[PATCH] lockdep: annotate ->s_lock
[GitHub/mt8127/android_kernel_alcatel_ttab.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
12 /*
13 * It's silly to have NR_OPEN bigger than NR_FILE, but you can change
14 * the file limit at runtime and only root can increase the per-process
15 * nr_file rlimit, so it's safe to set up a ridiculously high absolute
16 * upper limit on files-per-process.
17 *
18 * Some programs (notably those using select()) may have to be
19 * recompiled to take full advantage of the new limits..
20 */
21
22 /* Fixed constants first: */
23 #undef NR_OPEN
24 #define NR_OPEN (1024*1024) /* Absolute upper limit on fd num */
25 #define INR_OPEN 1024 /* Initial setting for nfile rlimits */
26
27 #define BLOCK_SIZE_BITS 10
28 #define BLOCK_SIZE (1<<BLOCK_SIZE_BITS)
29
30 /* And dynamically-tunable limits and defaults: */
31 struct files_stat_struct {
32 int nr_files; /* read only */
33 int nr_free_files; /* read only */
34 int max_files; /* tunable */
35 };
36 extern struct files_stat_struct files_stat;
37 extern int get_max_files(void);
38
39 struct inodes_stat_t {
40 int nr_inodes;
41 int nr_unused;
42 int dummy[5];
43 };
44 extern struct inodes_stat_t inodes_stat;
45
46 extern int leases_enable, lease_break_time;
47
48 #ifdef CONFIG_DNOTIFY
49 extern int dir_notify_enable;
50 #endif
51
52 #define NR_FILE 8192 /* this can well be larger on a larger system */
53
54 #define MAY_EXEC 1
55 #define MAY_WRITE 2
56 #define MAY_READ 4
57 #define MAY_APPEND 8
58
59 #define FMODE_READ 1
60 #define FMODE_WRITE 2
61
62 /* Internal kernel extensions */
63 #define FMODE_LSEEK 4
64 #define FMODE_PREAD 8
65 #define FMODE_PWRITE FMODE_PREAD /* These go hand in hand */
66
67 /* File is being opened for execution. Primary users of this flag are
68 distributed filesystems that can use it to achieve correct ETXTBUSY
69 behavior for cross-node execution/opening_for_writing of files */
70 #define FMODE_EXEC 16
71
72 #define RW_MASK 1
73 #define RWA_MASK 2
74 #define READ 0
75 #define WRITE 1
76 #define READA 2 /* read-ahead - don't block if no resources */
77 #define SWRITE 3 /* for ll_rw_block() - wait for buffer lock */
78 #define SPECIAL 4 /* For non-blockdevice requests in request queue */
79 #define READ_SYNC (READ | (1 << BIO_RW_SYNC))
80 #define WRITE_SYNC (WRITE | (1 << BIO_RW_SYNC))
81 #define WRITE_BARRIER ((1 << BIO_RW) | (1 << BIO_RW_BARRIER))
82
83 #define SEL_IN 1
84 #define SEL_OUT 2
85 #define SEL_EX 4
86
87 /* public flags for file_system_type */
88 #define FS_REQUIRES_DEV 1
89 #define FS_BINARY_MOUNTDATA 2
90 #define FS_REVAL_DOT 16384 /* Check the paths ".", ".." for staleness */
91 #define FS_ODD_RENAME 32768 /* Temporary stuff; will go away as soon
92 * as nfs_rename() will be cleaned up
93 */
94 /*
95 * These are the fs-independent mount-flags: up to 32 flags are supported
96 */
97 #define MS_RDONLY 1 /* Mount read-only */
98 #define MS_NOSUID 2 /* Ignore suid and sgid bits */
99 #define MS_NODEV 4 /* Disallow access to device special files */
100 #define MS_NOEXEC 8 /* Disallow program execution */
101 #define MS_SYNCHRONOUS 16 /* Writes are synced at once */
102 #define MS_REMOUNT 32 /* Alter flags of a mounted FS */
103 #define MS_MANDLOCK 64 /* Allow mandatory locks on an FS */
104 #define MS_DIRSYNC 128 /* Directory modifications are synchronous */
105 #define MS_NOATIME 1024 /* Do not update access times. */
106 #define MS_NODIRATIME 2048 /* Do not update directory access times */
107 #define MS_BIND 4096
108 #define MS_MOVE 8192
109 #define MS_REC 16384
110 #define MS_VERBOSE 32768 /* War is peace. Verbosity is silence.
111 MS_VERBOSE is deprecated. */
112 #define MS_SILENT 32768
113 #define MS_POSIXACL (1<<16) /* VFS does not apply the umask */
114 #define MS_UNBINDABLE (1<<17) /* change to unbindable */
115 #define MS_PRIVATE (1<<18) /* change to private */
116 #define MS_SLAVE (1<<19) /* change to slave */
117 #define MS_SHARED (1<<20) /* change to shared */
118 #define MS_ACTIVE (1<<30)
119 #define MS_NOUSER (1<<31)
120
121 /*
122 * Superblock flags that can be altered by MS_REMOUNT
123 */
124 #define MS_RMT_MASK (MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK)
125
126 /*
127 * Old magic mount flag and mask
128 */
129 #define MS_MGC_VAL 0xC0ED0000
130 #define MS_MGC_MSK 0xffff0000
131
132 /* Inode flags - they have nothing to superblock flags now */
133
134 #define S_SYNC 1 /* Writes are synced at once */
135 #define S_NOATIME 2 /* Do not update access times */
136 #define S_APPEND 4 /* Append-only file */
137 #define S_IMMUTABLE 8 /* Immutable file */
138 #define S_DEAD 16 /* removed, but still open directory */
139 #define S_NOQUOTA 32 /* Inode is not counted to quota */
140 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
141 #define S_NOCMTIME 128 /* Do not update file c/mtime */
142 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
143 #define S_PRIVATE 512 /* Inode is fs-internal */
144
145 /*
146 * Note that nosuid etc flags are inode-specific: setting some file-system
147 * flags just means all the inodes inherit those flags by default. It might be
148 * possible to override it selectively if you really wanted to with some
149 * ioctl() that is not currently implemented.
150 *
151 * Exception: MS_RDONLY is always applied to the entire file system.
152 *
153 * Unfortunately, it is possible to change a filesystems flags with it mounted
154 * with files in use. This means that all of the inodes will not have their
155 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
156 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
157 */
158 #define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))
159
160 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
161 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
162 ((inode)->i_flags & S_SYNC))
163 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
164 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
165 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
166
167 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
168 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
169 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
170 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
171
172 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
173 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
174 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
175 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
176
177 /* the read-only stuff doesn't really belong here, but any other place is
178 probably as bad and I don't want to create yet another include file. */
179
180 #define BLKROSET _IO(0x12,93) /* set device read-only (0 = read-write) */
181 #define BLKROGET _IO(0x12,94) /* get read-only status (0 = read_write) */
182 #define BLKRRPART _IO(0x12,95) /* re-read partition table */
183 #define BLKGETSIZE _IO(0x12,96) /* return device size /512 (long *arg) */
184 #define BLKFLSBUF _IO(0x12,97) /* flush buffer cache */
185 #define BLKRASET _IO(0x12,98) /* set read ahead for block device */
186 #define BLKRAGET _IO(0x12,99) /* get current read ahead setting */
187 #define BLKFRASET _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
188 #define BLKFRAGET _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
189 #define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
190 #define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
191 #define BLKSSZGET _IO(0x12,104)/* get block device sector size */
192 #if 0
193 #define BLKPG _IO(0x12,105)/* See blkpg.h */
194
195 /* Some people are morons. Do not use sizeof! */
196
197 #define BLKELVGET _IOR(0x12,106,size_t)/* elevator get */
198 #define BLKELVSET _IOW(0x12,107,size_t)/* elevator set */
199 /* This was here just to show that the number is taken -
200 probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
201 #endif
202 /* A jump here: 108-111 have been used for various private purposes. */
203 #define BLKBSZGET _IOR(0x12,112,size_t)
204 #define BLKBSZSET _IOW(0x12,113,size_t)
205 #define BLKGETSIZE64 _IOR(0x12,114,size_t) /* return device size in bytes (u64 *arg) */
206 #define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup)
207 #define BLKTRACESTART _IO(0x12,116)
208 #define BLKTRACESTOP _IO(0x12,117)
209 #define BLKTRACETEARDOWN _IO(0x12,118)
210
211 #define BMAP_IOCTL 1 /* obsolete - kept for compatibility */
212 #define FIBMAP _IO(0x00,1) /* bmap access */
213 #define FIGETBSZ _IO(0x00,2) /* get the block size used for bmap */
214
215 #define SYNC_FILE_RANGE_WAIT_BEFORE 1
216 #define SYNC_FILE_RANGE_WRITE 2
217 #define SYNC_FILE_RANGE_WAIT_AFTER 4
218
219 #ifdef __KERNEL__
220
221 #include <linux/linkage.h>
222 #include <linux/wait.h>
223 #include <linux/types.h>
224 #include <linux/kdev_t.h>
225 #include <linux/dcache.h>
226 #include <linux/stat.h>
227 #include <linux/cache.h>
228 #include <linux/kobject.h>
229 #include <linux/list.h>
230 #include <linux/radix-tree.h>
231 #include <linux/prio_tree.h>
232 #include <linux/init.h>
233 #include <linux/sched.h>
234 #include <linux/mutex.h>
235
236 #include <asm/atomic.h>
237 #include <asm/semaphore.h>
238 #include <asm/byteorder.h>
239
240 struct hd_geometry;
241 struct iovec;
242 struct nameidata;
243 struct kiocb;
244 struct pipe_inode_info;
245 struct poll_table_struct;
246 struct kstatfs;
247 struct vm_area_struct;
248 struct vfsmount;
249
250 extern void __init inode_init(unsigned long);
251 extern void __init inode_init_early(void);
252 extern void __init mnt_init(unsigned long);
253 extern void __init files_init(unsigned long);
254
255 struct buffer_head;
256 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
257 struct buffer_head *bh_result, int create);
258 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
259 ssize_t bytes, void *private);
260
261 /*
262 * Attribute flags. These should be or-ed together to figure out what
263 * has been changed!
264 */
265 #define ATTR_MODE 1
266 #define ATTR_UID 2
267 #define ATTR_GID 4
268 #define ATTR_SIZE 8
269 #define ATTR_ATIME 16
270 #define ATTR_MTIME 32
271 #define ATTR_CTIME 64
272 #define ATTR_ATIME_SET 128
273 #define ATTR_MTIME_SET 256
274 #define ATTR_FORCE 512 /* Not a change, but a change it */
275 #define ATTR_ATTR_FLAG 1024
276 #define ATTR_KILL_SUID 2048
277 #define ATTR_KILL_SGID 4096
278 #define ATTR_FILE 8192
279
280 /*
281 * This is the Inode Attributes structure, used for notify_change(). It
282 * uses the above definitions as flags, to know which values have changed.
283 * Also, in this manner, a Filesystem can look at only the values it cares
284 * about. Basically, these are the attributes that the VFS layer can
285 * request to change from the FS layer.
286 *
287 * Derek Atkins <warlord@MIT.EDU> 94-10-20
288 */
289 struct iattr {
290 unsigned int ia_valid;
291 umode_t ia_mode;
292 uid_t ia_uid;
293 gid_t ia_gid;
294 loff_t ia_size;
295 struct timespec ia_atime;
296 struct timespec ia_mtime;
297 struct timespec ia_ctime;
298
299 /*
300 * Not an attribute, but an auxilary info for filesystems wanting to
301 * implement an ftruncate() like method. NOTE: filesystem should
302 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
303 */
304 struct file *ia_file;
305 };
306
307 /*
308 * Includes for diskquotas.
309 */
310 #include <linux/quota.h>
311
312 /**
313 * enum positive_aop_returns - aop return codes with specific semantics
314 *
315 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
316 * completed, that the page is still locked, and
317 * should be considered active. The VM uses this hint
318 * to return the page to the active list -- it won't
319 * be a candidate for writeback again in the near
320 * future. Other callers must be careful to unlock
321 * the page if they get this return. Returned by
322 * writepage();
323 *
324 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
325 * unlocked it and the page might have been truncated.
326 * The caller should back up to acquiring a new page and
327 * trying again. The aop will be taking reasonable
328 * precautions not to livelock. If the caller held a page
329 * reference, it should drop it before retrying. Returned
330 * by readpage(), prepare_write(), and commit_write().
331 *
332 * address_space_operation functions return these large constants to indicate
333 * special semantics to the caller. These are much larger than the bytes in a
334 * page to allow for functions that return the number of bytes operated on in a
335 * given page.
336 */
337
338 enum positive_aop_returns {
339 AOP_WRITEPAGE_ACTIVATE = 0x80000,
340 AOP_TRUNCATED_PAGE = 0x80001,
341 };
342
343 /*
344 * oh the beauties of C type declarations.
345 */
346 struct page;
347 struct address_space;
348 struct writeback_control;
349
350 struct address_space_operations {
351 int (*writepage)(struct page *page, struct writeback_control *wbc);
352 int (*readpage)(struct file *, struct page *);
353 void (*sync_page)(struct page *);
354
355 /* Write back some dirty pages from this mapping. */
356 int (*writepages)(struct address_space *, struct writeback_control *);
357
358 /* Set a page dirty. Return true if this dirtied it */
359 int (*set_page_dirty)(struct page *page);
360
361 int (*readpages)(struct file *filp, struct address_space *mapping,
362 struct list_head *pages, unsigned nr_pages);
363
364 /*
365 * ext3 requires that a successful prepare_write() call be followed
366 * by a commit_write() call - they must be balanced
367 */
368 int (*prepare_write)(struct file *, struct page *, unsigned, unsigned);
369 int (*commit_write)(struct file *, struct page *, unsigned, unsigned);
370 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
371 sector_t (*bmap)(struct address_space *, sector_t);
372 void (*invalidatepage) (struct page *, unsigned long);
373 int (*releasepage) (struct page *, gfp_t);
374 ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
375 loff_t offset, unsigned long nr_segs);
376 struct page* (*get_xip_page)(struct address_space *, sector_t,
377 int);
378 /* migrate the contents of a page to the specified target */
379 int (*migratepage) (struct address_space *,
380 struct page *, struct page *);
381 };
382
383 struct backing_dev_info;
384 struct address_space {
385 struct inode *host; /* owner: inode, block_device */
386 struct radix_tree_root page_tree; /* radix tree of all pages */
387 rwlock_t tree_lock; /* and rwlock protecting it */
388 unsigned int i_mmap_writable;/* count VM_SHARED mappings */
389 struct prio_tree_root i_mmap; /* tree of private and shared mappings */
390 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
391 spinlock_t i_mmap_lock; /* protect tree, count, list */
392 unsigned int truncate_count; /* Cover race condition with truncate */
393 unsigned long nrpages; /* number of total pages */
394 pgoff_t writeback_index;/* writeback starts here */
395 const struct address_space_operations *a_ops; /* methods */
396 unsigned long flags; /* error bits/gfp mask */
397 struct backing_dev_info *backing_dev_info; /* device readahead, etc */
398 spinlock_t private_lock; /* for use by the address_space */
399 struct list_head private_list; /* ditto */
400 struct address_space *assoc_mapping; /* ditto */
401 } __attribute__((aligned(sizeof(long))));
402 /*
403 * On most architectures that alignment is already the case; but
404 * must be enforced here for CRIS, to let the least signficant bit
405 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
406 */
407
408 struct block_device {
409 dev_t bd_dev; /* not a kdev_t - it's a search key */
410 struct inode * bd_inode; /* will die */
411 int bd_openers;
412 struct mutex bd_mutex; /* open/close mutex */
413 struct mutex bd_mount_mutex; /* mount mutex */
414 struct list_head bd_inodes;
415 void * bd_holder;
416 int bd_holders;
417 #ifdef CONFIG_SYSFS
418 struct list_head bd_holder_list;
419 #endif
420 struct block_device * bd_contains;
421 unsigned bd_block_size;
422 struct hd_struct * bd_part;
423 /* number of times partitions within this device have been opened. */
424 unsigned bd_part_count;
425 int bd_invalidated;
426 struct gendisk * bd_disk;
427 struct list_head bd_list;
428 struct backing_dev_info *bd_inode_backing_dev_info;
429 /*
430 * Private data. You must have bd_claim'ed the block_device
431 * to use this. NOTE: bd_claim allows an owner to claim
432 * the same device multiple times, the owner must take special
433 * care to not mess up bd_private for that case.
434 */
435 unsigned long bd_private;
436 };
437
438 /*
439 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
440 * radix trees
441 */
442 #define PAGECACHE_TAG_DIRTY 0
443 #define PAGECACHE_TAG_WRITEBACK 1
444
445 int mapping_tagged(struct address_space *mapping, int tag);
446
447 /*
448 * Might pages of this file be mapped into userspace?
449 */
450 static inline int mapping_mapped(struct address_space *mapping)
451 {
452 return !prio_tree_empty(&mapping->i_mmap) ||
453 !list_empty(&mapping->i_mmap_nonlinear);
454 }
455
456 /*
457 * Might pages of this file have been modified in userspace?
458 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
459 * marks vma as VM_SHARED if it is shared, and the file was opened for
460 * writing i.e. vma may be mprotected writable even if now readonly.
461 */
462 static inline int mapping_writably_mapped(struct address_space *mapping)
463 {
464 return mapping->i_mmap_writable != 0;
465 }
466
467 /*
468 * Use sequence counter to get consistent i_size on 32-bit processors.
469 */
470 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
471 #include <linux/seqlock.h>
472 #define __NEED_I_SIZE_ORDERED
473 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
474 #else
475 #define i_size_ordered_init(inode) do { } while (0)
476 #endif
477
478 struct inode {
479 struct hlist_node i_hash;
480 struct list_head i_list;
481 struct list_head i_sb_list;
482 struct list_head i_dentry;
483 unsigned long i_ino;
484 atomic_t i_count;
485 umode_t i_mode;
486 unsigned int i_nlink;
487 uid_t i_uid;
488 gid_t i_gid;
489 dev_t i_rdev;
490 loff_t i_size;
491 struct timespec i_atime;
492 struct timespec i_mtime;
493 struct timespec i_ctime;
494 unsigned int i_blkbits;
495 unsigned long i_blksize;
496 unsigned long i_version;
497 blkcnt_t i_blocks;
498 unsigned short i_bytes;
499 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
500 struct mutex i_mutex;
501 struct rw_semaphore i_alloc_sem;
502 struct inode_operations *i_op;
503 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
504 struct super_block *i_sb;
505 struct file_lock *i_flock;
506 struct address_space *i_mapping;
507 struct address_space i_data;
508 #ifdef CONFIG_QUOTA
509 struct dquot *i_dquot[MAXQUOTAS];
510 #endif
511 /* These three should probably be a union */
512 struct list_head i_devices;
513 struct pipe_inode_info *i_pipe;
514 struct block_device *i_bdev;
515 struct cdev *i_cdev;
516 int i_cindex;
517
518 __u32 i_generation;
519
520 #ifdef CONFIG_DNOTIFY
521 unsigned long i_dnotify_mask; /* Directory notify events */
522 struct dnotify_struct *i_dnotify; /* for directory notifications */
523 #endif
524
525 #ifdef CONFIG_INOTIFY
526 struct list_head inotify_watches; /* watches on this inode */
527 struct mutex inotify_mutex; /* protects the watches list */
528 #endif
529
530 unsigned long i_state;
531 unsigned long dirtied_when; /* jiffies of first dirtying */
532
533 unsigned int i_flags;
534
535 atomic_t i_writecount;
536 void *i_security;
537 union {
538 void *generic_ip;
539 } u;
540 #ifdef __NEED_I_SIZE_ORDERED
541 seqcount_t i_size_seqcount;
542 #endif
543 };
544
545 /*
546 * inode->i_mutex nesting subclasses for the lock validator:
547 *
548 * 0: the object of the current VFS operation
549 * 1: parent
550 * 2: child/target
551 * 3: quota file
552 *
553 * The locking order between these classes is
554 * parent -> child -> normal -> quota
555 */
556 enum inode_i_mutex_lock_class
557 {
558 I_MUTEX_NORMAL,
559 I_MUTEX_PARENT,
560 I_MUTEX_CHILD,
561 I_MUTEX_QUOTA
562 };
563
564 /*
565 * NOTE: in a 32bit arch with a preemptable kernel and
566 * an UP compile the i_size_read/write must be atomic
567 * with respect to the local cpu (unlike with preempt disabled),
568 * but they don't need to be atomic with respect to other cpus like in
569 * true SMP (so they need either to either locally disable irq around
570 * the read or for example on x86 they can be still implemented as a
571 * cmpxchg8b without the need of the lock prefix). For SMP compiles
572 * and 64bit archs it makes no difference if preempt is enabled or not.
573 */
574 static inline loff_t i_size_read(struct inode *inode)
575 {
576 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
577 loff_t i_size;
578 unsigned int seq;
579
580 do {
581 seq = read_seqcount_begin(&inode->i_size_seqcount);
582 i_size = inode->i_size;
583 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
584 return i_size;
585 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
586 loff_t i_size;
587
588 preempt_disable();
589 i_size = inode->i_size;
590 preempt_enable();
591 return i_size;
592 #else
593 return inode->i_size;
594 #endif
595 }
596
597
598 static inline void i_size_write(struct inode *inode, loff_t i_size)
599 {
600 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
601 write_seqcount_begin(&inode->i_size_seqcount);
602 inode->i_size = i_size;
603 write_seqcount_end(&inode->i_size_seqcount);
604 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
605 preempt_disable();
606 inode->i_size = i_size;
607 preempt_enable();
608 #else
609 inode->i_size = i_size;
610 #endif
611 }
612
613 static inline unsigned iminor(struct inode *inode)
614 {
615 return MINOR(inode->i_rdev);
616 }
617
618 static inline unsigned imajor(struct inode *inode)
619 {
620 return MAJOR(inode->i_rdev);
621 }
622
623 extern struct block_device *I_BDEV(struct inode *inode);
624
625 struct fown_struct {
626 rwlock_t lock; /* protects pid, uid, euid fields */
627 int pid; /* pid or -pgrp where SIGIO should be sent */
628 uid_t uid, euid; /* uid/euid of process setting the owner */
629 void *security;
630 int signum; /* posix.1b rt signal to be delivered on IO */
631 };
632
633 /*
634 * Track a single file's readahead state
635 */
636 struct file_ra_state {
637 unsigned long start; /* Current window */
638 unsigned long size;
639 unsigned long flags; /* ra flags RA_FLAG_xxx*/
640 unsigned long cache_hit; /* cache hit count*/
641 unsigned long prev_page; /* Cache last read() position */
642 unsigned long ahead_start; /* Ahead window */
643 unsigned long ahead_size;
644 unsigned long ra_pages; /* Maximum readahead window */
645 unsigned long mmap_hit; /* Cache hit stat for mmap accesses */
646 unsigned long mmap_miss; /* Cache miss stat for mmap accesses */
647 };
648 #define RA_FLAG_MISS 0x01 /* a cache miss occured against this file */
649 #define RA_FLAG_INCACHE 0x02 /* file is already in cache */
650
651 struct file {
652 /*
653 * fu_list becomes invalid after file_free is called and queued via
654 * fu_rcuhead for RCU freeing
655 */
656 union {
657 struct list_head fu_list;
658 struct rcu_head fu_rcuhead;
659 } f_u;
660 struct dentry *f_dentry;
661 struct vfsmount *f_vfsmnt;
662 const struct file_operations *f_op;
663 atomic_t f_count;
664 unsigned int f_flags;
665 mode_t f_mode;
666 loff_t f_pos;
667 struct fown_struct f_owner;
668 unsigned int f_uid, f_gid;
669 struct file_ra_state f_ra;
670
671 unsigned long f_version;
672 void *f_security;
673
674 /* needed for tty driver, and maybe others */
675 void *private_data;
676
677 #ifdef CONFIG_EPOLL
678 /* Used by fs/eventpoll.c to link all the hooks to this file */
679 struct list_head f_ep_links;
680 spinlock_t f_ep_lock;
681 #endif /* #ifdef CONFIG_EPOLL */
682 struct address_space *f_mapping;
683 };
684 extern spinlock_t files_lock;
685 #define file_list_lock() spin_lock(&files_lock);
686 #define file_list_unlock() spin_unlock(&files_lock);
687
688 #define get_file(x) atomic_inc(&(x)->f_count)
689 #define file_count(x) atomic_read(&(x)->f_count)
690
691 #define MAX_NON_LFS ((1UL<<31) - 1)
692
693 /* Page cache limit. The filesystems should put that into their s_maxbytes
694 limits, otherwise bad things can happen in VM. */
695 #if BITS_PER_LONG==32
696 #define MAX_LFS_FILESIZE (((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
697 #elif BITS_PER_LONG==64
698 #define MAX_LFS_FILESIZE 0x7fffffffffffffffUL
699 #endif
700
701 #define FL_POSIX 1
702 #define FL_FLOCK 2
703 #define FL_ACCESS 8 /* not trying to lock, just looking */
704 #define FL_LEASE 32 /* lease held on this file */
705 #define FL_CLOSE 64 /* unlock on close */
706 #define FL_SLEEP 128 /* A blocking lock */
707
708 /*
709 * The POSIX file lock owner is determined by
710 * the "struct files_struct" in the thread group
711 * (or NULL for no owner - BSD locks).
712 *
713 * Lockd stuffs a "host" pointer into this.
714 */
715 typedef struct files_struct *fl_owner_t;
716
717 struct file_lock_operations {
718 void (*fl_insert)(struct file_lock *); /* lock insertion callback */
719 void (*fl_remove)(struct file_lock *); /* lock removal callback */
720 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
721 void (*fl_release_private)(struct file_lock *);
722 };
723
724 struct lock_manager_operations {
725 int (*fl_compare_owner)(struct file_lock *, struct file_lock *);
726 void (*fl_notify)(struct file_lock *); /* unblock callback */
727 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
728 void (*fl_release_private)(struct file_lock *);
729 void (*fl_break)(struct file_lock *);
730 int (*fl_mylease)(struct file_lock *, struct file_lock *);
731 int (*fl_change)(struct file_lock **, int);
732 };
733
734 /* that will die - we need it for nfs_lock_info */
735 #include <linux/nfs_fs_i.h>
736
737 struct file_lock {
738 struct file_lock *fl_next; /* singly linked list for this inode */
739 struct list_head fl_link; /* doubly linked list of all locks */
740 struct list_head fl_block; /* circular list of blocked processes */
741 fl_owner_t fl_owner;
742 unsigned int fl_pid;
743 wait_queue_head_t fl_wait;
744 struct file *fl_file;
745 unsigned char fl_flags;
746 unsigned char fl_type;
747 loff_t fl_start;
748 loff_t fl_end;
749
750 struct fasync_struct * fl_fasync; /* for lease break notifications */
751 unsigned long fl_break_time; /* for nonblocking lease breaks */
752
753 struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
754 struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
755 union {
756 struct nfs_lock_info nfs_fl;
757 struct nfs4_lock_info nfs4_fl;
758 } fl_u;
759 };
760
761 /* The following constant reflects the upper bound of the file/locking space */
762 #ifndef OFFSET_MAX
763 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
764 #define OFFSET_MAX INT_LIMIT(loff_t)
765 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
766 #endif
767
768 #include <linux/fcntl.h>
769
770 extern int fcntl_getlk(struct file *, struct flock __user *);
771 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
772 struct flock __user *);
773
774 #if BITS_PER_LONG == 32
775 extern int fcntl_getlk64(struct file *, struct flock64 __user *);
776 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
777 struct flock64 __user *);
778 #endif
779
780 extern void send_sigio(struct fown_struct *fown, int fd, int band);
781 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
782 extern int fcntl_getlease(struct file *filp);
783
784 /* fs/sync.c */
785 extern int do_sync_file_range(struct file *file, loff_t offset, loff_t endbyte,
786 unsigned int flags);
787
788 /* fs/locks.c */
789 extern void locks_init_lock(struct file_lock *);
790 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
791 extern void locks_remove_posix(struct file *, fl_owner_t);
792 extern void locks_remove_flock(struct file *);
793 extern int posix_test_lock(struct file *, struct file_lock *, struct file_lock *);
794 extern int posix_lock_file_conf(struct file *, struct file_lock *, struct file_lock *);
795 extern int posix_lock_file(struct file *, struct file_lock *);
796 extern int posix_lock_file_wait(struct file *, struct file_lock *);
797 extern int posix_unblock_lock(struct file *, struct file_lock *);
798 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
799 extern int __break_lease(struct inode *inode, unsigned int flags);
800 extern void lease_get_mtime(struct inode *, struct timespec *time);
801 extern int setlease(struct file *, long, struct file_lock **);
802 extern int lease_modify(struct file_lock **, int);
803 extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
804 extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
805
806 struct fasync_struct {
807 int magic;
808 int fa_fd;
809 struct fasync_struct *fa_next; /* singly linked list */
810 struct file *fa_file;
811 };
812
813 #define FASYNC_MAGIC 0x4601
814
815 /* SMP safe fasync helpers: */
816 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
817 /* can be called from interrupts */
818 extern void kill_fasync(struct fasync_struct **, int, int);
819 /* only for net: no internal synchronization */
820 extern void __kill_fasync(struct fasync_struct *, int, int);
821
822 extern int f_setown(struct file *filp, unsigned long arg, int force);
823 extern void f_delown(struct file *filp);
824 extern int send_sigurg(struct fown_struct *fown);
825
826 /*
827 * Umount options
828 */
829
830 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
831 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
832 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
833
834 extern struct list_head super_blocks;
835 extern spinlock_t sb_lock;
836
837 #define sb_entry(list) list_entry((list), struct super_block, s_list)
838 #define S_BIAS (1<<30)
839 struct super_block {
840 struct list_head s_list; /* Keep this first */
841 dev_t s_dev; /* search index; _not_ kdev_t */
842 unsigned long s_blocksize;
843 unsigned char s_blocksize_bits;
844 unsigned char s_dirt;
845 unsigned long long s_maxbytes; /* Max file size */
846 struct file_system_type *s_type;
847 struct super_operations *s_op;
848 struct dquot_operations *dq_op;
849 struct quotactl_ops *s_qcop;
850 struct export_operations *s_export_op;
851 unsigned long s_flags;
852 unsigned long s_magic;
853 struct dentry *s_root;
854 struct rw_semaphore s_umount;
855 struct mutex s_lock;
856 int s_count;
857 int s_syncing;
858 int s_need_sync_fs;
859 atomic_t s_active;
860 void *s_security;
861 struct xattr_handler **s_xattr;
862
863 struct list_head s_inodes; /* all inodes */
864 struct list_head s_dirty; /* dirty inodes */
865 struct list_head s_io; /* parked for writeback */
866 struct hlist_head s_anon; /* anonymous dentries for (nfs) exporting */
867 struct list_head s_files;
868
869 struct block_device *s_bdev;
870 struct list_head s_instances;
871 struct quota_info s_dquot; /* Diskquota specific options */
872
873 int s_frozen;
874 wait_queue_head_t s_wait_unfrozen;
875
876 char s_id[32]; /* Informational name */
877
878 void *s_fs_info; /* Filesystem private info */
879
880 /*
881 * The next field is for VFS *only*. No filesystems have any business
882 * even looking at it. You had been warned.
883 */
884 struct mutex s_vfs_rename_mutex; /* Kludge */
885
886 /* Granularity of c/m/atime in ns.
887 Cannot be worse than a second */
888 u32 s_time_gran;
889 };
890
891 extern struct timespec current_fs_time(struct super_block *sb);
892
893 /*
894 * Snapshotting support.
895 */
896 enum {
897 SB_UNFROZEN = 0,
898 SB_FREEZE_WRITE = 1,
899 SB_FREEZE_TRANS = 2,
900 };
901
902 #define vfs_check_frozen(sb, level) \
903 wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
904
905 static inline void get_fs_excl(void)
906 {
907 atomic_inc(&current->fs_excl);
908 }
909
910 static inline void put_fs_excl(void)
911 {
912 atomic_dec(&current->fs_excl);
913 }
914
915 static inline int has_fs_excl(void)
916 {
917 return atomic_read(&current->fs_excl);
918 }
919
920
921 /*
922 * Superblock locking.
923 */
924 static inline void lock_super(struct super_block * sb)
925 {
926 get_fs_excl();
927 mutex_lock(&sb->s_lock);
928 }
929
930 static inline void unlock_super(struct super_block * sb)
931 {
932 put_fs_excl();
933 mutex_unlock(&sb->s_lock);
934 }
935
936 /*
937 * VFS helper functions..
938 */
939 extern int vfs_permission(struct nameidata *, int);
940 extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *);
941 extern int vfs_mkdir(struct inode *, struct dentry *, int);
942 extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t);
943 extern int vfs_symlink(struct inode *, struct dentry *, const char *, int);
944 extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
945 extern int vfs_rmdir(struct inode *, struct dentry *);
946 extern int vfs_unlink(struct inode *, struct dentry *);
947 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
948
949 /*
950 * VFS dentry helper functions.
951 */
952 extern void dentry_unhash(struct dentry *dentry);
953
954 /*
955 * VFS file helper functions.
956 */
957 extern int file_permission(struct file *, int);
958
959 /*
960 * File types
961 *
962 * NOTE! These match bits 12..15 of stat.st_mode
963 * (ie "(i_mode >> 12) & 15").
964 */
965 #define DT_UNKNOWN 0
966 #define DT_FIFO 1
967 #define DT_CHR 2
968 #define DT_DIR 4
969 #define DT_BLK 6
970 #define DT_REG 8
971 #define DT_LNK 10
972 #define DT_SOCK 12
973 #define DT_WHT 14
974
975 #define OSYNC_METADATA (1<<0)
976 #define OSYNC_DATA (1<<1)
977 #define OSYNC_INODE (1<<2)
978 int generic_osync_inode(struct inode *, struct address_space *, int);
979
980 /*
981 * This is the "filldir" function type, used by readdir() to let
982 * the kernel specify what kind of dirent layout it wants to have.
983 * This allows the kernel to read directories into kernel space or
984 * to have different dirent layouts depending on the binary type.
985 */
986 typedef int (*filldir_t)(void *, const char *, int, loff_t, ino_t, unsigned);
987
988 struct block_device_operations {
989 int (*open) (struct inode *, struct file *);
990 int (*release) (struct inode *, struct file *);
991 int (*ioctl) (struct inode *, struct file *, unsigned, unsigned long);
992 long (*unlocked_ioctl) (struct file *, unsigned, unsigned long);
993 long (*compat_ioctl) (struct file *, unsigned, unsigned long);
994 int (*direct_access) (struct block_device *, sector_t, unsigned long *);
995 int (*media_changed) (struct gendisk *);
996 int (*revalidate_disk) (struct gendisk *);
997 int (*getgeo)(struct block_device *, struct hd_geometry *);
998 struct module *owner;
999 };
1000
1001 /*
1002 * "descriptor" for what we're up to with a read for sendfile().
1003 * This allows us to use the same read code yet
1004 * have multiple different users of the data that
1005 * we read from a file.
1006 *
1007 * The simplest case just copies the data to user
1008 * mode.
1009 */
1010 typedef struct {
1011 size_t written;
1012 size_t count;
1013 union {
1014 char __user * buf;
1015 void *data;
1016 } arg;
1017 int error;
1018 } read_descriptor_t;
1019
1020 typedef int (*read_actor_t)(read_descriptor_t *, struct page *, unsigned long, unsigned long);
1021
1022 /* These macros are for out of kernel modules to test that
1023 * the kernel supports the unlocked_ioctl and compat_ioctl
1024 * fields in struct file_operations. */
1025 #define HAVE_COMPAT_IOCTL 1
1026 #define HAVE_UNLOCKED_IOCTL 1
1027
1028 /*
1029 * NOTE:
1030 * read, write, poll, fsync, readv, writev, unlocked_ioctl and compat_ioctl
1031 * can be called without the big kernel lock held in all filesystems.
1032 */
1033 struct file_operations {
1034 struct module *owner;
1035 loff_t (*llseek) (struct file *, loff_t, int);
1036 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1037 ssize_t (*aio_read) (struct kiocb *, char __user *, size_t, loff_t);
1038 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1039 ssize_t (*aio_write) (struct kiocb *, const char __user *, size_t, loff_t);
1040 int (*readdir) (struct file *, void *, filldir_t);
1041 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1042 int (*ioctl) (struct inode *, struct file *, unsigned int, unsigned long);
1043 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1044 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1045 int (*mmap) (struct file *, struct vm_area_struct *);
1046 int (*open) (struct inode *, struct file *);
1047 int (*flush) (struct file *, fl_owner_t id);
1048 int (*release) (struct inode *, struct file *);
1049 int (*fsync) (struct file *, struct dentry *, int datasync);
1050 int (*aio_fsync) (struct kiocb *, int datasync);
1051 int (*fasync) (int, struct file *, int);
1052 int (*lock) (struct file *, int, struct file_lock *);
1053 ssize_t (*readv) (struct file *, const struct iovec *, unsigned long, loff_t *);
1054 ssize_t (*writev) (struct file *, const struct iovec *, unsigned long, loff_t *);
1055 ssize_t (*sendfile) (struct file *, loff_t *, size_t, read_actor_t, void *);
1056 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1057 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1058 int (*check_flags)(int);
1059 int (*dir_notify)(struct file *filp, unsigned long arg);
1060 int (*flock) (struct file *, int, struct file_lock *);
1061 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1062 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1063 };
1064
1065 struct inode_operations {
1066 int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
1067 struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
1068 int (*link) (struct dentry *,struct inode *,struct dentry *);
1069 int (*unlink) (struct inode *,struct dentry *);
1070 int (*symlink) (struct inode *,struct dentry *,const char *);
1071 int (*mkdir) (struct inode *,struct dentry *,int);
1072 int (*rmdir) (struct inode *,struct dentry *);
1073 int (*mknod) (struct inode *,struct dentry *,int,dev_t);
1074 int (*rename) (struct inode *, struct dentry *,
1075 struct inode *, struct dentry *);
1076 int (*readlink) (struct dentry *, char __user *,int);
1077 void * (*follow_link) (struct dentry *, struct nameidata *);
1078 void (*put_link) (struct dentry *, struct nameidata *, void *);
1079 void (*truncate) (struct inode *);
1080 int (*permission) (struct inode *, int, struct nameidata *);
1081 int (*setattr) (struct dentry *, struct iattr *);
1082 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1083 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1084 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1085 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1086 int (*removexattr) (struct dentry *, const char *);
1087 void (*truncate_range)(struct inode *, loff_t, loff_t);
1088 };
1089
1090 struct seq_file;
1091
1092 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1093 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1094 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1095 unsigned long, loff_t *);
1096 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1097 unsigned long, loff_t *);
1098
1099 /*
1100 * NOTE: write_inode, delete_inode, clear_inode, put_inode can be called
1101 * without the big kernel lock held in all filesystems.
1102 */
1103 struct super_operations {
1104 struct inode *(*alloc_inode)(struct super_block *sb);
1105 void (*destroy_inode)(struct inode *);
1106
1107 void (*read_inode) (struct inode *);
1108
1109 void (*dirty_inode) (struct inode *);
1110 int (*write_inode) (struct inode *, int);
1111 void (*put_inode) (struct inode *);
1112 void (*drop_inode) (struct inode *);
1113 void (*delete_inode) (struct inode *);
1114 void (*put_super) (struct super_block *);
1115 void (*write_super) (struct super_block *);
1116 int (*sync_fs)(struct super_block *sb, int wait);
1117 void (*write_super_lockfs) (struct super_block *);
1118 void (*unlockfs) (struct super_block *);
1119 int (*statfs) (struct dentry *, struct kstatfs *);
1120 int (*remount_fs) (struct super_block *, int *, char *);
1121 void (*clear_inode) (struct inode *);
1122 void (*umount_begin) (struct vfsmount *, int);
1123
1124 int (*show_options)(struct seq_file *, struct vfsmount *);
1125 int (*show_stats)(struct seq_file *, struct vfsmount *);
1126
1127 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1128 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1129 };
1130
1131 /* Inode state bits. Protected by inode_lock. */
1132 #define I_DIRTY_SYNC 1 /* Not dirty enough for O_DATASYNC */
1133 #define I_DIRTY_DATASYNC 2 /* Data-related inode changes pending */
1134 #define I_DIRTY_PAGES 4 /* Data-related inode changes pending */
1135 #define __I_LOCK 3
1136 #define I_LOCK (1 << __I_LOCK)
1137 #define I_FREEING 16
1138 #define I_CLEAR 32
1139 #define I_NEW 64
1140 #define I_WILL_FREE 128
1141
1142 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1143
1144 extern void __mark_inode_dirty(struct inode *, int);
1145 static inline void mark_inode_dirty(struct inode *inode)
1146 {
1147 __mark_inode_dirty(inode, I_DIRTY);
1148 }
1149
1150 static inline void mark_inode_dirty_sync(struct inode *inode)
1151 {
1152 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1153 }
1154
1155 static inline void inode_inc_link_count(struct inode *inode)
1156 {
1157 inode->i_nlink++;
1158 mark_inode_dirty(inode);
1159 }
1160
1161 static inline void inode_dec_link_count(struct inode *inode)
1162 {
1163 inode->i_nlink--;
1164 mark_inode_dirty(inode);
1165 }
1166
1167 extern void touch_atime(struct vfsmount *mnt, struct dentry *dentry);
1168 static inline void file_accessed(struct file *file)
1169 {
1170 if (!(file->f_flags & O_NOATIME))
1171 touch_atime(file->f_vfsmnt, file->f_dentry);
1172 }
1173
1174 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1175
1176 /**
1177 * struct export_operations - for nfsd to communicate with file systems
1178 * @decode_fh: decode a file handle fragment and return a &struct dentry
1179 * @encode_fh: encode a file handle fragment from a dentry
1180 * @get_name: find the name for a given inode in a given directory
1181 * @get_parent: find the parent of a given directory
1182 * @get_dentry: find a dentry for the inode given a file handle sub-fragment
1183 * @find_exported_dentry:
1184 * set by the exporting module to a standard helper function.
1185 *
1186 * Description:
1187 * The export_operations structure provides a means for nfsd to communicate
1188 * with a particular exported file system - particularly enabling nfsd and
1189 * the filesystem to co-operate when dealing with file handles.
1190 *
1191 * export_operations contains two basic operation for dealing with file
1192 * handles, decode_fh() and encode_fh(), and allows for some other
1193 * operations to be defined which standard helper routines use to get
1194 * specific information from the filesystem.
1195 *
1196 * nfsd encodes information use to determine which filesystem a filehandle
1197 * applies to in the initial part of the file handle. The remainder, termed
1198 * a file handle fragment, is controlled completely by the filesystem. The
1199 * standard helper routines assume that this fragment will contain one or
1200 * two sub-fragments, one which identifies the file, and one which may be
1201 * used to identify the (a) directory containing the file.
1202 *
1203 * In some situations, nfsd needs to get a dentry which is connected into a
1204 * specific part of the file tree. To allow for this, it passes the
1205 * function acceptable() together with a @context which can be used to see
1206 * if the dentry is acceptable. As there can be multiple dentrys for a
1207 * given file, the filesystem should check each one for acceptability before
1208 * looking for the next. As soon as an acceptable one is found, it should
1209 * be returned.
1210 *
1211 * decode_fh:
1212 * @decode_fh is given a &struct super_block (@sb), a file handle fragment
1213 * (@fh, @fh_len) and an acceptability testing function (@acceptable,
1214 * @context). It should return a &struct dentry which refers to the same
1215 * file that the file handle fragment refers to, and which passes the
1216 * acceptability test. If it cannot, it should return a %NULL pointer if
1217 * the file was found but no acceptable &dentries were available, or a
1218 * %ERR_PTR error code indicating why it couldn't be found (e.g. %ENOENT or
1219 * %ENOMEM).
1220 *
1221 * encode_fh:
1222 * @encode_fh should store in the file handle fragment @fh (using at most
1223 * @max_len bytes) information that can be used by @decode_fh to recover the
1224 * file refered to by the &struct dentry @de. If the @connectable flag is
1225 * set, the encode_fh() should store sufficient information so that a good
1226 * attempt can be made to find not only the file but also it's place in the
1227 * filesystem. This typically means storing a reference to de->d_parent in
1228 * the filehandle fragment. encode_fh() should return the number of bytes
1229 * stored or a negative error code such as %-ENOSPC
1230 *
1231 * get_name:
1232 * @get_name should find a name for the given @child in the given @parent
1233 * directory. The name should be stored in the @name (with the
1234 * understanding that it is already pointing to a a %NAME_MAX+1 sized
1235 * buffer. get_name() should return %0 on success, a negative error code
1236 * or error. @get_name will be called without @parent->i_mutex held.
1237 *
1238 * get_parent:
1239 * @get_parent should find the parent directory for the given @child which
1240 * is also a directory. In the event that it cannot be found, or storage
1241 * space cannot be allocated, a %ERR_PTR should be returned.
1242 *
1243 * get_dentry:
1244 * Given a &super_block (@sb) and a pointer to a file-system specific inode
1245 * identifier, possibly an inode number, (@inump) get_dentry() should find
1246 * the identified inode and return a dentry for that inode. Any suitable
1247 * dentry can be returned including, if necessary, a new dentry created with
1248 * d_alloc_root. The caller can then find any other extant dentrys by
1249 * following the d_alias links. If a new dentry was created using
1250 * d_alloc_root, DCACHE_NFSD_DISCONNECTED should be set, and the dentry
1251 * should be d_rehash()ed.
1252 *
1253 * If the inode cannot be found, either a %NULL pointer or an %ERR_PTR code
1254 * can be returned. The @inump will be whatever was passed to
1255 * nfsd_find_fh_dentry() in either the @obj or @parent parameters.
1256 *
1257 * Locking rules:
1258 * get_parent is called with child->d_inode->i_mutex down
1259 * get_name is not (which is possibly inconsistent)
1260 */
1261
1262 struct export_operations {
1263 struct dentry *(*decode_fh)(struct super_block *sb, __u32 *fh, int fh_len, int fh_type,
1264 int (*acceptable)(void *context, struct dentry *de),
1265 void *context);
1266 int (*encode_fh)(struct dentry *de, __u32 *fh, int *max_len,
1267 int connectable);
1268
1269 /* the following are only called from the filesystem itself */
1270 int (*get_name)(struct dentry *parent, char *name,
1271 struct dentry *child);
1272 struct dentry * (*get_parent)(struct dentry *child);
1273 struct dentry * (*get_dentry)(struct super_block *sb, void *inump);
1274
1275 /* This is set by the exporting module to a standard helper */
1276 struct dentry * (*find_exported_dentry)(
1277 struct super_block *sb, void *obj, void *parent,
1278 int (*acceptable)(void *context, struct dentry *de),
1279 void *context);
1280
1281
1282 };
1283
1284 extern struct dentry *
1285 find_exported_dentry(struct super_block *sb, void *obj, void *parent,
1286 int (*acceptable)(void *context, struct dentry *de),
1287 void *context);
1288
1289 struct file_system_type {
1290 const char *name;
1291 int fs_flags;
1292 int (*get_sb) (struct file_system_type *, int,
1293 const char *, void *, struct vfsmount *);
1294 void (*kill_sb) (struct super_block *);
1295 struct module *owner;
1296 struct file_system_type * next;
1297 struct list_head fs_supers;
1298 struct lock_class_key s_lock_key;
1299 };
1300
1301 extern int get_sb_bdev(struct file_system_type *fs_type,
1302 int flags, const char *dev_name, void *data,
1303 int (*fill_super)(struct super_block *, void *, int),
1304 struct vfsmount *mnt);
1305 extern int get_sb_single(struct file_system_type *fs_type,
1306 int flags, void *data,
1307 int (*fill_super)(struct super_block *, void *, int),
1308 struct vfsmount *mnt);
1309 extern int get_sb_nodev(struct file_system_type *fs_type,
1310 int flags, void *data,
1311 int (*fill_super)(struct super_block *, void *, int),
1312 struct vfsmount *mnt);
1313 void generic_shutdown_super(struct super_block *sb);
1314 void kill_block_super(struct super_block *sb);
1315 void kill_anon_super(struct super_block *sb);
1316 void kill_litter_super(struct super_block *sb);
1317 void deactivate_super(struct super_block *sb);
1318 int set_anon_super(struct super_block *s, void *data);
1319 struct super_block *sget(struct file_system_type *type,
1320 int (*test)(struct super_block *,void *),
1321 int (*set)(struct super_block *,void *),
1322 void *data);
1323 extern int get_sb_pseudo(struct file_system_type *, char *,
1324 struct super_operations *ops, unsigned long,
1325 struct vfsmount *mnt);
1326 extern int simple_set_mnt(struct vfsmount *mnt, struct super_block *sb);
1327 int __put_super(struct super_block *sb);
1328 int __put_super_and_need_restart(struct super_block *sb);
1329 void unnamed_dev_init(void);
1330
1331 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1332 #define fops_get(fops) \
1333 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1334 #define fops_put(fops) \
1335 do { if (fops) module_put((fops)->owner); } while(0)
1336
1337 extern int register_filesystem(struct file_system_type *);
1338 extern int unregister_filesystem(struct file_system_type *);
1339 extern struct vfsmount *kern_mount(struct file_system_type *);
1340 extern int may_umount_tree(struct vfsmount *);
1341 extern int may_umount(struct vfsmount *);
1342 extern void umount_tree(struct vfsmount *, int, struct list_head *);
1343 extern void release_mounts(struct list_head *);
1344 extern long do_mount(char *, char *, char *, unsigned long, void *);
1345 extern struct vfsmount *copy_tree(struct vfsmount *, struct dentry *, int);
1346 extern void mnt_set_mountpoint(struct vfsmount *, struct dentry *,
1347 struct vfsmount *);
1348
1349 extern int vfs_statfs(struct dentry *, struct kstatfs *);
1350
1351 /* /sys/fs */
1352 extern struct subsystem fs_subsys;
1353
1354 #define FLOCK_VERIFY_READ 1
1355 #define FLOCK_VERIFY_WRITE 2
1356
1357 extern int locks_mandatory_locked(struct inode *);
1358 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1359
1360 /*
1361 * Candidates for mandatory locking have the setgid bit set
1362 * but no group execute bit - an otherwise meaningless combination.
1363 */
1364 #define MANDATORY_LOCK(inode) \
1365 (IS_MANDLOCK(inode) && ((inode)->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID)
1366
1367 static inline int locks_verify_locked(struct inode *inode)
1368 {
1369 if (MANDATORY_LOCK(inode))
1370 return locks_mandatory_locked(inode);
1371 return 0;
1372 }
1373
1374 extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1375
1376 static inline int locks_verify_truncate(struct inode *inode,
1377 struct file *filp,
1378 loff_t size)
1379 {
1380 if (inode->i_flock && MANDATORY_LOCK(inode))
1381 return locks_mandatory_area(
1382 FLOCK_VERIFY_WRITE, inode, filp,
1383 size < inode->i_size ? size : inode->i_size,
1384 (size < inode->i_size ? inode->i_size - size
1385 : size - inode->i_size)
1386 );
1387 return 0;
1388 }
1389
1390 static inline int break_lease(struct inode *inode, unsigned int mode)
1391 {
1392 if (inode->i_flock)
1393 return __break_lease(inode, mode);
1394 return 0;
1395 }
1396
1397 /* fs/open.c */
1398
1399 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
1400 struct file *filp);
1401 extern long do_sys_open(int fdf, const char __user *filename, int flags,
1402 int mode);
1403 extern struct file *filp_open(const char *, int, int);
1404 extern struct file * dentry_open(struct dentry *, struct vfsmount *, int);
1405 extern int filp_close(struct file *, fl_owner_t id);
1406 extern char * getname(const char __user *);
1407
1408 /* fs/dcache.c */
1409 extern void __init vfs_caches_init_early(void);
1410 extern void __init vfs_caches_init(unsigned long);
1411
1412 #define __getname() kmem_cache_alloc(names_cachep, SLAB_KERNEL)
1413 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
1414 #ifndef CONFIG_AUDITSYSCALL
1415 #define putname(name) __putname(name)
1416 #else
1417 extern void putname(const char *name);
1418 #endif
1419
1420 extern int register_blkdev(unsigned int, const char *);
1421 extern int unregister_blkdev(unsigned int, const char *);
1422 extern struct block_device *bdget(dev_t);
1423 extern void bd_set_size(struct block_device *, loff_t size);
1424 extern void bd_forget(struct inode *inode);
1425 extern void bdput(struct block_device *);
1426 extern struct block_device *open_by_devnum(dev_t, unsigned);
1427 extern const struct file_operations def_blk_fops;
1428 extern const struct address_space_operations def_blk_aops;
1429 extern const struct file_operations def_chr_fops;
1430 extern const struct file_operations bad_sock_fops;
1431 extern const struct file_operations def_fifo_fops;
1432 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
1433 extern int blkdev_ioctl(struct inode *, struct file *, unsigned, unsigned long);
1434 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
1435 extern int blkdev_get(struct block_device *, mode_t, unsigned);
1436 extern int blkdev_put(struct block_device *);
1437 extern int bd_claim(struct block_device *, void *);
1438 extern void bd_release(struct block_device *);
1439 #ifdef CONFIG_SYSFS
1440 extern int bd_claim_by_disk(struct block_device *, void *, struct gendisk *);
1441 extern void bd_release_from_disk(struct block_device *, struct gendisk *);
1442 #else
1443 #define bd_claim_by_disk(bdev, holder, disk) bd_claim(bdev, holder)
1444 #define bd_release_from_disk(bdev, disk) bd_release(bdev)
1445 #endif
1446
1447 /* fs/char_dev.c */
1448 #define CHRDEV_MAJOR_HASH_SIZE 255
1449 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
1450 extern int register_chrdev_region(dev_t, unsigned, const char *);
1451 extern int register_chrdev(unsigned int, const char *,
1452 const struct file_operations *);
1453 extern int unregister_chrdev(unsigned int, const char *);
1454 extern void unregister_chrdev_region(dev_t, unsigned);
1455 extern int chrdev_open(struct inode *, struct file *);
1456 extern void chrdev_show(struct seq_file *,off_t);
1457
1458 /* fs/block_dev.c */
1459 #define BLKDEV_MAJOR_HASH_SIZE 255
1460 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
1461 extern const char *__bdevname(dev_t, char *buffer);
1462 extern const char *bdevname(struct block_device *bdev, char *buffer);
1463 extern struct block_device *lookup_bdev(const char *);
1464 extern struct block_device *open_bdev_excl(const char *, int, void *);
1465 extern void close_bdev_excl(struct block_device *);
1466 extern void blkdev_show(struct seq_file *,off_t);
1467
1468 extern void init_special_inode(struct inode *, umode_t, dev_t);
1469
1470 /* Invalid inode operations -- fs/bad_inode.c */
1471 extern void make_bad_inode(struct inode *);
1472 extern int is_bad_inode(struct inode *);
1473
1474 extern const struct file_operations read_fifo_fops;
1475 extern const struct file_operations write_fifo_fops;
1476 extern const struct file_operations rdwr_fifo_fops;
1477
1478 extern int fs_may_remount_ro(struct super_block *);
1479
1480 /*
1481 * return READ, READA, or WRITE
1482 */
1483 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
1484
1485 /*
1486 * return data direction, READ or WRITE
1487 */
1488 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
1489
1490 extern int check_disk_change(struct block_device *);
1491 extern int invalidate_inodes(struct super_block *);
1492 extern int __invalidate_device(struct block_device *);
1493 extern int invalidate_partition(struct gendisk *, int);
1494 unsigned long invalidate_mapping_pages(struct address_space *mapping,
1495 pgoff_t start, pgoff_t end);
1496 unsigned long invalidate_inode_pages(struct address_space *mapping);
1497 static inline void invalidate_remote_inode(struct inode *inode)
1498 {
1499 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
1500 S_ISLNK(inode->i_mode))
1501 invalidate_inode_pages(inode->i_mapping);
1502 }
1503 extern int invalidate_inode_pages2(struct address_space *mapping);
1504 extern int invalidate_inode_pages2_range(struct address_space *mapping,
1505 pgoff_t start, pgoff_t end);
1506 extern int write_inode_now(struct inode *, int);
1507 extern int filemap_fdatawrite(struct address_space *);
1508 extern int filemap_flush(struct address_space *);
1509 extern int filemap_fdatawait(struct address_space *);
1510 extern int filemap_write_and_wait(struct address_space *mapping);
1511 extern int filemap_write_and_wait_range(struct address_space *mapping,
1512 loff_t lstart, loff_t lend);
1513 extern int wait_on_page_writeback_range(struct address_space *mapping,
1514 pgoff_t start, pgoff_t end);
1515 extern int __filemap_fdatawrite_range(struct address_space *mapping,
1516 loff_t start, loff_t end, int sync_mode);
1517
1518 extern long do_fsync(struct file *file, int datasync);
1519 extern void sync_supers(void);
1520 extern void sync_filesystems(int wait);
1521 extern void emergency_sync(void);
1522 extern void emergency_remount(void);
1523 extern int do_remount_sb(struct super_block *sb, int flags,
1524 void *data, int force);
1525 extern sector_t bmap(struct inode *, sector_t);
1526 extern int notify_change(struct dentry *, struct iattr *);
1527 extern int permission(struct inode *, int, struct nameidata *);
1528 extern int generic_permission(struct inode *, int,
1529 int (*check_acl)(struct inode *, int));
1530
1531 extern int get_write_access(struct inode *);
1532 extern int deny_write_access(struct file *);
1533 static inline void put_write_access(struct inode * inode)
1534 {
1535 atomic_dec(&inode->i_writecount);
1536 }
1537 static inline void allow_write_access(struct file *file)
1538 {
1539 if (file)
1540 atomic_inc(&file->f_dentry->d_inode->i_writecount);
1541 }
1542 extern int do_pipe(int *);
1543
1544 extern int open_namei(int dfd, const char *, int, int, struct nameidata *);
1545 extern int may_open(struct nameidata *, int, int);
1546
1547 extern int kernel_read(struct file *, unsigned long, char *, unsigned long);
1548 extern struct file * open_exec(const char *);
1549
1550 /* fs/dcache.c -- generic fs support functions */
1551 extern int is_subdir(struct dentry *, struct dentry *);
1552 extern ino_t find_inode_number(struct dentry *, struct qstr *);
1553
1554 #include <linux/err.h>
1555
1556 /* needed for stackable file system support */
1557 extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
1558
1559 extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
1560
1561 extern void inode_init_once(struct inode *);
1562 extern void iput(struct inode *);
1563 extern struct inode * igrab(struct inode *);
1564 extern ino_t iunique(struct super_block *, ino_t);
1565 extern int inode_needs_sync(struct inode *inode);
1566 extern void generic_delete_inode(struct inode *inode);
1567 extern void generic_drop_inode(struct inode *inode);
1568
1569 extern struct inode *ilookup5_nowait(struct super_block *sb,
1570 unsigned long hashval, int (*test)(struct inode *, void *),
1571 void *data);
1572 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
1573 int (*test)(struct inode *, void *), void *data);
1574 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
1575
1576 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
1577 extern struct inode * iget_locked(struct super_block *, unsigned long);
1578 extern void unlock_new_inode(struct inode *);
1579
1580 static inline struct inode *iget(struct super_block *sb, unsigned long ino)
1581 {
1582 struct inode *inode = iget_locked(sb, ino);
1583
1584 if (inode && (inode->i_state & I_NEW)) {
1585 sb->s_op->read_inode(inode);
1586 unlock_new_inode(inode);
1587 }
1588
1589 return inode;
1590 }
1591
1592 extern void __iget(struct inode * inode);
1593 extern void clear_inode(struct inode *);
1594 extern void destroy_inode(struct inode *);
1595 extern struct inode *new_inode(struct super_block *);
1596 extern int remove_suid(struct dentry *);
1597 extern void remove_dquot_ref(struct super_block *, int, struct list_head *);
1598
1599 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
1600 extern void remove_inode_hash(struct inode *);
1601 static inline void insert_inode_hash(struct inode *inode) {
1602 __insert_inode_hash(inode, inode->i_ino);
1603 }
1604
1605 extern struct file * get_empty_filp(void);
1606 extern void file_move(struct file *f, struct list_head *list);
1607 extern void file_kill(struct file *f);
1608 struct bio;
1609 extern void submit_bio(int, struct bio *);
1610 extern int bdev_read_only(struct block_device *);
1611 extern int set_blocksize(struct block_device *, int);
1612 extern int sb_set_blocksize(struct super_block *, int);
1613 extern int sb_min_blocksize(struct super_block *, int);
1614
1615 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
1616 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
1617 extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
1618 extern int file_send_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
1619 extern ssize_t generic_file_read(struct file *, char __user *, size_t, loff_t *);
1620 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
1621 extern ssize_t generic_file_write(struct file *, const char __user *, size_t, loff_t *);
1622 extern ssize_t generic_file_aio_read(struct kiocb *, char __user *, size_t, loff_t);
1623 extern ssize_t __generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t *);
1624 extern ssize_t generic_file_aio_write(struct kiocb *, const char __user *, size_t, loff_t);
1625 extern ssize_t generic_file_aio_write_nolock(struct kiocb *, const struct iovec *,
1626 unsigned long, loff_t *);
1627 extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
1628 unsigned long *, loff_t, loff_t *, size_t, size_t);
1629 extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
1630 unsigned long, loff_t, loff_t *, size_t, ssize_t);
1631 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
1632 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
1633 ssize_t generic_file_write_nolock(struct file *file, const struct iovec *iov,
1634 unsigned long nr_segs, loff_t *ppos);
1635 extern ssize_t generic_file_sendfile(struct file *, loff_t *, size_t, read_actor_t, void *);
1636 extern void do_generic_mapping_read(struct address_space *mapping,
1637 struct file_ra_state *, struct file *,
1638 loff_t *, read_descriptor_t *, read_actor_t);
1639
1640 /* fs/splice.c */
1641 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
1642 struct pipe_inode_info *, size_t, unsigned int);
1643 extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
1644 struct file *, loff_t *, size_t, unsigned int);
1645 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
1646 struct file *out, loff_t *, size_t len, unsigned int flags);
1647 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
1648 size_t len, unsigned int flags);
1649
1650 extern void
1651 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
1652 extern ssize_t generic_file_readv(struct file *filp, const struct iovec *iov,
1653 unsigned long nr_segs, loff_t *ppos);
1654 ssize_t generic_file_writev(struct file *filp, const struct iovec *iov,
1655 unsigned long nr_segs, loff_t *ppos);
1656 extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
1657 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
1658 extern loff_t remote_llseek(struct file *file, loff_t offset, int origin);
1659 extern int generic_file_open(struct inode * inode, struct file * filp);
1660 extern int nonseekable_open(struct inode * inode, struct file * filp);
1661
1662 #ifdef CONFIG_FS_XIP
1663 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
1664 loff_t *ppos);
1665 extern ssize_t xip_file_sendfile(struct file *in_file, loff_t *ppos,
1666 size_t count, read_actor_t actor,
1667 void *target);
1668 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
1669 extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
1670 size_t len, loff_t *ppos);
1671 extern int xip_truncate_page(struct address_space *mapping, loff_t from);
1672 #else
1673 static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
1674 {
1675 return 0;
1676 }
1677 #endif
1678
1679 static inline void do_generic_file_read(struct file * filp, loff_t *ppos,
1680 read_descriptor_t * desc,
1681 read_actor_t actor)
1682 {
1683 do_generic_mapping_read(filp->f_mapping,
1684 &filp->f_ra,
1685 filp,
1686 ppos,
1687 desc,
1688 actor);
1689 }
1690
1691 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
1692 struct block_device *bdev, const struct iovec *iov, loff_t offset,
1693 unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
1694 int lock_type);
1695
1696 enum {
1697 DIO_LOCKING = 1, /* need locking between buffered and direct access */
1698 DIO_NO_LOCKING, /* bdev; no locking at all between buffered/direct */
1699 DIO_OWN_LOCKING, /* filesystem locks buffered and direct internally */
1700 };
1701
1702 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
1703 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1704 loff_t offset, unsigned long nr_segs, get_block_t get_block,
1705 dio_iodone_t end_io)
1706 {
1707 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1708 nr_segs, get_block, end_io, DIO_LOCKING);
1709 }
1710
1711 static inline ssize_t blockdev_direct_IO_no_locking(int rw, struct kiocb *iocb,
1712 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1713 loff_t offset, unsigned long nr_segs, get_block_t get_block,
1714 dio_iodone_t end_io)
1715 {
1716 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1717 nr_segs, get_block, end_io, DIO_NO_LOCKING);
1718 }
1719
1720 static inline ssize_t blockdev_direct_IO_own_locking(int rw, struct kiocb *iocb,
1721 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1722 loff_t offset, unsigned long nr_segs, get_block_t get_block,
1723 dio_iodone_t end_io)
1724 {
1725 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1726 nr_segs, get_block, end_io, DIO_OWN_LOCKING);
1727 }
1728
1729 extern const struct file_operations generic_ro_fops;
1730
1731 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
1732
1733 extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
1734 extern int vfs_follow_link(struct nameidata *, const char *);
1735 extern int page_readlink(struct dentry *, char __user *, int);
1736 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
1737 extern void page_put_link(struct dentry *, struct nameidata *, void *);
1738 extern int __page_symlink(struct inode *inode, const char *symname, int len,
1739 gfp_t gfp_mask);
1740 extern int page_symlink(struct inode *inode, const char *symname, int len);
1741 extern struct inode_operations page_symlink_inode_operations;
1742 extern int generic_readlink(struct dentry *, char __user *, int);
1743 extern void generic_fillattr(struct inode *, struct kstat *);
1744 extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1745 void inode_add_bytes(struct inode *inode, loff_t bytes);
1746 void inode_sub_bytes(struct inode *inode, loff_t bytes);
1747 loff_t inode_get_bytes(struct inode *inode);
1748 void inode_set_bytes(struct inode *inode, loff_t bytes);
1749
1750 extern int vfs_readdir(struct file *, filldir_t, void *);
1751
1752 extern int vfs_stat(char __user *, struct kstat *);
1753 extern int vfs_lstat(char __user *, struct kstat *);
1754 extern int vfs_stat_fd(int dfd, char __user *, struct kstat *);
1755 extern int vfs_lstat_fd(int dfd, char __user *, struct kstat *);
1756 extern int vfs_fstat(unsigned int, struct kstat *);
1757
1758 extern int vfs_ioctl(struct file *, unsigned int, unsigned int, unsigned long);
1759
1760 extern struct file_system_type *get_fs_type(const char *name);
1761 extern struct super_block *get_super(struct block_device *);
1762 extern struct super_block *user_get_super(dev_t);
1763 extern void drop_super(struct super_block *sb);
1764
1765 extern int dcache_dir_open(struct inode *, struct file *);
1766 extern int dcache_dir_close(struct inode *, struct file *);
1767 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
1768 extern int dcache_readdir(struct file *, void *, filldir_t);
1769 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1770 extern int simple_statfs(struct dentry *, struct kstatfs *);
1771 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
1772 extern int simple_unlink(struct inode *, struct dentry *);
1773 extern int simple_rmdir(struct inode *, struct dentry *);
1774 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1775 extern int simple_sync_file(struct file *, struct dentry *, int);
1776 extern int simple_empty(struct dentry *);
1777 extern int simple_readpage(struct file *file, struct page *page);
1778 extern int simple_prepare_write(struct file *file, struct page *page,
1779 unsigned offset, unsigned to);
1780 extern int simple_commit_write(struct file *file, struct page *page,
1781 unsigned offset, unsigned to);
1782
1783 extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
1784 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
1785 extern const struct file_operations simple_dir_operations;
1786 extern struct inode_operations simple_dir_inode_operations;
1787 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
1788 struct dentry *d_alloc_name(struct dentry *, const char *);
1789 extern int simple_fill_super(struct super_block *, int, struct tree_descr *);
1790 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
1791 extern void simple_release_fs(struct vfsmount **mount, int *count);
1792
1793 extern ssize_t simple_read_from_buffer(void __user *, size_t, loff_t *, const void *, size_t);
1794
1795 #ifdef CONFIG_MIGRATION
1796 extern int buffer_migrate_page(struct address_space *,
1797 struct page *, struct page *);
1798 #else
1799 #define buffer_migrate_page NULL
1800 #endif
1801
1802 extern int inode_change_ok(struct inode *, struct iattr *);
1803 extern int __must_check inode_setattr(struct inode *, struct iattr *);
1804
1805 extern void file_update_time(struct file *file);
1806
1807 static inline ino_t parent_ino(struct dentry *dentry)
1808 {
1809 ino_t res;
1810
1811 spin_lock(&dentry->d_lock);
1812 res = dentry->d_parent->d_inode->i_ino;
1813 spin_unlock(&dentry->d_lock);
1814 return res;
1815 }
1816
1817 /* kernel/fork.c */
1818 extern int unshare_files(void);
1819
1820 /* Transaction based IO helpers */
1821
1822 /*
1823 * An argresp is stored in an allocated page and holds the
1824 * size of the argument or response, along with its content
1825 */
1826 struct simple_transaction_argresp {
1827 ssize_t size;
1828 char data[0];
1829 };
1830
1831 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
1832
1833 char *simple_transaction_get(struct file *file, const char __user *buf,
1834 size_t size);
1835 ssize_t simple_transaction_read(struct file *file, char __user *buf,
1836 size_t size, loff_t *pos);
1837 int simple_transaction_release(struct inode *inode, struct file *file);
1838
1839 static inline void simple_transaction_set(struct file *file, size_t n)
1840 {
1841 struct simple_transaction_argresp *ar = file->private_data;
1842
1843 BUG_ON(n > SIMPLE_TRANSACTION_LIMIT);
1844
1845 /*
1846 * The barrier ensures that ar->size will really remain zero until
1847 * ar->data is ready for reading.
1848 */
1849 smp_mb();
1850 ar->size = n;
1851 }
1852
1853 /*
1854 * simple attribute files
1855 *
1856 * These attributes behave similar to those in sysfs:
1857 *
1858 * Writing to an attribute immediately sets a value, an open file can be
1859 * written to multiple times.
1860 *
1861 * Reading from an attribute creates a buffer from the value that might get
1862 * read with multiple read calls. When the attribute has been read
1863 * completely, no further read calls are possible until the file is opened
1864 * again.
1865 *
1866 * All attributes contain a text representation of a numeric value
1867 * that are accessed with the get() and set() functions.
1868 */
1869 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
1870 static int __fops ## _open(struct inode *inode, struct file *file) \
1871 { \
1872 __simple_attr_check_format(__fmt, 0ull); \
1873 return simple_attr_open(inode, file, __get, __set, __fmt); \
1874 } \
1875 static struct file_operations __fops = { \
1876 .owner = THIS_MODULE, \
1877 .open = __fops ## _open, \
1878 .release = simple_attr_close, \
1879 .read = simple_attr_read, \
1880 .write = simple_attr_write, \
1881 };
1882
1883 static inline void __attribute__((format(printf, 1, 2)))
1884 __simple_attr_check_format(const char *fmt, ...)
1885 {
1886 /* don't do anything, just let the compiler check the arguments; */
1887 }
1888
1889 int simple_attr_open(struct inode *inode, struct file *file,
1890 u64 (*get)(void *), void (*set)(void *, u64),
1891 const char *fmt);
1892 int simple_attr_close(struct inode *inode, struct file *file);
1893 ssize_t simple_attr_read(struct file *file, char __user *buf,
1894 size_t len, loff_t *ppos);
1895 ssize_t simple_attr_write(struct file *file, const char __user *buf,
1896 size_t len, loff_t *ppos);
1897
1898
1899 #ifdef CONFIG_SECURITY
1900 static inline char *alloc_secdata(void)
1901 {
1902 return (char *)get_zeroed_page(GFP_KERNEL);
1903 }
1904
1905 static inline void free_secdata(void *secdata)
1906 {
1907 free_page((unsigned long)secdata);
1908 }
1909 #else
1910 static inline char *alloc_secdata(void)
1911 {
1912 return (char *)1;
1913 }
1914
1915 static inline void free_secdata(void *secdata)
1916 { }
1917 #endif /* CONFIG_SECURITY */
1918
1919 #endif /* __KERNEL__ */
1920 #endif /* _LINUX_FS_H */