on the filesystem.
1751 */
1752 static inline void drop_nlink(struct inode *inode)
1753 {
1754 inode->i_nlink--;
1755 }
1756
1757 /**
1758 * clear_nlink - directly zero an inode's link count
1759 * @inode: inode
1760 *
1761 * This is a low-level filesystem helper to replace any
1762 * direct filesystem manipulation of i_nlink. See
1763 * drop_nlink() for why we care about i_nlink hitting zero.
1764 */
1765 static inline void clear_nlink(struct inode *inode)
1766 {
1767 inode->i_nlink = 0;
1768 }
1769
1770 static inline void inode_dec_link_count(struct inode *inode)
1771 {
1772 drop_nlink(inode);
1773 mark_inode_dirty(inode);
1774 }
1775
1776 /**
1777 * inode_inc_iversion - increments i_version
1778 * @inode: inode that need to be updated
1779 *
1780 * Every time the inode is modified, the i_version field will be incremented.
1781 * The filesystem has to be mounted with i_version flag
1782 */
1783
1784 static inline void inode_inc_iversion(struct inode *inode)
1785 {
1786 spin_lock(&inode->i_lock);
1787 inode->i_version++;
1788 spin_unlock(&inode->i_lock);
1789 }
1790
1791 extern void touch_atime(struct vfsmount *mnt, struct dentry *dentry);
1792 static inline void file_accessed(struct file *file)
1793 {
1794 if (!(file->f_flags & O_NOATIME))
1795 touch_atime(file->f_path.mnt, file->f_path.dentry);
1796 }
1797
1798 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1799 int sync_inode_metadata(struct inode *inode, int wait);
1800
1801 struct file_system_type {
1802 const char *name;
1803 int fs_flags;
1804 struct dentry *(*mount) (struct file_system_type *, int,
1805 const char *, void *);
1806 void (*kill_sb) (struct super_block *);
1807 struct module *owner;
1808 struct file_system_type * next;
1809 struct list_head fs_supers;
1810
1811 struct lock_class_key s_lock_key;
1812 struct lock_class_key s_umount_key;
1813 struct lock_class_key s_vfs_rename_key;
1814
1815 struct lock_class_key i_lock_key;
1816 struct lock_class_key i_mutex_key;
1817 struct lock_class_key i_mutex_dir_key;
1818 struct lock_class_key i_alloc_sem_key;
1819 };
1820
1821 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1822 void *data, int (*fill_super)(struct super_block *, void *, int));
1823 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1824 int flags, const char *dev_name, void *data,
1825 int (*fill_super)(struct super_block *, void *, int));
1826 extern struct dentry *mount_single(struct file_system_type *fs_type,
1827 int flags, void *data,
1828 int (*fill_super)(struct super_block *, void *, int));
1829 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1830 int flags, void *data,
1831 int (*fill_super)(struct super_block *, void *, int));
1832 void generic_shutdown_super(struct super_block *sb);
1833 void kill_block_super(struct super_block *sb);
1834 void kill_anon_super(struct super_block *sb);
1835 void kill_litter_super(struct super_block *sb);
1836 void deactivate_super(struct super_block *sb);
1837 void deactivate_locked_super(struct super_block *sb);
1838 int set_anon_super(struct super_block *s, void *data);
1839 struct super_block *sget(struct file_system_type *type,
1840 int (*test)(struct super_block *,void *),
1841 int (*set)(struct super_block *,void *),
1842 void *data);
1843 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
1844 const struct super_operations *ops,
1845 const struct dentry_operations *dops,
1846 unsigned long);
1847
1848 static inline void sb_mark_dirty(struct super_block *sb)
1849 {
1850 sb->s_dirt = 1;
1851 }
1852 static inline void sb_mark_clean(struct super_block *sb)
1853 {
1854 sb->s_dirt = 0;
1855 }
1856 static inline int sb_is_dirty(struct super_block *sb)
1857 {
1858 return sb->s_dirt;
1859 }
1860
1861 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1862 #define fops_get(fops) \
1863 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1864 #define fops_put(fops) \
1865 do { if (fops) module_put((fops)->owner); } while(0)
1866
1867 extern int register_filesystem(struct file_system_type *);
1868 extern int unregister_filesystem(struct file_system_type *);
1869 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1870 #define kern_mount(type) kern_mount_data(type, NULL)
1871 extern int may_umount_tree(struct vfsmount *);
1872 extern int may_umount(struct vfsmount *);
1873 extern long do_mount(char *, char *, char *, unsigned long, void *);
1874 extern struct vfsmount *collect_mounts(struct path *);
1875 extern void drop_collected_mounts(struct vfsmount *);
1876 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1877 struct vfsmount *);
1878 extern int vfs_statfs(struct path *, struct kstatfs *);
1879 extern int user_statfs(const char __user *, struct kstatfs *);
1880 extern int fd_statfs(int, struct kstatfs *);
1881 extern int statfs_by_dentry(struct dentry *, struct kstatfs *);
1882 extern int freeze_super(struct super_block *super);
1883 extern int thaw_super(struct super_block *super);
1884
1885 extern int current_umask(void);
1886
1887 /* /sys/fs */
1888 extern struct kobject *fs_kobj;
1889
1890 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
1891 extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1892
1893 #define FLOCK_VERIFY_READ 1
1894 #define FLOCK_VERIFY_WRITE 2
1895
1896 #ifdef CONFIG_FILE_LOCKING
1897 extern int locks_mandatory_locked(struct inode *);
1898 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1899
1900 /*
1901 * Candidates for mandatory locking have the setgid bit set
1902 * but no group execute bit - an otherwise meaningless combination.
1903 */
1904
1905 static inline int __mandatory_lock(struct inode *ino)
1906 {
1907 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1908 }
1909
1910 /*
1911 * ... and these candidates should be on MS_MANDLOCK mounted fs,
1912 * otherwise these will be advisory locks
1913 */
1914
1915 static inline int mandatory_lock(struct inode *ino)
1916 {
1917 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1918 }
1919
1920 static inline int locks_verify_locked(struct inode *inode)
1921 {
1922 if (mandatory_lock(inode))
1923 return locks_mandatory_locked(inode);
1924 return 0;
1925 }
1926
1927 static inline int locks_verify_truncate(struct inode *inode,
1928 struct file *filp,
1929 loff_t size)
1930 {
1931 if (inode->i_flock && mandatory_lock(inode))
1932 return locks_mandatory_area(
1933 FLOCK_VERIFY_WRITE, inode, filp,
1934 size < inode->i_size ? size : inode->i_size,
1935 (size < inode->i_size ? inode->i_size - size
1936 : size - inode->i_size)
1937 );
1938 return 0;
1939 }
1940
1941 static inline int break_lease(struct inode *inode, unsigned int mode)
1942 {
1943 if (inode->i_flock)
1944 return __break_lease(inode, mode);
1945 return 0;
1946 }
1947 #else /* !CONFIG_FILE_LOCKING */
1948 static inline int locks_mandatory_locked(struct inode *inode)
1949 {
1950 return 0;
1951 }
1952
1953 static inline int locks_mandatory_area(int rw, struct inode *inode,
1954 struct file *filp, loff_t offset,
1955 size_t count)
1956 {
1957 return 0;
1958 }
1959
1960 static inline int __mandatory_lock(struct inode *inode)
1961 {
1962 return 0;
1963 }
1964
1965 static inline int mandatory_lock(struct inode *inode)
1966 {
1967 return 0;
1968 }
1969
1970 static inline int locks_verify_locked(struct inode *inode)
1971 {
1972 return 0;
1973 }
1974
1975 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
1976 size_t size)
1977 {
1978 return 0;
1979 }
1980
1981 static inline int break_lease(struct inode *inode, unsigned int mode)
1982 {
1983 return 0;
1984 }
1985
1986 #endif /* CONFIG_FILE_LOCKING */
1987
1988 /* fs/open.c */
1989
1990 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
1991 struct file *filp);
1992 extern int do_fallocate(struct file *file, int mode, loff_t offset,
1993 loff_t len);
1994 extern long do_sys_open(int dfd, const char __user *filename, int flags,
1995 int mode);
1996 extern struct file *filp_open(const char *, int, int);
1997 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
1998 const char *, int);
1999 extern struct file * dentry_open(struct dentry *, struct vfsmount *, int,
2000 const struct cred *);
2001 extern int filp_close(struct file *, fl_owner_t id);
2002 extern char * getname(const char __user *);
2003
2004 /* fs/ioctl.c */
2005
2006 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2007
2008 /* fs/dcache.c */
2009 extern void __init vfs_caches_init_early(void);
2010 extern void __init vfs_caches_init(unsigned long);
2011
2012 extern struct kmem_cache *names_cachep;
2013
2014 #define __getname_gfp(gfp) kmem_cache_alloc(names_cachep, (gfp))
2015 #define __getname() __getname_gfp(GFP_KERNEL)
2016 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2017 #ifndef CONFIG_AUDITSYSCALL
2018 #define putname(name) __putname(name)
2019 #else
2020 extern void putname(const char *name);
2021 #endif
2022
2023 #ifdef CONFIG_BLOCK
2024 extern int register_blkdev(unsigned int, const char *);
2025 extern void unregister_blkdev(unsigned int, const char *);
2026 extern struct block_device *bdget(dev_t);
2027 extern struct block_device *bdgrab(struct block_device *bdev);
2028 extern void bd_set_size(struct block_device *, loff_t size);
2029 extern void bd_forget(struct inode *inode);
2030 extern void bdput(struct block_device *);
2031 extern void invalidate_bdev(struct block_device *);
2032 extern int sync_blockdev(struct block_device *bdev);
2033 extern struct super_block *freeze_bdev(struct block_device *);
2034 extern void emergency_thaw_all(void);
2035 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2036 extern int fsync_bdev(struct block_device *);
2037 #else
2038 static inline void bd_forget(struct inode *inode) {}
2039 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2040 static inline void invalidate_bdev(struct block_device *bdev) {}
2041
2042 static inline struct super_block *freeze_bdev(struct block_device *sb)
2043 {
2044 return NULL;
2045 }
2046
2047 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2048 {
2049 return 0;
2050 }
2051 #endif
2052 extern int sync_filesystem(struct super_block *);
2053 extern const struct file_operations def_blk_fops;
2054 extern const struct file_operations def_chr_fops;
2055 extern const struct file_operations bad_sock_fops;
2056 extern const struct file_operations def_fifo_fops;
2057 #ifdef CONFIG_BLOCK
2058 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2059 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2060 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2061 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2062 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2063 void *holder);
2064 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2065 void *holder);
2066 extern int blkdev_put(struct block_device *bdev, fmode_t mode);
2067 #ifdef CONFIG_SYSFS
2068 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2069 extern void bd_unlink_disk_holder(struct block_device *bdev,
2070 struct gendisk *disk);
2071 #else
2072 static inline int bd_link_disk_holder(struct block_device *bdev,
2073 struct gendisk *disk)
2074 {
2075 return 0;
2076 }
2077 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2078 struct gendisk *disk)
2079 {
2080 }
2081 #endif
2082 #endif
2083
2084 /* fs/char_dev.c */
2085 #define CHRDEV_MAJOR_HASH_SIZE 255
2086 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2087 extern int register_chrdev_region(dev_t, unsigned, const char *);
2088 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2089 unsigned int count, const char *name,
2090 const struct file_operations *fops);
2091 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2092 unsigned int count, const char *name);
2093 extern void unregister_chrdev_region(dev_t, unsigned);
2094 extern void chrdev_show(struct seq_file *,off_t);
2095
2096 static inline int register_chrdev(unsigned int major, const char *name,
2097 const struct file_operations *fops)
2098 {
2099 return __register_chrdev(major, 0, 256, name, fops);
2100 }
2101
2102 static inline void unregister_chrdev(unsigned int major, const char *name)
2103 {
2104 __unregister_chrdev(major, 0, 256, name);
2105 }
2106
2107 /* fs/block_dev.c */
2108 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2109 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2110
2111 #ifdef CONFIG_BLOCK
2112 #define BLKDEV_MAJOR_HASH_SIZE 255
2113 extern const char *__bdevname(dev_t, char *buffer);
2114 extern const char *bdevname(struct block_device *bdev, char *buffer);
2115 extern struct block_device *lookup_bdev(const char *);
2116 extern void blkdev_show(struct seq_file *,off_t);
2117
2118 #else
2119 #define BLKDEV_MAJOR_HASH_SIZE 0
2120 #endif
2121
2122 extern void init_special_inode(struct inode *, umode_t, dev_t);
2123
2124 /* Invalid inode operations -- fs/bad_inode.c */
2125 extern void make_bad_inode(struct inode *);
2126 extern int is_bad_inode(struct inode *);
2127
2128 extern const struct file_operations read_pipefifo_fops;
2129 extern const struct file_operations write_pipefifo_fops;
2130 extern const struct file_operations rdwr_pipefifo_fops;
2131
2132 extern int fs_may_remount_ro(struct super_block *);
2133
2134 #ifdef CONFIG_BLOCK
2135 /*
2136 * return READ, READA, or WRITE
2137 */
2138 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2139
2140 /*
2141 * return data direction, READ or WRITE
2142 */
2143 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2144
2145 extern void check_disk_size_change(struct gendisk *disk,
2146 struct block_device *bdev);
2147 extern int revalidate_disk(struct gendisk *);
2148 extern int check_disk_change(struct block_device *);
2149 extern int __invalidate_device(struct block_device *, bool);
2150 extern int invalidate_partition(struct gendisk *, int);
2151 #endif
2152 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2153 pgoff_t start, pgoff_t end);
2154
2155 static inline void invalidate_remote_inode(struct inode *inode)
2156 {
2157 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2158 S_ISLNK(inode->i_mode))
2159 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2160 }
2161 extern int invalidate_inode_pages2(struct address_space *mapping);
2162 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2163 pgoff_t start, pgoff_t end);
2164 extern int write_inode_now(struct inode *, int);
2165 extern int filemap_fdatawrite(struct address_space *);
2166 extern int filemap_flush(struct address_space *);
2167 extern int filemap_fdatawait(struct address_space *);
2168 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2169 loff_t lend);
2170 extern int filemap_write_and_wait(struct address_space *mapping);
2171 extern int filemap_write_and_wait_range(struct address_space *mapping,
2172 loff_t lstart, loff_t lend);
2173 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2174 loff_t start, loff_t end, int sync_mode);
2175 extern int filemap_fdatawrite_range(struct address_space *mapping,
2176 loff_t start, loff_t end);
2177
2178 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2179 int datasync);
2180 extern int vfs_fsync(struct file *file, int datasync);
2181 extern int generic_write_sync(struct file *file, loff_t pos, loff_t count);
2182 extern void sync_supers(void);
2183 extern void emergency_sync(void);
2184 extern void emergency_remount(void);
2185 #ifdef CONFIG_BLOCK
2186 extern sector_t bmap(struct inode *, sector_t);
2187 #endif
2188 extern int notify_change(struct dentry *, struct iattr *);
2189 extern int inode_permission(struct inode *, int);
2190 extern int generic_permission(struct inode *, int, unsigned int,
2191 int (*check_acl)(struct inode *, int, unsigned int));
2192
2193 static inline bool execute_ok(struct inode *inode)
2194 {
2195 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2196 }
2197
2198 extern int get_write_access(struct inode *);
2199 extern int deny_write_access(struct file *);
2200 static inline void put_write_access(struct inode * inode)
2201 {
2202 atomic_dec(&inode->i_writecount);
2203 }
2204 static inline void allow_write_access(struct file *file)
2205 {
2206 if (file)
2207 atomic_inc(&file->f_path.dentry->d_inode->i_writecount);
2208 }
2209 #ifdef CONFIG_IMA
2210 static inline void i_readcount_dec(struct inode *inode)
2211 {
2212 BUG_ON(!atomic_read(&inode->i_readcount));
2213 atomic_dec(&inode->i_readcount);
2214 }
2215 static inline void i_readcount_inc(struct inode *inode)
2216 {
2217 atomic_inc(&inode->i_readcount);
2218 }
2219 #else
2220 static inline void i_readcount_dec(struct inode *inode)
2221 {
2222 return;
2223 }
2224 static inline void i_readcount_inc(struct inode *inode)
2225 {
2226 return;
2227 }
2228 #endif
2229 extern int do_pipe_flags(int *, int);
2230 extern struct file *create_read_pipe(struct file *f, int flags);
2231 extern struct file *create_write_pipe(int flags);
2232 extern void free_write_pipe(struct file *);
2233
2234 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2235 extern struct file * open_exec(const char *);
2236
2237 /* fs/dcache.c -- generic fs support functions */
2238 extern int is_subdir(struct dentry *, struct dentry *);
2239 extern int path_is_under(struct path *, struct path *);
2240 extern ino_t find_inode_number(struct dentry *, struct qstr *);
2241
2242 #include <linux/err.h>
2243
2244 /* needed for stackable file system support */
2245 extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
2246
2247 extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
2248
2249 extern int inode_init_always(struct super_block *, struct inode *);
2250 extern void inode_init_once(struct inode *);
2251 extern void address_space_init_once(struct address_space *mapping);
2252 extern void ihold(struct inode * inode);
2253 extern void iput(struct inode *);
2254 extern struct inode * igrab(struct inode *);
2255 extern ino_t iunique(struct super_block *, ino_t);
2256 extern int inode_needs_sync(struct inode *inode);
2257 extern int generic_delete_inode(struct inode *inode);
2258 extern int generic_drop_inode(struct inode *inode);
2259
2260 extern struct inode *ilookup5_nowait(struct super_block *sb,
2261 unsigned long hashval, int (*test)(struct inode *, void *),
2262 void *data);
2263 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2264 int (*test)(struct inode *, void *), void *data);
2265 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2266
2267 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2268 extern struct inode * iget_locked(struct super_block *, unsigned long);
2269 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2270 extern int insert_inode_locked(struct inode *);
2271 extern void unlock_new_inode(struct inode *);
2272 extern unsigned int get_next_ino(void);
2273
2274 extern void __iget(struct inode * inode);
2275 extern void iget_failed(struct inode *);
2276 extern void end_writeback(struct inode *);
2277 extern void __destroy_inode(struct inode *);
2278 extern struct inode *new_inode(struct super_block *);
2279 extern void free_inode_nonrcu(struct inode *inode);
2280 extern int should_remove_suid(struct dentry *);
2281 extern int file_remove_suid(struct file *);
2282
2283 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2284 extern void remove_inode_hash(struct inode *);
2285 static inline void insert_inode_hash(struct inode *inode)
2286 {
2287 __insert_inode_hash(inode, inode->i_ino);
2288 }
2289 extern void inode_sb_list_add(struct inode *inode);
2290
2291 #ifdef CONFIG_BLOCK
2292 extern void submit_bio(int, struct bio *);
2293 extern int bdev_read_only(struct block_device *);
2294 #endif
2295 extern int set_blocksize(struct block_device *, int);
2296 extern int sb_set_blocksize(struct super_block *, int);
2297 extern int sb_min_blocksize(struct super_block *, int);
2298
2299 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2300 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2301 extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
2302 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
2303 extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2304 extern ssize_t __generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long,
2305 loff_t *);
2306 extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2307 extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
2308 unsigned long *, loff_t, loff_t *, size_t, size_t);
2309 extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
2310 unsigned long, loff_t, loff_t *, size_t, ssize_t);
2311 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2312 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2313 extern int generic_segment_checks(const struct iovec *iov,
2314 unsigned long *nr_segs, size_t *count, int access_flags);
2315
2316 /* fs/block_dev.c */
2317 extern ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov,
2318 unsigned long nr_segs, loff_t pos);
2319 extern int blkdev_fsync(struct file *filp, int datasync);
2320
2321 /* fs/splice.c */
2322 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2323 struct pipe_inode_info *, size_t, unsigned int);
2324 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2325 struct pipe_inode_info *, size_t, unsigned int);
2326 extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
2327 struct file *, loff_t *, size_t, unsigned int);
2328 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2329 struct file *out, loff_t *, size_t len, unsigned int flags);
2330 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2331 size_t len, unsigned int flags);
2332
2333 extern void
2334 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2335 extern loff_t noop_llseek(struct file *file, loff_t offset, int origin);
2336 extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
2337 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
2338 extern loff_t generic_file_llseek_unlocked(struct file *file, loff_t offset,
2339 int origin);
2340 extern int generic_file_open(struct inode * inode, struct file * filp);
2341 extern int nonseekable_open(struct inode * inode, struct file * filp);
2342
2343 #ifdef CONFIG_FS_XIP
2344 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2345 loff_t *ppos);
2346 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
2347 extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2348 size_t len, loff_t *ppos);
2349 extern int xip_truncate_page(struct address_space *mapping, loff_t from);
2350 #else
2351 static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2352 {
2353 return 0;
2354 }
2355 #endif
2356
2357 #ifdef CONFIG_BLOCK
2358 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2359 loff_t file_offset);
2360
2361 enum {
2362 /* need locking between buffered and direct access */
2363 DIO_LOCKING = 0x01,
2364
2365 /* filesystem does not support filling holes */
2366 DIO_SKIP_HOLES = 0x02,
2367 };
2368
2369 void dio_end_io(struct bio *bio, int error);
2370
2371 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2372 struct block_device *bdev, const struct iovec *iov, loff_t offset,
2373 unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
2374 dio_submit_t submit_io, int flags);
2375
2376 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
2377 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
2378 loff_t offset, unsigned long nr_segs, get_block_t get_block,
2379 dio_iodone_t end_io)
2380 {
2381 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
2382 nr_segs, get_block, end_io, NULL,
2383 DIO_LOCKING | DIO_SKIP_HOLES);
2384 }
2385 #endif
2386
2387 extern const struct file_operations generic_ro_fops;
2388
2389 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2390
2391 extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
2392 extern int vfs_follow_link(struct nameidata *, const char *);
2393 extern int page_readlink(struct dentry *, char __user *, int);
2394 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2395 extern void page_put_link(struct dentry *, struct nameidata *, void *);
2396 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2397 int nofs);
2398 extern int page_symlink(struct inode *inode, const char *symname, int len);
2399 extern const struct inode_operations page_symlink_inode_operations;
2400 extern int generic_readlink(struct dentry *, char __user *, int);
2401 extern void generic_fillattr(struct inode *, struct kstat *);
2402 extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2403 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2404 void inode_add_bytes(struct inode *inode, loff_t bytes);
2405 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2406 loff_t inode_get_bytes(struct inode *inode);
2407 void inode_set_bytes(struct inode *inode, loff_t bytes);
2408
2409 extern int vfs_readdir(struct file *, filldir_t, void *);
2410
2411 extern int vfs_stat(const char __user *, struct kstat *);
2412 extern int vfs_lstat(const char __user *, struct kstat *);
2413 extern int vfs_fstat(unsigned int, struct kstat *);
2414 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2415
2416 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2417 unsigned long arg);
2418 extern int __generic_block_fiemap(struct inode *inode,
2419 struct fiemap_extent_info *fieinfo,
2420 loff_t start, loff_t len,
2421 get_block_t *get_block);
2422 extern int generic_block_fiemap(struct inode *inode,
2423 struct fiemap_extent_info *fieinfo, u64 start,
2424 u64 len, get_block_t *get_block);
2425
2426 extern void get_filesystem(struct file_system_type *fs);
2427 extern void put_filesystem(struct file_system_type *fs);
2428 extern struct file_system_type *get_fs_type(const char *name);
2429 extern struct super_block *get_super(struct block_device *);
2430 extern struct super_block *get_active_super(struct block_device *bdev);
2431 extern struct super_block *user_get_super(dev_t);
2432 extern void drop_super(struct super_block *sb);
2433 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2434 extern void iterate_supers_type(struct file_system_type *,
2435 void (*)(struct super_block *, void *), void *);
2436
2437 extern int dcache_dir_open(struct inode *, struct file *);
2438 extern int dcache_dir_close(struct inode *, struct file *);
2439 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2440 extern int dcache_readdir(struct file *, void *, filldir_t);
2441 extern int simple_setattr(struct dentry *, struct iattr *);
2442 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2443 extern int simple_statfs(struct dentry *, struct kstatfs *);
2444 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2445 extern int simple_unlink(struct inode *, struct dentry *);
2446 extern int simple_rmdir(struct inode *, struct dentry *);
2447 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2448 extern int noop_fsync(struct file *, int);
2449 extern int simple_empty(struct dentry *);
2450 extern int simple_readpage(struct file *file, struct page *page);
2451 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2452 loff_t pos, unsigned len, unsigned flags,
2453 struct page **pagep, void **fsdata);
2454 extern int simple_write_end(struct file *file, struct address_space *mapping,
2455 loff_t pos, unsigned len, unsigned copied,
2456 struct page *page, void *fsdata);
2457
2458 extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
2459 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2460 extern const struct file_operations simple_dir_operations;
2461 extern const struct inode_operations simple_dir_inode_operations;
2462 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2463 struct dentry *d_alloc_name(struct dentry *, const char *);
2464 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2465 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2466 extern void simple_release_fs(struct vfsmount **mount, int *count);
2467
2468 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2469 loff_t *ppos, const void *from, size_t available);
2470 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2471 const void __user *from, size_t count);
2472
2473 extern int generic_file_fsync(struct file *, int);
2474
2475 extern int generic_check_addressable(unsigned, u64);
2476
2477 #ifdef CONFIG_MIGRATION
2478 extern int buffer_migrate_page(struct address_space *,
2479 struct page *, struct page *);
2480 #else
2481 #define buffer_migrate_page NULL
2482 #endif
2483
2484 extern int inode_change_ok(const struct inode *, struct iattr *);
2485 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2486 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2487
2488 extern void file_update_time(struct file *file);
2489
2490 extern int generic_show_options(struct seq_file *m, struct vfsmount *mnt);
2491 extern void save_mount_options(struct super_block *sb, char *options);
2492 extern void replace_mount_options(struct super_block *sb, char *options);
2493
2494 static inline ino_t parent_ino(struct dentry *dentry)
2495 {
2496 ino_t res;
2497
2498 /*
2499 * Don't strictly need d_lock here? If the parent ino could change
2500 * then surely we'd have a deeper race in the caller?
2501 */
2502 spin_lock(&dentry->d_lock);
2503 res = dentry->d_parent->d_inode->i_ino;
2504 spin_unlock(&dentry->d_lock);
2505 return res;
2506 }
2507
2508 /* Transaction based IO helpers */
2509
2510 /*
2511 * An argresp is stored in an allocated page and holds the
2512 * size of the argument or response, along with its content
2513 */
2514 struct simple_transaction_argresp {
2515 ssize_t size;
2516 char data[0];
2517 };
2518
2519 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2520
2521 char *simple_transaction_get(struct file *file, const char __user *buf,
2522 size_t size);
2523 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2524 size_t size, loff_t *pos);
2525 int simple_transaction_release(struct inode *inode, struct file *file);
2526
2527 void simple_transaction_set(struct file *file, size_t n);
2528
2529 /*
2530 * simple attribute files
2531 *
2532 * These attributes behave similar to those in sysfs:
2533 *
2534 * Writing to an attribute immediately sets a value, an open file can be
2535 * written to multiple times.
2536 *
2537 * Reading from an attribute creates a buffer from the value that might get
2538 * read with multiple read calls. When the attribute has been read
2539 * completely, no further read calls are possible until the file is opened
2540 * again.
2541 *
2542 * All attributes contain a text representation of a numeric value
2543 * that are accessed with the get() and set() functions.
2544 */
2545 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2546 static int __fops ## _open(struct inode *inode, struct file *file) \
2547 { \
2548 __simple_attr_check_format(__fmt, 0ull); \
2549 return simple_attr_open(inode, file, __get, __set, __fmt); \
2550 } \
2551 static const struct file_operations __fops = { \
2552 .owner = THIS_MODULE, \
2553 .open = __fops ## _open, \
2554 .release = simple_attr_release, \
2555 .read = simple_attr_read, \
2556 .write = simple_attr_write, \
2557 .llseek = generic_file_llseek, \
2558 };
2559
2560 static inline void __attribute__((format(printf, 1, 2)))
2561 __simple_attr_check_format(const char *fmt, ...)
2562 {
2563 /* don't do anything, just let the compiler check the arguments; */
2564 }
2565
2566 int simple_attr_open(struct inode *inode, struct file *file,
2567 int (*get)(void *, u64 *), int (*set)(void *, u64),
2568 const char *fmt);
2569 int simple_attr_release(struct inode *inode, struct file *file);
2570 ssize_t simple_attr_read(struct file *file, char __user *buf,
2571 size_t len, loff_t *ppos);
2572 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2573 size_t len, loff_t *ppos);
2574
2575 struct ctl_table;
2576 int proc_nr_files(struct ctl_table *table, int write,
2577 void __user *buffer, size_t *lenp, loff_t *ppos);
2578 int proc_nr_dentry(struct ctl_table *table, int write,
2579 void __user *buffer, size_t *lenp, loff_t *ppos);
2580 int proc_nr_inodes(struct ctl_table *table, int write,
2581 void __user *buffer, size_t *lenp, loff_t *ppos);
2582 int __init get_filesystem_list(char *buf);
2583
2584 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
2585 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
2586
2587 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2588 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2589 (flag & __FMODE_NONOTIFY)))
2590
2591 static inline int is_sxid(mode_t mode)
2592 {
2593 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
2594 }
2595
2596 static inline void inode_has_no_xattr(struct inode *inode)
2597 {
2598 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
2599 inode->i_flags |= S_NOSEC;
2600 }
2601
2602 #endif /* __KERNEL__ */
2603 #endif /* _LINUX_FS_H */