d57bc5df7225af64e2a59f2756f988e7680e6cdb
[GitHub/mt8127/android_kernel_alcatel_ttab.git] / include / linux / fs.h
1 #ifndef _LINUX_FS_H
2 #define _LINUX_FS_H
3
4
5 #include <linux/linkage.h>
6 #include <linux/wait.h>
7 #include <linux/kdev_t.h>
8 #include <linux/dcache.h>
9 #include <linux/path.h>
10 #include <linux/stat.h>
11 #include <linux/cache.h>
12 #include <linux/list.h>
13 #include <linux/radix-tree.h>
14 #include <linux/rbtree.h>
15 #include <linux/init.h>
16 #include <linux/pid.h>
17 #include <linux/bug.h>
18 #include <linux/mutex.h>
19 #include <linux/capability.h>
20 #include <linux/semaphore.h>
21 #include <linux/fiemap.h>
22 #include <linux/rculist_bl.h>
23 #include <linux/atomic.h>
24 #include <linux/shrinker.h>
25 #include <linux/migrate_mode.h>
26 #include <linux/uidgid.h>
27 #include <linux/lockdep.h>
28 #include <linux/percpu-rwsem.h>
29 #include <linux/blk_types.h>
30
31 #include <asm/byteorder.h>
32 #include <uapi/linux/fs.h>
33
34 struct export_operations;
35 struct hd_geometry;
36 struct iovec;
37 struct nameidata;
38 struct kiocb;
39 struct kobject;
40 struct pipe_inode_info;
41 struct poll_table_struct;
42 struct kstatfs;
43 struct vm_area_struct;
44 struct vfsmount;
45 struct cred;
46 struct swap_info_struct;
47 struct seq_file;
48
49 extern void __init inode_init(void);
50 extern void __init inode_init_early(void);
51 extern void __init files_init(unsigned long);
52
53 extern struct files_stat_struct files_stat;
54 extern unsigned long get_max_files(void);
55 extern int sysctl_nr_open;
56 extern struct inodes_stat_t inodes_stat;
57 extern int leases_enable, lease_break_time;
58 extern int sysctl_protected_symlinks;
59 extern int sysctl_protected_hardlinks;
60
61 struct buffer_head;
62 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
63 struct buffer_head *bh_result, int create);
64 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
65 ssize_t bytes, void *private, int ret,
66 bool is_async);
67
68 #define MAY_EXEC 0x00000001
69 #define MAY_WRITE 0x00000002
70 #define MAY_READ 0x00000004
71 #define MAY_APPEND 0x00000008
72 #define MAY_ACCESS 0x00000010
73 #define MAY_OPEN 0x00000020
74 #define MAY_CHDIR 0x00000040
75 /* called from RCU mode, don't block */
76 #define MAY_NOT_BLOCK 0x00000080
77
78 /*
79 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
80 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
81 */
82
83 /* file is open for reading */
84 #define FMODE_READ ((__force fmode_t)0x1)
85 /* file is open for writing */
86 #define FMODE_WRITE ((__force fmode_t)0x2)
87 /* file is seekable */
88 #define FMODE_LSEEK ((__force fmode_t)0x4)
89 /* file can be accessed using pread */
90 #define FMODE_PREAD ((__force fmode_t)0x8)
91 /* file can be accessed using pwrite */
92 #define FMODE_PWRITE ((__force fmode_t)0x10)
93 /* File is opened for execution with sys_execve / sys_uselib */
94 #define FMODE_EXEC ((__force fmode_t)0x20)
95 /* File is opened with O_NDELAY (only set for block devices) */
96 #define FMODE_NDELAY ((__force fmode_t)0x40)
97 /* File is opened with O_EXCL (only set for block devices) */
98 #define FMODE_EXCL ((__force fmode_t)0x80)
99 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
100 (specialy hack for floppy.c) */
101 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
102 /* 32bit hashes as llseek() offset (for directories) */
103 #define FMODE_32BITHASH ((__force fmode_t)0x200)
104 /* 64bit hashes as llseek() offset (for directories) */
105 #define FMODE_64BITHASH ((__force fmode_t)0x400)
106
107 /*
108 * Don't update ctime and mtime.
109 *
110 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
111 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
112 */
113 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
114
115 /* Expect random access pattern */
116 #define FMODE_RANDOM ((__force fmode_t)0x1000)
117
118 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
119 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
120
121 /* File is opened with O_PATH; almost nothing can be done with it */
122 #define FMODE_PATH ((__force fmode_t)0x4000)
123
124 /* File was opened by fanotify and shouldn't generate fanotify events */
125 #define FMODE_NONOTIFY ((__force fmode_t)0x1000000)
126
127 /*
128 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
129 * that indicates that they should check the contents of the iovec are
130 * valid, but not check the memory that the iovec elements
131 * points too.
132 */
133 #define CHECK_IOVEC_ONLY -1
134
135 /*
136 * The below are the various read and write types that we support. Some of
137 * them include behavioral modifiers that send information down to the
138 * block layer and IO scheduler. Terminology:
139 *
140 * The block layer uses device plugging to defer IO a little bit, in
141 * the hope that we will see more IO very shortly. This increases
142 * coalescing of adjacent IO and thus reduces the number of IOs we
143 * have to send to the device. It also allows for better queuing,
144 * if the IO isn't mergeable. If the caller is going to be waiting
145 * for the IO, then he must ensure that the device is unplugged so
146 * that the IO is dispatched to the driver.
147 *
148 * All IO is handled async in Linux. This is fine for background
149 * writes, but for reads or writes that someone waits for completion
150 * on, we want to notify the block layer and IO scheduler so that they
151 * know about it. That allows them to make better scheduling
152 * decisions. So when the below references 'sync' and 'async', it
153 * is referencing this priority hint.
154 *
155 * With that in mind, the available types are:
156 *
157 * READ A normal read operation. Device will be plugged.
158 * READ_SYNC A synchronous read. Device is not plugged, caller can
159 * immediately wait on this read without caring about
160 * unplugging.
161 * READA Used for read-ahead operations. Lower priority, and the
162 * block layer could (in theory) choose to ignore this
163 * request if it runs into resource problems.
164 * WRITE A normal async write. Device will be plugged.
165 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
166 * the hint that someone will be waiting on this IO
167 * shortly. The write equivalent of READ_SYNC.
168 * WRITE_ODIRECT Special case write for O_DIRECT only.
169 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
170 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
171 * non-volatile media on completion.
172 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
173 * by a cache flush and data is guaranteed to be on
174 * non-volatile media on completion.
175 *
176 */
177 #define RW_MASK REQ_WRITE
178 #define RWA_MASK REQ_RAHEAD
179
180 #define READ 0
181 #define WRITE RW_MASK
182 #define READA RWA_MASK
183 #define KERNEL_READ (READ|REQ_KERNEL)
184 #define KERNEL_WRITE (WRITE|REQ_KERNEL)
185
186 #define READ_SYNC (READ | REQ_SYNC)
187 #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
188 #define WRITE_ODIRECT (WRITE | REQ_SYNC)
189 #define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
190 #define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
191 #define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
192
193 /*
194 * Attribute flags. These should be or-ed together to figure out what
195 * has been changed!
196 */
197 #define ATTR_MODE (1 << 0)
198 #define ATTR_UID (1 << 1)
199 #define ATTR_GID (1 << 2)
200 #define ATTR_SIZE (1 << 3)
201 #define ATTR_ATIME (1 << 4)
202 #define ATTR_MTIME (1 << 5)
203 #define ATTR_CTIME (1 << 6)
204 #define ATTR_ATIME_SET (1 << 7)
205 #define ATTR_MTIME_SET (1 << 8)
206 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
207 #define ATTR_ATTR_FLAG (1 << 10)
208 #define ATTR_KILL_SUID (1 << 11)
209 #define ATTR_KILL_SGID (1 << 12)
210 #define ATTR_FILE (1 << 13)
211 #define ATTR_KILL_PRIV (1 << 14)
212 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
213 #define ATTR_TIMES_SET (1 << 16)
214
215 /*
216 * This is the Inode Attributes structure, used for notify_change(). It
217 * uses the above definitions as flags, to know which values have changed.
218 * Also, in this manner, a Filesystem can look at only the values it cares
219 * about. Basically, these are the attributes that the VFS layer can
220 * request to change from the FS layer.
221 *
222 * Derek Atkins <warlord@MIT.EDU> 94-10-20
223 */
224 struct iattr {
225 unsigned int ia_valid;
226 umode_t ia_mode;
227 kuid_t ia_uid;
228 kgid_t ia_gid;
229 loff_t ia_size;
230 struct timespec ia_atime;
231 struct timespec ia_mtime;
232 struct timespec ia_ctime;
233
234 /*
235 * Not an attribute, but an auxiliary info for filesystems wanting to
236 * implement an ftruncate() like method. NOTE: filesystem should
237 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
238 */
239 struct file *ia_file;
240 };
241
242 /*
243 * Includes for diskquotas.
244 */
245 #include <linux/quota.h>
246
247 /**
248 * enum positive_aop_returns - aop return codes with specific semantics
249 *
250 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
251 * completed, that the page is still locked, and
252 * should be considered active. The VM uses this hint
253 * to return the page to the active list -- it won't
254 * be a candidate for writeback again in the near
255 * future. Other callers must be careful to unlock
256 * the page if they get this return. Returned by
257 * writepage();
258 *
259 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
260 * unlocked it and the page might have been truncated.
261 * The caller should back up to acquiring a new page and
262 * trying again. The aop will be taking reasonable
263 * precautions not to livelock. If the caller held a page
264 * reference, it should drop it before retrying. Returned
265 * by readpage().
266 *
267 * address_space_operation functions return these large constants to indicate
268 * special semantics to the caller. These are much larger than the bytes in a
269 * page to allow for functions that return the number of bytes operated on in a
270 * given page.
271 */
272
273 enum positive_aop_returns {
274 AOP_WRITEPAGE_ACTIVATE = 0x80000,
275 AOP_TRUNCATED_PAGE = 0x80001,
276 };
277
278 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
279 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
280 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
281 * helper code (eg buffer layer)
282 * to clear GFP_FS from alloc */
283
284 /*
285 * oh the beauties of C type declarations.
286 */
287 struct page;
288 struct address_space;
289 struct writeback_control;
290
291 struct iov_iter {
292 const struct iovec *iov;
293 unsigned long nr_segs;
294 size_t iov_offset;
295 size_t count;
296 };
297
298 size_t iov_iter_copy_from_user_atomic(struct page *page,
299 struct iov_iter *i, unsigned long offset, size_t bytes);
300 size_t iov_iter_copy_from_user(struct page *page,
301 struct iov_iter *i, unsigned long offset, size_t bytes);
302 void iov_iter_advance(struct iov_iter *i, size_t bytes);
303 int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes);
304 size_t iov_iter_single_seg_count(const struct iov_iter *i);
305
306 static inline void iov_iter_init(struct iov_iter *i,
307 const struct iovec *iov, unsigned long nr_segs,
308 size_t count, size_t written)
309 {
310 i->iov = iov;
311 i->nr_segs = nr_segs;
312 i->iov_offset = 0;
313 i->count = count + written;
314
315 iov_iter_advance(i, written);
316 }
317
318 static inline size_t iov_iter_count(struct iov_iter *i)
319 {
320 return i->count;
321 }
322
323 /*
324 * "descriptor" for what we're up to with a read.
325 * This allows us to use the same read code yet
326 * have multiple different users of the data that
327 * we read from a file.
328 *
329 * The simplest case just copies the data to user
330 * mode.
331 */
332 typedef struct {
333 size_t written;
334 size_t count;
335 union {
336 char __user *buf;
337 void *data;
338 } arg;
339 int error;
340 } read_descriptor_t;
341
342 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
343 unsigned long, unsigned long);
344
345 struct address_space_operations {
346 int (*writepage)(struct page *page, struct writeback_control *wbc);
347 int (*readpage)(struct file *, struct page *);
348
349 /* Write back some dirty pages from this mapping. */
350 int (*writepages)(struct address_space *, struct writeback_control *);
351
352 /* Set a page dirty. Return true if this dirtied it */
353 int (*set_page_dirty)(struct page *page);
354
355 int (*readpages)(struct file *filp, struct address_space *mapping,
356 struct list_head *pages, unsigned nr_pages);
357
358 int (*write_begin)(struct file *, struct address_space *mapping,
359 loff_t pos, unsigned len, unsigned flags,
360 struct page **pagep, void **fsdata);
361 int (*write_end)(struct file *, struct address_space *mapping,
362 loff_t pos, unsigned len, unsigned copied,
363 struct page *page, void *fsdata);
364
365 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
366 sector_t (*bmap)(struct address_space *, sector_t);
367 void (*invalidatepage) (struct page *, unsigned long);
368 int (*releasepage) (struct page *, gfp_t);
369 void (*freepage)(struct page *);
370 ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
371 loff_t offset, unsigned long nr_segs);
372 int (*get_xip_mem)(struct address_space *, pgoff_t, int,
373 void **, unsigned long *);
374 /*
375 * migrate the contents of a page to the specified target. If sync
376 * is false, it must not block.
377 */
378 int (*migratepage) (struct address_space *,
379 struct page *, struct page *, enum migrate_mode);
380 int (*launder_page) (struct page *);
381 int (*is_partially_uptodate) (struct page *, read_descriptor_t *,
382 unsigned long);
383 int (*error_remove_page)(struct address_space *, struct page *);
384
385 /* swapfile support */
386 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
387 sector_t *span);
388 void (*swap_deactivate)(struct file *file);
389 };
390
391 extern const struct address_space_operations empty_aops;
392
393 /*
394 * pagecache_write_begin/pagecache_write_end must be used by general code
395 * to write into the pagecache.
396 */
397 int pagecache_write_begin(struct file *, struct address_space *mapping,
398 loff_t pos, unsigned len, unsigned flags,
399 struct page **pagep, void **fsdata);
400
401 int pagecache_write_end(struct file *, struct address_space *mapping,
402 loff_t pos, unsigned len, unsigned copied,
403 struct page *page, void *fsdata);
404
405 struct backing_dev_info;
406 struct address_space {
407 struct inode *host; /* owner: inode, block_device */
408 struct radix_tree_root page_tree; /* radix tree of all pages */
409 spinlock_t tree_lock; /* and lock protecting it */
410 unsigned int i_mmap_writable;/* count VM_SHARED mappings */
411 struct rb_root i_mmap; /* tree of private and shared mappings */
412 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
413 struct mutex i_mmap_mutex; /* protect tree, count, list */
414 /* Protected by tree_lock together with the radix tree */
415 unsigned long nrpages; /* number of total pages */
416 pgoff_t writeback_index;/* writeback starts here */
417 const struct address_space_operations *a_ops; /* methods */
418 unsigned long flags; /* error bits/gfp mask */
419 struct backing_dev_info *backing_dev_info; /* device readahead, etc */
420 spinlock_t private_lock; /* for use by the address_space */
421 struct list_head private_list; /* ditto */
422 void *private_data; /* ditto */
423 } __attribute__((aligned(sizeof(long))));
424 /*
425 * On most architectures that alignment is already the case; but
426 * must be enforced here for CRIS, to let the least significant bit
427 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
428 */
429 struct request_queue;
430
431 struct block_device {
432 dev_t bd_dev; /* not a kdev_t - it's a search key */
433 int bd_openers;
434 struct inode * bd_inode; /* will die */
435 struct super_block * bd_super;
436 struct mutex bd_mutex; /* open/close mutex */
437 struct list_head bd_inodes;
438 void * bd_claiming;
439 void * bd_holder;
440 int bd_holders;
441 bool bd_write_holder;
442 #ifdef CONFIG_SYSFS
443 struct list_head bd_holder_disks;
444 #endif
445 struct block_device * bd_contains;
446 unsigned bd_block_size;
447 struct hd_struct * bd_part;
448 /* number of times partitions within this device have been opened. */
449 unsigned bd_part_count;
450 int bd_invalidated;
451 struct gendisk * bd_disk;
452 struct request_queue * bd_queue;
453 struct list_head bd_list;
454 /*
455 * Private data. You must have bd_claim'ed the block_device
456 * to use this. NOTE: bd_claim allows an owner to claim
457 * the same device multiple times, the owner must take special
458 * care to not mess up bd_private for that case.
459 */
460 unsigned long bd_private;
461
462 /* The counter of freeze processes */
463 int bd_fsfreeze_count;
464 /* Mutex for freeze */
465 struct mutex bd_fsfreeze_mutex;
466 };
467
468 /*
469 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
470 * radix trees
471 */
472 #define PAGECACHE_TAG_DIRTY 0
473 #define PAGECACHE_TAG_WRITEBACK 1
474 #define PAGECACHE_TAG_TOWRITE 2
475
476 int mapping_tagged(struct address_space *mapping, int tag);
477
478 /*
479 * Might pages of this file be mapped into userspace?
480 */
481 static inline int mapping_mapped(struct address_space *mapping)
482 {
483 return !RB_EMPTY_ROOT(&mapping->i_mmap) ||
484 !list_empty(&mapping->i_mmap_nonlinear);
485 }
486
487 /*
488 * Might pages of this file have been modified in userspace?
489 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
490 * marks vma as VM_SHARED if it is shared, and the file was opened for
491 * writing i.e. vma may be mprotected writable even if now readonly.
492 */
493 static inline int mapping_writably_mapped(struct address_space *mapping)
494 {
495 return mapping->i_mmap_writable != 0;
496 }
497
498 /*
499 * Use sequence counter to get consistent i_size on 32-bit processors.
500 */
501 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
502 #include <linux/seqlock.h>
503 #define __NEED_I_SIZE_ORDERED
504 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
505 #else
506 #define i_size_ordered_init(inode) do { } while (0)
507 #endif
508
509 struct posix_acl;
510 #define ACL_NOT_CACHED ((void *)(-1))
511
512 #define IOP_FASTPERM 0x0001
513 #define IOP_LOOKUP 0x0002
514 #define IOP_NOFOLLOW 0x0004
515
516 /*
517 * Keep mostly read-only and often accessed (especially for
518 * the RCU path lookup and 'stat' data) fields at the beginning
519 * of the 'struct inode'
520 */
521 struct inode {
522 umode_t i_mode;
523 unsigned short i_opflags;
524 kuid_t i_uid;
525 kgid_t i_gid;
526 unsigned int i_flags;
527
528 #ifdef CONFIG_FS_POSIX_ACL
529 struct posix_acl *i_acl;
530 struct posix_acl *i_default_acl;
531 #endif
532
533 const struct inode_operations *i_op;
534 struct super_block *i_sb;
535 struct address_space *i_mapping;
536
537 #ifdef CONFIG_SECURITY
538 void *i_security;
539 #endif
540
541 /* Stat data, not accessed from path walking */
542 unsigned long i_ino;
543 /*
544 * Filesystems may only read i_nlink directly. They shall use the
545 * following functions for modification:
546 *
547 * (set|clear|inc|drop)_nlink
548 * inode_(inc|dec)_link_count
549 */
550 union {
551 const unsigned int i_nlink;
552 unsigned int __i_nlink;
553 };
554 dev_t i_rdev;
555 loff_t i_size;
556 struct timespec i_atime;
557 struct timespec i_mtime;
558 struct timespec i_ctime;
559 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
560 unsigned short i_bytes;
561 unsigned int i_blkbits;
562 blkcnt_t i_blocks;
563
564 #ifdef __NEED_I_SIZE_ORDERED
565 seqcount_t i_size_seqcount;
566 #endif
567
568 /* Misc */
569 unsigned long i_state;
570 struct mutex i_mutex;
571
572 unsigned long dirtied_when; /* jiffies of first dirtying */
573
574 struct hlist_node i_hash;
575 struct list_head i_wb_list; /* backing dev IO list */
576 struct list_head i_lru; /* inode LRU list */
577 struct list_head i_sb_list;
578 union {
579 struct hlist_head i_dentry;
580 struct rcu_head i_rcu;
581 };
582 u64 i_version;
583 atomic_t i_count;
584 atomic_t i_dio_count;
585 atomic_t i_writecount;
586 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
587 struct file_lock *i_flock;
588 struct address_space i_data;
589 #ifdef CONFIG_QUOTA
590 struct dquot *i_dquot[MAXQUOTAS];
591 #endif
592 struct list_head i_devices;
593 union {
594 struct pipe_inode_info *i_pipe;
595 struct block_device *i_bdev;
596 struct cdev *i_cdev;
597 };
598
599 __u32 i_generation;
600
601 #ifdef CONFIG_FSNOTIFY
602 __u32 i_fsnotify_mask; /* all events this inode cares about */
603 struct hlist_head i_fsnotify_marks;
604 #endif
605
606 #ifdef CONFIG_IMA
607 atomic_t i_readcount; /* struct files open RO */
608 #endif
609 void *i_private; /* fs or device private pointer */
610 };
611
612 static inline int inode_unhashed(struct inode *inode)
613 {
614 return hlist_unhashed(&inode->i_hash);
615 }
616
617 /*
618 * inode->i_mutex nesting subclasses for the lock validator:
619 *
620 * 0: the object of the current VFS operation
621 * 1: parent
622 * 2: child/target
623 * 3: quota file
624 *
625 * The locking order between these classes is
626 * parent -> child -> normal -> xattr -> quota
627 */
628 enum inode_i_mutex_lock_class
629 {
630 I_MUTEX_NORMAL,
631 I_MUTEX_PARENT,
632 I_MUTEX_CHILD,
633 I_MUTEX_XATTR,
634 I_MUTEX_QUOTA
635 };
636
637 /*
638 * NOTE: in a 32bit arch with a preemptable kernel and
639 * an UP compile the i_size_read/write must be atomic
640 * with respect to the local cpu (unlike with preempt disabled),
641 * but they don't need to be atomic with respect to other cpus like in
642 * true SMP (so they need either to either locally disable irq around
643 * the read or for example on x86 they can be still implemented as a
644 * cmpxchg8b without the need of the lock prefix). For SMP compiles
645 * and 64bit archs it makes no difference if preempt is enabled or not.
646 */
647 static inline loff_t i_size_read(const struct inode *inode)
648 {
649 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
650 loff_t i_size;
651 unsigned int seq;
652
653 do {
654 seq = read_seqcount_begin(&inode->i_size_seqcount);
655 i_size = inode->i_size;
656 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
657 return i_size;
658 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
659 loff_t i_size;
660
661 preempt_disable();
662 i_size = inode->i_size;
663 preempt_enable();
664 return i_size;
665 #else
666 return inode->i_size;
667 #endif
668 }
669
670 /*
671 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
672 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
673 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
674 */
675 static inline void i_size_write(struct inode *inode, loff_t i_size)
676 {
677 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
678 preempt_disable();
679 write_seqcount_begin(&inode->i_size_seqcount);
680 inode->i_size = i_size;
681 write_seqcount_end(&inode->i_size_seqcount);
682 preempt_enable();
683 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
684 preempt_disable();
685 inode->i_size = i_size;
686 preempt_enable();
687 #else
688 inode->i_size = i_size;
689 #endif
690 }
691
692 /* Helper functions so that in most cases filesystems will
693 * not need to deal directly with kuid_t and kgid_t and can
694 * instead deal with the raw numeric values that are stored
695 * in the filesystem.
696 */
697 static inline uid_t i_uid_read(const struct inode *inode)
698 {
699 return from_kuid(&init_user_ns, inode->i_uid);
700 }
701
702 static inline gid_t i_gid_read(const struct inode *inode)
703 {
704 return from_kgid(&init_user_ns, inode->i_gid);
705 }
706
707 static inline void i_uid_write(struct inode *inode, uid_t uid)
708 {
709 inode->i_uid = make_kuid(&init_user_ns, uid);
710 }
711
712 static inline void i_gid_write(struct inode *inode, gid_t gid)
713 {
714 inode->i_gid = make_kgid(&init_user_ns, gid);
715 }
716
717 static inline unsigned iminor(const struct inode *inode)
718 {
719 return MINOR(inode->i_rdev);
720 }
721
722 static inline unsigned imajor(const struct inode *inode)
723 {
724 return MAJOR(inode->i_rdev);
725 }
726
727 extern struct block_device *I_BDEV(struct inode *inode);
728
729 struct fown_struct {
730 rwlock_t lock; /* protects pid, uid, euid fields */
731 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
732 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
733 kuid_t uid, euid; /* uid/euid of process setting the owner */
734 int signum; /* posix.1b rt signal to be delivered on IO */
735 };
736
737 /*
738 * Track a single file's readahead state
739 */
740 struct file_ra_state {
741 pgoff_t start; /* where readahead started */
742 unsigned int size; /* # of readahead pages */
743 unsigned int async_size; /* do asynchronous readahead when
744 there are only # of pages ahead */
745
746 unsigned int ra_pages; /* Maximum readahead window */
747 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
748 loff_t prev_pos; /* Cache last read() position */
749 };
750
751 /*
752 * Check if @index falls in the readahead windows.
753 */
754 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
755 {
756 return (index >= ra->start &&
757 index < ra->start + ra->size);
758 }
759
760 #define FILE_MNT_WRITE_TAKEN 1
761 #define FILE_MNT_WRITE_RELEASED 2
762
763 struct file {
764 /*
765 * fu_list becomes invalid after file_free is called and queued via
766 * fu_rcuhead for RCU freeing
767 */
768 union {
769 struct list_head fu_list;
770 struct rcu_head fu_rcuhead;
771 } f_u;
772 struct path f_path;
773 #define f_dentry f_path.dentry
774 struct inode *f_inode; /* cached value */
775 const struct file_operations *f_op;
776
777 /*
778 * Protects f_ep_links, f_flags, f_pos vs i_size in lseek SEEK_CUR.
779 * Must not be taken from IRQ context.
780 */
781 spinlock_t f_lock;
782 #ifdef CONFIG_SMP
783 int f_sb_list_cpu;
784 #endif
785 atomic_long_t f_count;
786 unsigned int f_flags;
787 fmode_t f_mode;
788 loff_t f_pos;
789 struct fown_struct f_owner;
790 const struct cred *f_cred;
791 struct file_ra_state f_ra;
792
793 u64 f_version;
794 #ifdef CONFIG_SECURITY
795 void *f_security;
796 #endif
797 /* needed for tty driver, and maybe others */
798 void *private_data;
799
800 #ifdef CONFIG_EPOLL
801 /* Used by fs/eventpoll.c to link all the hooks to this file */
802 struct list_head f_ep_links;
803 struct list_head f_tfile_llink;
804 #endif /* #ifdef CONFIG_EPOLL */
805 struct address_space *f_mapping;
806 #ifdef CONFIG_DEBUG_WRITECOUNT
807 unsigned long f_mnt_write_state;
808 #endif
809 };
810
811 struct file_handle {
812 __u32 handle_bytes;
813 int handle_type;
814 /* file identifier */
815 unsigned char f_handle[0];
816 };
817
818 static inline struct file *get_file(struct file *f)
819 {
820 atomic_long_inc(&f->f_count);
821 return f;
822 }
823 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
824 #define file_count(x) atomic_long_read(&(x)->f_count)
825
826 #ifdef CONFIG_DEBUG_WRITECOUNT
827 static inline void file_take_write(struct file *f)
828 {
829 WARN_ON(f->f_mnt_write_state != 0);
830 f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN;
831 }
832 static inline void file_release_write(struct file *f)
833 {
834 f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED;
835 }
836 static inline void file_reset_write(struct file *f)
837 {
838 f->f_mnt_write_state = 0;
839 }
840 static inline void file_check_state(struct file *f)
841 {
842 /*
843 * At this point, either both or neither of these bits
844 * should be set.
845 */
846 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN);
847 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED);
848 }
849 static inline int file_check_writeable(struct file *f)
850 {
851 if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN)
852 return 0;
853 printk(KERN_WARNING "writeable file with no "
854 "mnt_want_write()\n");
855 WARN_ON(1);
856 return -EINVAL;
857 }
858 #else /* !CONFIG_DEBUG_WRITECOUNT */
859 static inline void file_take_write(struct file *filp) {}
860 static inline void file_release_write(struct file *filp) {}
861 static inline void file_reset_write(struct file *filp) {}
862 static inline void file_check_state(struct file *filp) {}
863 static inline int file_check_writeable(struct file *filp)
864 {
865 return 0;
866 }
867 #endif /* CONFIG_DEBUG_WRITECOUNT */
868
869 #define MAX_NON_LFS ((1UL<<31) - 1)
870
871 /* Page cache limit. The filesystems should put that into their s_maxbytes
872 limits, otherwise bad things can happen in VM. */
873 #if BITS_PER_LONG==32
874 #define MAX_LFS_FILESIZE (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
875 #elif BITS_PER_LONG==64
876 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
877 #endif
878
879 #define FL_POSIX 1
880 #define FL_FLOCK 2
881 #define FL_ACCESS 8 /* not trying to lock, just looking */
882 #define FL_EXISTS 16 /* when unlocking, test for existence */
883 #define FL_LEASE 32 /* lease held on this file */
884 #define FL_CLOSE 64 /* unlock on close */
885 #define FL_SLEEP 128 /* A blocking lock */
886 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
887 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
888
889 /*
890 * Special return value from posix_lock_file() and vfs_lock_file() for
891 * asynchronous locking.
892 */
893 #define FILE_LOCK_DEFERRED 1
894
895 /*
896 * The POSIX file lock owner is determined by
897 * the "struct files_struct" in the thread group
898 * (or NULL for no owner - BSD locks).
899 *
900 * Lockd stuffs a "host" pointer into this.
901 */
902 typedef struct files_struct *fl_owner_t;
903
904 struct file_lock_operations {
905 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
906 void (*fl_release_private)(struct file_lock *);
907 };
908
909 struct lock_manager_operations {
910 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
911 void (*lm_notify)(struct file_lock *); /* unblock callback */
912 int (*lm_grant)(struct file_lock *, struct file_lock *, int);
913 void (*lm_break)(struct file_lock *);
914 int (*lm_change)(struct file_lock **, int);
915 };
916
917 struct lock_manager {
918 struct list_head list;
919 };
920
921 struct net;
922 void locks_start_grace(struct net *, struct lock_manager *);
923 void locks_end_grace(struct lock_manager *);
924 int locks_in_grace(struct net *);
925
926 /* that will die - we need it for nfs_lock_info */
927 #include <linux/nfs_fs_i.h>
928
929 struct file_lock {
930 struct file_lock *fl_next; /* singly linked list for this inode */
931 struct list_head fl_link; /* doubly linked list of all locks */
932 struct list_head fl_block; /* circular list of blocked processes */
933 fl_owner_t fl_owner;
934 unsigned int fl_flags;
935 unsigned char fl_type;
936 unsigned int fl_pid;
937 struct pid *fl_nspid;
938 wait_queue_head_t fl_wait;
939 struct file *fl_file;
940 loff_t fl_start;
941 loff_t fl_end;
942
943 struct fasync_struct * fl_fasync; /* for lease break notifications */
944 /* for lease breaks: */
945 unsigned long fl_break_time;
946 unsigned long fl_downgrade_time;
947
948 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
949 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
950 union {
951 struct nfs_lock_info nfs_fl;
952 struct nfs4_lock_info nfs4_fl;
953 struct {
954 struct list_head link; /* link in AFS vnode's pending_locks list */
955 int state; /* state of grant or error if -ve */
956 } afs;
957 } fl_u;
958 };
959
960 /* The following constant reflects the upper bound of the file/locking space */
961 #ifndef OFFSET_MAX
962 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
963 #define OFFSET_MAX INT_LIMIT(loff_t)
964 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
965 #endif
966
967 #include <linux/fcntl.h>
968
969 extern void send_sigio(struct fown_struct *fown, int fd, int band);
970
971 #ifdef CONFIG_FILE_LOCKING
972 extern int fcntl_getlk(struct file *, struct flock __user *);
973 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
974 struct flock __user *);
975
976 #if BITS_PER_LONG == 32
977 extern int fcntl_getlk64(struct file *, struct flock64 __user *);
978 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
979 struct flock64 __user *);
980 #endif
981
982 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
983 extern int fcntl_getlease(struct file *filp);
984
985 /* fs/locks.c */
986 void locks_free_lock(struct file_lock *fl);
987 extern void locks_init_lock(struct file_lock *);
988 extern struct file_lock * locks_alloc_lock(void);
989 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
990 extern void __locks_copy_lock(struct file_lock *, const struct file_lock *);
991 extern void locks_remove_posix(struct file *, fl_owner_t);
992 extern void locks_remove_flock(struct file *);
993 extern void locks_release_private(struct file_lock *);
994 extern void posix_test_lock(struct file *, struct file_lock *);
995 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
996 extern int posix_lock_file_wait(struct file *, struct file_lock *);
997 extern int posix_unblock_lock(struct file *, struct file_lock *);
998 extern int vfs_test_lock(struct file *, struct file_lock *);
999 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1000 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1001 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
1002 extern int __break_lease(struct inode *inode, unsigned int flags);
1003 extern void lease_get_mtime(struct inode *, struct timespec *time);
1004 extern int generic_setlease(struct file *, long, struct file_lock **);
1005 extern int vfs_setlease(struct file *, long, struct file_lock **);
1006 extern int lease_modify(struct file_lock **, int);
1007 extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
1008 extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
1009 extern void locks_delete_block(struct file_lock *waiter);
1010 extern void lock_flocks(void);
1011 extern void unlock_flocks(void);
1012 #else /* !CONFIG_FILE_LOCKING */
1013 static inline int fcntl_getlk(struct file *file, struct flock __user *user)
1014 {
1015 return -EINVAL;
1016 }
1017
1018 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1019 unsigned int cmd, struct flock __user *user)
1020 {
1021 return -EACCES;
1022 }
1023
1024 #if BITS_PER_LONG == 32
1025 static inline int fcntl_getlk64(struct file *file, struct flock64 __user *user)
1026 {
1027 return -EINVAL;
1028 }
1029
1030 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1031 unsigned int cmd, struct flock64 __user *user)
1032 {
1033 return -EACCES;
1034 }
1035 #endif
1036 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1037 {
1038 return 0;
1039 }
1040
1041 static inline int fcntl_getlease(struct file *filp)
1042 {
1043 return 0;
1044 }
1045
1046 static inline void locks_init_lock(struct file_lock *fl)
1047 {
1048 return;
1049 }
1050
1051 static inline void __locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1052 {
1053 return;
1054 }
1055
1056 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1057 {
1058 return;
1059 }
1060
1061 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1062 {
1063 return;
1064 }
1065
1066 static inline void locks_remove_flock(struct file *filp)
1067 {
1068 return;
1069 }
1070
1071 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1072 {
1073 return;
1074 }
1075
1076 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1077 struct file_lock *conflock)
1078 {
1079 return -ENOLCK;
1080 }
1081
1082 static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1083 {
1084 return -ENOLCK;
1085 }
1086
1087 static inline int posix_unblock_lock(struct file *filp,
1088 struct file_lock *waiter)
1089 {
1090 return -ENOENT;
1091 }
1092
1093 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1094 {
1095 return 0;
1096 }
1097
1098 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1099 struct file_lock *fl, struct file_lock *conf)
1100 {
1101 return -ENOLCK;
1102 }
1103
1104 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1105 {
1106 return 0;
1107 }
1108
1109 static inline int flock_lock_file_wait(struct file *filp,
1110 struct file_lock *request)
1111 {
1112 return -ENOLCK;
1113 }
1114
1115 static inline int __break_lease(struct inode *inode, unsigned int mode)
1116 {
1117 return 0;
1118 }
1119
1120 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1121 {
1122 return;
1123 }
1124
1125 static inline int generic_setlease(struct file *filp, long arg,
1126 struct file_lock **flp)
1127 {
1128 return -EINVAL;
1129 }
1130
1131 static inline int vfs_setlease(struct file *filp, long arg,
1132 struct file_lock **lease)
1133 {
1134 return -EINVAL;
1135 }
1136
1137 static inline int lease_modify(struct file_lock **before, int arg)
1138 {
1139 return -EINVAL;
1140 }
1141
1142 static inline int lock_may_read(struct inode *inode, loff_t start,
1143 unsigned long len)
1144 {
1145 return 1;
1146 }
1147
1148 static inline int lock_may_write(struct inode *inode, loff_t start,
1149 unsigned long len)
1150 {
1151 return 1;
1152 }
1153
1154 static inline void locks_delete_block(struct file_lock *waiter)
1155 {
1156 }
1157
1158 static inline void lock_flocks(void)
1159 {
1160 }
1161
1162 static inline void unlock_flocks(void)
1163 {
1164 }
1165
1166 #endif /* !CONFIG_FILE_LOCKING */
1167
1168
1169 struct fasync_struct {
1170 spinlock_t fa_lock;
1171 int magic;
1172 int fa_fd;
1173 struct fasync_struct *fa_next; /* singly linked list */
1174 struct file *fa_file;
1175 struct rcu_head fa_rcu;
1176 };
1177
1178 #define FASYNC_MAGIC 0x4601
1179
1180 /* SMP safe fasync helpers: */
1181 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1182 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1183 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1184 extern struct fasync_struct *fasync_alloc(void);
1185 extern void fasync_free(struct fasync_struct *);
1186
1187 /* can be called from interrupts */
1188 extern void kill_fasync(struct fasync_struct **, int, int);
1189
1190 extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1191 extern int f_setown(struct file *filp, unsigned long arg, int force);
1192 extern void f_delown(struct file *filp);
1193 extern pid_t f_getown(struct file *filp);
1194 extern int send_sigurg(struct fown_struct *fown);
1195
1196 struct mm_struct;
1197
1198 /*
1199 * Umount options
1200 */
1201
1202 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1203 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1204 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1205 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1206 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1207
1208 extern struct list_head super_blocks;
1209 extern spinlock_t sb_lock;
1210
1211 /* Possible states of 'frozen' field */
1212 enum {
1213 SB_UNFROZEN = 0, /* FS is unfrozen */
1214 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1215 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1216 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1217 * internal threads if needed) */
1218 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1219 };
1220
1221 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1222
1223 struct sb_writers {
1224 /* Counters for counting writers at each level */
1225 struct percpu_counter counter[SB_FREEZE_LEVELS];
1226 wait_queue_head_t wait; /* queue for waiting for
1227 writers / faults to finish */
1228 int frozen; /* Is sb frozen? */
1229 wait_queue_head_t wait_unfrozen; /* queue for waiting for
1230 sb to be thawed */
1231 #ifdef CONFIG_DEBUG_LOCK_ALLOC
1232 struct lockdep_map lock_map[SB_FREEZE_LEVELS];
1233 #endif
1234 };
1235
1236 struct super_block {
1237 struct list_head s_list; /* Keep this first */
1238 dev_t s_dev; /* search index; _not_ kdev_t */
1239 unsigned char s_blocksize_bits;
1240 unsigned long s_blocksize;
1241 loff_t s_maxbytes; /* Max file size */
1242 struct file_system_type *s_type;
1243 const struct super_operations *s_op;
1244 const struct dquot_operations *dq_op;
1245 const struct quotactl_ops *s_qcop;
1246 const struct export_operations *s_export_op;
1247 unsigned long s_flags;
1248 unsigned long s_magic;
1249 struct dentry *s_root;
1250 struct rw_semaphore s_umount;
1251 int s_count;
1252 atomic_t s_active;
1253 #ifdef CONFIG_SECURITY
1254 void *s_security;
1255 #endif
1256 const struct xattr_handler **s_xattr;
1257
1258 struct list_head s_inodes; /* all inodes */
1259 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1260 #ifdef CONFIG_SMP
1261 struct list_head __percpu *s_files;
1262 #else
1263 struct list_head s_files;
1264 #endif
1265 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1266 /* s_dentry_lru, s_nr_dentry_unused protected by dcache.c lru locks */
1267 struct list_head s_dentry_lru; /* unused dentry lru */
1268 int s_nr_dentry_unused; /* # of dentry on lru */
1269
1270 /* s_inode_lru_lock protects s_inode_lru and s_nr_inodes_unused */
1271 spinlock_t s_inode_lru_lock ____cacheline_aligned_in_smp;
1272 struct list_head s_inode_lru; /* unused inode lru */
1273 int s_nr_inodes_unused; /* # of inodes on lru */
1274
1275 struct block_device *s_bdev;
1276 struct backing_dev_info *s_bdi;
1277 struct mtd_info *s_mtd;
1278 struct hlist_node s_instances;
1279 struct quota_info s_dquot; /* Diskquota specific options */
1280
1281 struct sb_writers s_writers;
1282
1283 char s_id[32]; /* Informational name */
1284 u8 s_uuid[16]; /* UUID */
1285
1286 void *s_fs_info; /* Filesystem private info */
1287 unsigned int s_max_links;
1288 fmode_t s_mode;
1289
1290 /* Granularity of c/m/atime in ns.
1291 Cannot be worse than a second */
1292 u32 s_time_gran;
1293
1294 /*
1295 * The next field is for VFS *only*. No filesystems have any business
1296 * even looking at it. You had been warned.
1297 */
1298 struct mutex s_vfs_rename_mutex; /* Kludge */
1299
1300 /*
1301 * Filesystem subtype. If non-empty the filesystem type field
1302 * in /proc/mounts will be "type.subtype"
1303 */
1304 char *s_subtype;
1305
1306 /*
1307 * Saved mount options for lazy filesystems using
1308 * generic_show_options()
1309 */
1310 char __rcu *s_options;
1311 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1312
1313 /*
1314 * Saved pool identifier for cleancache (-1 means none)
1315 */
1316 int cleancache_poolid;
1317
1318 struct shrinker s_shrink; /* per-sb shrinker handle */
1319
1320 /* Number of inodes with nlink == 0 but still referenced */
1321 atomic_long_t s_remove_count;
1322
1323 /* Being remounted read-only */
1324 int s_readonly_remount;
1325 };
1326
1327 /* superblock cache pruning functions */
1328 extern void prune_icache_sb(struct super_block *sb, int nr_to_scan);
1329 extern void prune_dcache_sb(struct super_block *sb, int nr_to_scan);
1330
1331 extern struct timespec current_fs_time(struct super_block *sb);
1332
1333 /*
1334 * Snapshotting support.
1335 */
1336
1337 void __sb_end_write(struct super_block *sb, int level);
1338 int __sb_start_write(struct super_block *sb, int level, bool wait);
1339
1340 /**
1341 * sb_end_write - drop write access to a superblock
1342 * @sb: the super we wrote to
1343 *
1344 * Decrement number of writers to the filesystem. Wake up possible waiters
1345 * wanting to freeze the filesystem.
1346 */
1347 static inline void sb_end_write(struct super_block *sb)
1348 {
1349 __sb_end_write(sb, SB_FREEZE_WRITE);
1350 }
1351
1352 /**
1353 * sb_end_pagefault - drop write access to a superblock from a page fault
1354 * @sb: the super we wrote to
1355 *
1356 * Decrement number of processes handling write page fault to the filesystem.
1357 * Wake up possible waiters wanting to freeze the filesystem.
1358 */
1359 static inline void sb_end_pagefault(struct super_block *sb)
1360 {
1361 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1362 }
1363
1364 /**
1365 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1366 * @sb: the super we wrote to
1367 *
1368 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1369 * waiters wanting to freeze the filesystem.
1370 */
1371 static inline void sb_end_intwrite(struct super_block *sb)
1372 {
1373 __sb_end_write(sb, SB_FREEZE_FS);
1374 }
1375
1376 /**
1377 * sb_start_write - get write access to a superblock
1378 * @sb: the super we write to
1379 *
1380 * When a process wants to write data or metadata to a file system (i.e. dirty
1381 * a page or an inode), it should embed the operation in a sb_start_write() -
1382 * sb_end_write() pair to get exclusion against file system freezing. This
1383 * function increments number of writers preventing freezing. If the file
1384 * system is already frozen, the function waits until the file system is
1385 * thawed.
1386 *
1387 * Since freeze protection behaves as a lock, users have to preserve
1388 * ordering of freeze protection and other filesystem locks. Generally,
1389 * freeze protection should be the outermost lock. In particular, we have:
1390 *
1391 * sb_start_write
1392 * -> i_mutex (write path, truncate, directory ops, ...)
1393 * -> s_umount (freeze_super, thaw_super)
1394 */
1395 static inline void sb_start_write(struct super_block *sb)
1396 {
1397 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1398 }
1399
1400 static inline int sb_start_write_trylock(struct super_block *sb)
1401 {
1402 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1403 }
1404
1405 /**
1406 * sb_start_pagefault - get write access to a superblock from a page fault
1407 * @sb: the super we write to
1408 *
1409 * When a process starts handling write page fault, it should embed the
1410 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1411 * exclusion against file system freezing. This is needed since the page fault
1412 * is going to dirty a page. This function increments number of running page
1413 * faults preventing freezing. If the file system is already frozen, the
1414 * function waits until the file system is thawed.
1415 *
1416 * Since page fault freeze protection behaves as a lock, users have to preserve
1417 * ordering of freeze protection and other filesystem locks. It is advised to
1418 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1419 * handling code implies lock dependency:
1420 *
1421 * mmap_sem
1422 * -> sb_start_pagefault
1423 */
1424 static inline void sb_start_pagefault(struct super_block *sb)
1425 {
1426 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1427 }
1428
1429 /*
1430 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1431 * @sb: the super we write to
1432 *
1433 * This is the third level of protection against filesystem freezing. It is
1434 * free for use by a filesystem. The only requirement is that it must rank
1435 * below sb_start_pagefault.
1436 *
1437 * For example filesystem can call sb_start_intwrite() when starting a
1438 * transaction which somewhat eases handling of freezing for internal sources
1439 * of filesystem changes (internal fs threads, discarding preallocation on file
1440 * close, etc.).
1441 */
1442 static inline void sb_start_intwrite(struct super_block *sb)
1443 {
1444 __sb_start_write(sb, SB_FREEZE_FS, true);
1445 }
1446
1447
1448 extern bool inode_owner_or_capable(const struct inode *inode);
1449
1450 /*
1451 * VFS helper functions..
1452 */
1453 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1454 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1455 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1456 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1457 extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
1458 extern int vfs_rmdir(struct inode *, struct dentry *);
1459 extern int vfs_unlink(struct inode *, struct dentry *);
1460 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1461
1462 /*
1463 * VFS dentry helper functions.
1464 */
1465 extern void dentry_unhash(struct dentry *dentry);
1466
1467 /*
1468 * VFS file helper functions.
1469 */
1470 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1471 umode_t mode);
1472 /*
1473 * VFS FS_IOC_FIEMAP helper definitions.
1474 */
1475 struct fiemap_extent_info {
1476 unsigned int fi_flags; /* Flags as passed from user */
1477 unsigned int fi_extents_mapped; /* Number of mapped extents */
1478 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1479 struct fiemap_extent __user *fi_extents_start; /* Start of
1480 fiemap_extent array */
1481 };
1482 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1483 u64 phys, u64 len, u32 flags);
1484 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1485
1486 /*
1487 * File types
1488 *
1489 * NOTE! These match bits 12..15 of stat.st_mode
1490 * (ie "(i_mode >> 12) & 15").
1491 */
1492 #define DT_UNKNOWN 0
1493 #define DT_FIFO 1
1494 #define DT_CHR 2
1495 #define DT_DIR 4
1496 #define DT_BLK 6
1497 #define DT_REG 8
1498 #define DT_LNK 10
1499 #define DT_SOCK 12
1500 #define DT_WHT 14
1501
1502 /*
1503 * This is the "filldir" function type, used by readdir() to let
1504 * the kernel specify what kind of dirent layout it wants to have.
1505 * This allows the kernel to read directories into kernel space or
1506 * to have different dirent layouts depending on the binary type.
1507 */
1508 typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
1509 struct block_device_operations;
1510
1511 /* These macros are for out of kernel modules to test that
1512 * the kernel supports the unlocked_ioctl and compat_ioctl
1513 * fields in struct file_operations. */
1514 #define HAVE_COMPAT_IOCTL 1
1515 #define HAVE_UNLOCKED_IOCTL 1
1516
1517 struct file_operations {
1518 struct module *owner;
1519 loff_t (*llseek) (struct file *, loff_t, int);
1520 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1521 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1522 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1523 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1524 int (*readdir) (struct file *, void *, filldir_t);
1525 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1526 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1527 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1528 int (*mmap) (struct file *, struct vm_area_struct *);
1529 int (*open) (struct inode *, struct file *);
1530 int (*flush) (struct file *, fl_owner_t id);
1531 int (*release) (struct inode *, struct file *);
1532 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1533 int (*aio_fsync) (struct kiocb *, int datasync);
1534 int (*fasync) (int, struct file *, int);
1535 int (*lock) (struct file *, int, struct file_lock *);
1536 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1537 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1538 int (*check_flags)(int);
1539 int (*flock) (struct file *, int, struct file_lock *);
1540 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1541 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1542 int (*setlease)(struct file *, long, struct file_lock **);
1543 long (*fallocate)(struct file *file, int mode, loff_t offset,
1544 loff_t len);
1545 int (*show_fdinfo)(struct seq_file *m, struct file *f);
1546 };
1547
1548 struct inode_operations {
1549 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1550 void * (*follow_link) (struct dentry *, struct nameidata *);
1551 int (*permission) (struct inode *, int);
1552 struct posix_acl * (*get_acl)(struct inode *, int);
1553
1554 int (*readlink) (struct dentry *, char __user *,int);
1555 void (*put_link) (struct dentry *, struct nameidata *, void *);
1556
1557 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1558 int (*link) (struct dentry *,struct inode *,struct dentry *);
1559 int (*unlink) (struct inode *,struct dentry *);
1560 int (*symlink) (struct inode *,struct dentry *,const char *);
1561 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1562 int (*rmdir) (struct inode *,struct dentry *);
1563 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1564 int (*rename) (struct inode *, struct dentry *,
1565 struct inode *, struct dentry *);
1566 int (*setattr) (struct dentry *, struct iattr *);
1567 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1568 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1569 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1570 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1571 int (*removexattr) (struct dentry *, const char *);
1572 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1573 u64 len);
1574 int (*update_time)(struct inode *, struct timespec *, int);
1575 int (*atomic_open)(struct inode *, struct dentry *,
1576 struct file *, unsigned open_flag,
1577 umode_t create_mode, int *opened);
1578 } ____cacheline_aligned;
1579
1580 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1581 unsigned long nr_segs, unsigned long fast_segs,
1582 struct iovec *fast_pointer,
1583 struct iovec **ret_pointer);
1584
1585 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1586 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1587 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1588 unsigned long, loff_t *);
1589 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1590 unsigned long, loff_t *);
1591
1592 struct super_operations {
1593 struct inode *(*alloc_inode)(struct super_block *sb);
1594 void (*destroy_inode)(struct inode *);
1595
1596 void (*dirty_inode) (struct inode *, int flags);
1597 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1598 int (*drop_inode) (struct inode *);
1599 void (*evict_inode) (struct inode *);
1600 void (*put_super) (struct super_block *);
1601 int (*sync_fs)(struct super_block *sb, int wait);
1602 int (*freeze_fs) (struct super_block *);
1603 int (*unfreeze_fs) (struct super_block *);
1604 int (*statfs) (struct dentry *, struct kstatfs *);
1605 int (*remount_fs) (struct super_block *, int *, char *);
1606 void (*umount_begin) (struct super_block *);
1607
1608 int (*show_options)(struct seq_file *, struct dentry *);
1609 int (*show_devname)(struct seq_file *, struct dentry *);
1610 int (*show_path)(struct seq_file *, struct dentry *);
1611 int (*show_stats)(struct seq_file *, struct dentry *);
1612 #ifdef CONFIG_QUOTA
1613 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1614 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1615 #endif
1616 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1617 int (*nr_cached_objects)(struct super_block *);
1618 void (*free_cached_objects)(struct super_block *, int);
1619 };
1620
1621 /*
1622 * Inode flags - they have no relation to superblock flags now
1623 */
1624 #define S_SYNC 1 /* Writes are synced at once */
1625 #define S_NOATIME 2 /* Do not update access times */
1626 #define S_APPEND 4 /* Append-only file */
1627 #define S_IMMUTABLE 8 /* Immutable file */
1628 #define S_DEAD 16 /* removed, but still open directory */
1629 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1630 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1631 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1632 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1633 #define S_PRIVATE 512 /* Inode is fs-internal */
1634 #define S_IMA 1024 /* Inode has an associated IMA struct */
1635 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1636 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1637
1638 /*
1639 * Note that nosuid etc flags are inode-specific: setting some file-system
1640 * flags just means all the inodes inherit those flags by default. It might be
1641 * possible to override it selectively if you really wanted to with some
1642 * ioctl() that is not currently implemented.
1643 *
1644 * Exception: MS_RDONLY is always applied to the entire file system.
1645 *
1646 * Unfortunately, it is possible to change a filesystems flags with it mounted
1647 * with files in use. This means that all of the inodes will not have their
1648 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1649 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1650 */
1651 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1652
1653 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1654 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1655 ((inode)->i_flags & S_SYNC))
1656 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1657 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1658 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1659 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1660 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1661
1662 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1663 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1664 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1665 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1666
1667 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1668 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1669 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1670 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1671 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1672 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1673 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1674
1675 /*
1676 * Inode state bits. Protected by inode->i_lock
1677 *
1678 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1679 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1680 *
1681 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1682 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1683 * various stages of removing an inode.
1684 *
1685 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1686 *
1687 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1688 * fdatasync(). i_atime is the usual cause.
1689 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1690 * these changes separately from I_DIRTY_SYNC so that we
1691 * don't have to write inode on fdatasync() when only
1692 * mtime has changed in it.
1693 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1694 * I_NEW Serves as both a mutex and completion notification.
1695 * New inodes set I_NEW. If two processes both create
1696 * the same inode, one of them will release its inode and
1697 * wait for I_NEW to be released before returning.
1698 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1699 * also cause waiting on I_NEW, without I_NEW actually
1700 * being set. find_inode() uses this to prevent returning
1701 * nearly-dead inodes.
1702 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1703 * is zero. I_FREEING must be set when I_WILL_FREE is
1704 * cleared.
1705 * I_FREEING Set when inode is about to be freed but still has dirty
1706 * pages or buffers attached or the inode itself is still
1707 * dirty.
1708 * I_CLEAR Added by clear_inode(). In this state the inode is
1709 * clean and can be destroyed. Inode keeps I_FREEING.
1710 *
1711 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1712 * prohibited for many purposes. iget() must wait for
1713 * the inode to be completely released, then create it
1714 * anew. Other functions will just ignore such inodes,
1715 * if appropriate. I_NEW is used for waiting.
1716 *
1717 * I_SYNC Writeback of inode is running. The bit is set during
1718 * data writeback, and cleared with a wakeup on the bit
1719 * address once it is done. The bit is also used to pin
1720 * the inode in memory for flusher thread.
1721 *
1722 * I_REFERENCED Marks the inode as recently references on the LRU list.
1723 *
1724 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1725 *
1726 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1727 */
1728 #define I_DIRTY_SYNC (1 << 0)
1729 #define I_DIRTY_DATASYNC (1 << 1)
1730 #define I_DIRTY_PAGES (1 << 2)
1731 #define __I_NEW 3
1732 #define I_NEW (1 << __I_NEW)
1733 #define I_WILL_FREE (1 << 4)
1734 #define I_FREEING (1 << 5)
1735 #define I_CLEAR (1 << 6)
1736 #define __I_SYNC 7
1737 #define I_SYNC (1 << __I_SYNC)
1738 #define I_REFERENCED (1 << 8)
1739 #define __I_DIO_WAKEUP 9
1740 #define I_DIO_WAKEUP (1 << I_DIO_WAKEUP)
1741
1742 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1743
1744 extern void __mark_inode_dirty(struct inode *, int);
1745 static inline void mark_inode_dirty(struct inode *inode)
1746 {
1747 __mark_inode_dirty(inode, I_DIRTY);
1748 }
1749
1750 static inline void mark_inode_dirty_sync(struct inode *inode)
1751 {
1752 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1753 }
1754
1755 extern void inc_nlink(struct inode *inode);
1756 extern void drop_nlink(struct inode *inode);
1757 extern void clear_nlink(struct inode *inode);
1758 extern void set_nlink(struct inode *inode, unsigned int nlink);
1759
1760 static inline void inode_inc_link_count(struct inode *inode)
1761 {
1762 inc_nlink(inode);
1763 mark_inode_dirty(inode);
1764 }
1765
1766 static inline void inode_dec_link_count(struct inode *inode)
1767 {
1768 drop_nlink(inode);
1769 mark_inode_dirty(inode);
1770 }
1771
1772 /**
1773 * inode_inc_iversion - increments i_version
1774 * @inode: inode that need to be updated
1775 *
1776 * Every time the inode is modified, the i_version field will be incremented.
1777 * The filesystem has to be mounted with i_version flag
1778 */
1779
1780 static inline void inode_inc_iversion(struct inode *inode)
1781 {
1782 spin_lock(&inode->i_lock);
1783 inode->i_version++;
1784 spin_unlock(&inode->i_lock);
1785 }
1786
1787 enum file_time_flags {
1788 S_ATIME = 1,
1789 S_MTIME = 2,
1790 S_CTIME = 4,
1791 S_VERSION = 8,
1792 };
1793
1794 extern void touch_atime(struct path *);
1795 static inline void file_accessed(struct file *file)
1796 {
1797 if (!(file->f_flags & O_NOATIME))
1798 touch_atime(&file->f_path);
1799 }
1800
1801 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1802 int sync_inode_metadata(struct inode *inode, int wait);
1803
1804 struct file_system_type {
1805 const char *name;
1806 int fs_flags;
1807 #define FS_REQUIRES_DEV 1
1808 #define FS_BINARY_MOUNTDATA 2
1809 #define FS_HAS_SUBTYPE 4
1810 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
1811 #define FS_USERNS_DEV_MOUNT 16 /* A userns mount does not imply MNT_NODEV */
1812 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
1813 struct dentry *(*mount) (struct file_system_type *, int,
1814 const char *, void *);
1815 void (*kill_sb) (struct super_block *);
1816 struct module *owner;
1817 struct file_system_type * next;
1818 struct hlist_head fs_supers;
1819
1820 struct lock_class_key s_lock_key;
1821 struct lock_class_key s_umount_key;
1822 struct lock_class_key s_vfs_rename_key;
1823 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
1824
1825 struct lock_class_key i_lock_key;
1826 struct lock_class_key i_mutex_key;
1827 struct lock_class_key i_mutex_dir_key;
1828 };
1829
1830 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
1831
1832 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1833 void *data, int (*fill_super)(struct super_block *, void *, int));
1834 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1835 int flags, const char *dev_name, void *data,
1836 int (*fill_super)(struct super_block *, void *, int));
1837 extern struct dentry *mount_single(struct file_system_type *fs_type,
1838 int flags, void *data,
1839 int (*fill_super)(struct super_block *, void *, int));
1840 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1841 int flags, void *data,
1842 int (*fill_super)(struct super_block *, void *, int));
1843 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
1844 void generic_shutdown_super(struct super_block *sb);
1845 void kill_block_super(struct super_block *sb);
1846 void kill_anon_super(struct super_block *sb);
1847 void kill_litter_super(struct super_block *sb);
1848 void deactivate_super(struct super_block *sb);
1849 void deactivate_locked_super(struct super_block *sb);
1850 int set_anon_super(struct super_block *s, void *data);
1851 int get_anon_bdev(dev_t *);
1852 void free_anon_bdev(dev_t);
1853 struct super_block *sget(struct file_system_type *type,
1854 int (*test)(struct super_block *,void *),
1855 int (*set)(struct super_block *,void *),
1856 int flags, void *data);
1857 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
1858 const struct super_operations *ops,
1859 const struct dentry_operations *dops,
1860 unsigned long);
1861
1862 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1863 #define fops_get(fops) \
1864 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1865 #define fops_put(fops) \
1866 do { if (fops) module_put((fops)->owner); } while(0)
1867
1868 extern int register_filesystem(struct file_system_type *);
1869 extern int unregister_filesystem(struct file_system_type *);
1870 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1871 #define kern_mount(type) kern_mount_data(type, NULL)
1872 extern void kern_unmount(struct vfsmount *mnt);
1873 extern int may_umount_tree(struct vfsmount *);
1874 extern int may_umount(struct vfsmount *);
1875 extern long do_mount(const char *, const char *, const char *, unsigned long, void *);
1876 extern struct vfsmount *collect_mounts(struct path *);
1877 extern void drop_collected_mounts(struct vfsmount *);
1878 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1879 struct vfsmount *);
1880 extern int vfs_statfs(struct path *, struct kstatfs *);
1881 extern int user_statfs(const char __user *, struct kstatfs *);
1882 extern int fd_statfs(int, struct kstatfs *);
1883 extern int vfs_ustat(dev_t, struct kstatfs *);
1884 extern int freeze_super(struct super_block *super);
1885 extern int thaw_super(struct super_block *super);
1886 extern bool our_mnt(struct vfsmount *mnt);
1887
1888 extern int current_umask(void);
1889
1890 /* /sys/fs */
1891 extern struct kobject *fs_kobj;
1892
1893 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
1894 extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1895
1896 #define FLOCK_VERIFY_READ 1
1897 #define FLOCK_VERIFY_WRITE 2
1898
1899 #ifdef CONFIG_FILE_LOCKING
1900 extern int locks_mandatory_locked(struct inode *);
1901 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1902
1903 /*
1904 * Candidates for mandatory locking have the setgid bit set
1905 * but no group execute bit - an otherwise meaningless combination.
1906 */
1907
1908 static inline int __mandatory_lock(struct inode *ino)
1909 {
1910 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1911 }
1912
1913 /*
1914 * ... and these candidates should be on MS_MANDLOCK mounted fs,
1915 * otherwise these will be advisory locks
1916 */
1917
1918 static inline int mandatory_lock(struct inode *ino)
1919 {
1920 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1921 }
1922
1923 static inline int locks_verify_locked(struct inode *inode)
1924 {
1925 if (mandatory_lock(inode))
1926 return locks_mandatory_locked(inode);
1927 return 0;
1928 }
1929
1930 static inline int locks_verify_truncate(struct inode *inode,
1931 struct file *filp,
1932 loff_t size)
1933 {
1934 if (inode->i_flock && mandatory_lock(inode))
1935 return locks_mandatory_area(
1936 FLOCK_VERIFY_WRITE, inode, filp,
1937 size < inode->i_size ? size : inode->i_size,
1938 (size < inode->i_size ? inode->i_size - size
1939 : size - inode->i_size)
1940 );
1941 return 0;
1942 }
1943
1944 static inline int break_lease(struct inode *inode, unsigned int mode)
1945 {
1946 if (inode->i_flock)
1947 return __break_lease(inode, mode);
1948 return 0;
1949 }
1950 #else /* !CONFIG_FILE_LOCKING */
1951 static inline int locks_mandatory_locked(struct inode *inode)
1952 {
1953 return 0;
1954 }
1955
1956 static inline int locks_mandatory_area(int rw, struct inode *inode,
1957 struct file *filp, loff_t offset,
1958 size_t count)
1959 {
1960 return 0;
1961 }
1962
1963 static inline int __mandatory_lock(struct inode *inode)
1964 {
1965 return 0;
1966 }
1967
1968 static inline int mandatory_lock(struct inode *inode)
1969 {
1970 return 0;
1971 }
1972
1973 static inline int locks_verify_locked(struct inode *inode)
1974 {
1975 return 0;
1976 }
1977
1978 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
1979 size_t size)
1980 {
1981 return 0;
1982 }
1983
1984 static inline int break_lease(struct inode *inode, unsigned int mode)
1985 {
1986 return 0;
1987 }
1988
1989 #endif /* CONFIG_FILE_LOCKING */
1990
1991 /* fs/open.c */
1992 struct audit_names;
1993 struct filename {
1994 const char *name; /* pointer to actual string */
1995 const __user char *uptr; /* original userland pointer */
1996 struct audit_names *aname;
1997 bool separate; /* should "name" be freed? */
1998 };
1999
2000 extern long vfs_truncate(struct path *, loff_t);
2001 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2002 struct file *filp);
2003 extern int do_fallocate(struct file *file, int mode, loff_t offset,
2004 loff_t len);
2005 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2006 umode_t mode);
2007 extern struct file *file_open_name(struct filename *, int, umode_t);
2008 extern struct file *filp_open(const char *, int, umode_t);
2009 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2010 const char *, int);
2011 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2012 extern int filp_close(struct file *, fl_owner_t id);
2013
2014 extern struct filename *getname(const char __user *);
2015
2016 enum {
2017 FILE_CREATED = 1,
2018 FILE_OPENED = 2
2019 };
2020 extern int finish_open(struct file *file, struct dentry *dentry,
2021 int (*open)(struct inode *, struct file *),
2022 int *opened);
2023 extern int finish_no_open(struct file *file, struct dentry *dentry);
2024
2025 /* fs/ioctl.c */
2026
2027 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2028
2029 /* fs/dcache.c */
2030 extern void __init vfs_caches_init_early(void);
2031 extern void __init vfs_caches_init(unsigned long);
2032
2033 extern struct kmem_cache *names_cachep;
2034
2035 extern void final_putname(struct filename *name);
2036
2037 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2038 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2039 #ifndef CONFIG_AUDITSYSCALL
2040 #define putname(name) final_putname(name)
2041 #else
2042 extern void putname(struct filename *name);
2043 #endif
2044
2045 #ifdef CONFIG_BLOCK
2046 extern int register_blkdev(unsigned int, const char *);
2047 extern void unregister_blkdev(unsigned int, const char *);
2048 extern struct block_device *bdget(dev_t);
2049 extern struct block_device *bdgrab(struct block_device *bdev);
2050 extern void bd_set_size(struct block_device *, loff_t size);
2051 extern void bd_forget(struct inode *inode);
2052 extern void bdput(struct block_device *);
2053 extern void invalidate_bdev(struct block_device *);
2054 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2055 extern int sync_blockdev(struct block_device *bdev);
2056 extern void kill_bdev(struct block_device *);
2057 extern struct super_block *freeze_bdev(struct block_device *);
2058 extern void emergency_thaw_all(void);
2059 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2060 extern int fsync_bdev(struct block_device *);
2061 #else
2062 static inline void bd_forget(struct inode *inode) {}
2063 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2064 static inline void kill_bdev(struct block_device *bdev) {}
2065 static inline void invalidate_bdev(struct block_device *bdev) {}
2066
2067 static inline struct super_block *freeze_bdev(struct block_device *sb)
2068 {
2069 return NULL;
2070 }
2071
2072 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2073 {
2074 return 0;
2075 }
2076
2077 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2078 {
2079 }
2080 #endif
2081 extern int sync_filesystem(struct super_block *);
2082 extern const struct file_operations def_blk_fops;
2083 extern const struct file_operations def_chr_fops;
2084 extern const struct file_operations bad_sock_fops;
2085 #ifdef CONFIG_BLOCK
2086 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2087 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2088 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2089 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2090 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2091 void *holder);
2092 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2093 void *holder);
2094 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2095 #ifdef CONFIG_SYSFS
2096 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2097 extern void bd_unlink_disk_holder(struct block_device *bdev,
2098 struct gendisk *disk);
2099 #else
2100 static inline int bd_link_disk_holder(struct block_device *bdev,
2101 struct gendisk *disk)
2102 {
2103 return 0;
2104 }
2105 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2106 struct gendisk *disk)
2107 {
2108 }
2109 #endif
2110 #endif
2111
2112 /* fs/char_dev.c */
2113 #define CHRDEV_MAJOR_HASH_SIZE 255
2114 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2115 extern int register_chrdev_region(dev_t, unsigned, const char *);
2116 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2117 unsigned int count, const char *name,
2118 const struct file_operations *fops);
2119 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2120 unsigned int count, const char *name);
2121 extern void unregister_chrdev_region(dev_t, unsigned);
2122 extern void chrdev_show(struct seq_file *,off_t);
2123
2124 static inline int register_chrdev(unsigned int major, const char *name,
2125 const struct file_operations *fops)
2126 {
2127 return __register_chrdev(major, 0, 256, name, fops);
2128 }
2129
2130 static inline void unregister_chrdev(unsigned int major, const char *name)
2131 {
2132 __unregister_chrdev(major, 0, 256, name);
2133 }
2134
2135 /* fs/block_dev.c */
2136 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2137 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2138
2139 #ifdef CONFIG_BLOCK
2140 #define BLKDEV_MAJOR_HASH_SIZE 255
2141 extern const char *__bdevname(dev_t, char *buffer);
2142 extern const char *bdevname(struct block_device *bdev, char *buffer);
2143 extern struct block_device *lookup_bdev(const char *);
2144 extern void blkdev_show(struct seq_file *,off_t);
2145
2146 #else
2147 #define BLKDEV_MAJOR_HASH_SIZE 0
2148 #endif
2149
2150 extern void init_special_inode(struct inode *, umode_t, dev_t);
2151
2152 /* Invalid inode operations -- fs/bad_inode.c */
2153 extern void make_bad_inode(struct inode *);
2154 extern int is_bad_inode(struct inode *);
2155
2156 #ifdef CONFIG_BLOCK
2157 /*
2158 * return READ, READA, or WRITE
2159 */
2160 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2161
2162 /*
2163 * return data direction, READ or WRITE
2164 */
2165 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2166
2167 extern void check_disk_size_change(struct gendisk *disk,
2168 struct block_device *bdev);
2169 extern int revalidate_disk(struct gendisk *);
2170 extern int check_disk_change(struct block_device *);
2171 extern int __invalidate_device(struct block_device *, bool);
2172 extern int invalidate_partition(struct gendisk *, int);
2173 #endif
2174 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2175 pgoff_t start, pgoff_t end);
2176
2177 static inline void invalidate_remote_inode(struct inode *inode)
2178 {
2179 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2180 S_ISLNK(inode->i_mode))
2181 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2182 }
2183 extern int invalidate_inode_pages2(struct address_space *mapping);
2184 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2185 pgoff_t start, pgoff_t end);
2186 extern int write_inode_now(struct inode *, int);
2187 extern int filemap_fdatawrite(struct address_space *);
2188 extern int filemap_flush(struct address_space *);
2189 extern int filemap_fdatawait(struct address_space *);
2190 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2191 loff_t lend);
2192 extern int filemap_write_and_wait(struct address_space *mapping);
2193 extern int filemap_write_and_wait_range(struct address_space *mapping,
2194 loff_t lstart, loff_t lend);
2195 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2196 loff_t start, loff_t end, int sync_mode);
2197 extern int filemap_fdatawrite_range(struct address_space *mapping,
2198 loff_t start, loff_t end);
2199
2200 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2201 int datasync);
2202 extern int vfs_fsync(struct file *file, int datasync);
2203 extern int generic_write_sync(struct file *file, loff_t pos, loff_t count);
2204 extern void emergency_sync(void);
2205 extern void emergency_remount(void);
2206 #ifdef CONFIG_BLOCK
2207 extern sector_t bmap(struct inode *, sector_t);
2208 #endif
2209 extern int notify_change(struct dentry *, struct iattr *);
2210 extern int inode_permission(struct inode *, int);
2211 extern int generic_permission(struct inode *, int);
2212
2213 static inline bool execute_ok(struct inode *inode)
2214 {
2215 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2216 }
2217
2218 static inline struct inode *file_inode(struct file *f)
2219 {
2220 return f->f_inode;
2221 }
2222
2223 static inline void file_start_write(struct file *file)
2224 {
2225 if (!S_ISREG(file_inode(file)->i_mode))
2226 return;
2227 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2228 }
2229
2230 static inline bool file_start_write_trylock(struct file *file)
2231 {
2232 if (!S_ISREG(file_inode(file)->i_mode))
2233 return true;
2234 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2235 }
2236
2237 static inline void file_end_write(struct file *file)
2238 {
2239 if (!S_ISREG(file_inode(file)->i_mode))
2240 return;
2241 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2242 }
2243
2244 /*
2245 * get_write_access() gets write permission for a file.
2246 * put_write_access() releases this write permission.
2247 * This is used for regular files.
2248 * We cannot support write (and maybe mmap read-write shared) accesses and
2249 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2250 * can have the following values:
2251 * 0: no writers, no VM_DENYWRITE mappings
2252 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2253 * > 0: (i_writecount) users are writing to the file.
2254 *
2255 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2256 * except for the cases where we don't hold i_writecount yet. Then we need to
2257 * use {get,deny}_write_access() - these functions check the sign and refuse
2258 * to do the change if sign is wrong.
2259 */
2260 static inline int get_write_access(struct inode *inode)
2261 {
2262 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2263 }
2264 static inline int deny_write_access(struct file *file)
2265 {
2266 struct inode *inode = file_inode(file);
2267 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2268 }
2269 static inline void put_write_access(struct inode * inode)
2270 {
2271 atomic_dec(&inode->i_writecount);
2272 }
2273 static inline void allow_write_access(struct file *file)
2274 {
2275 if (file)
2276 atomic_inc(&file_inode(file)->i_writecount);
2277 }
2278 #ifdef CONFIG_IMA
2279 static inline void i_readcount_dec(struct inode *inode)
2280 {
2281 BUG_ON(!atomic_read(&inode->i_readcount));
2282 atomic_dec(&inode->i_readcount);
2283 }
2284 static inline void i_readcount_inc(struct inode *inode)
2285 {
2286 atomic_inc(&inode->i_readcount);
2287 }
2288 #else
2289 static inline void i_readcount_dec(struct inode *inode)
2290 {
2291 return;
2292 }
2293 static inline void i_readcount_inc(struct inode *inode)
2294 {
2295 return;
2296 }
2297 #endif
2298 extern int do_pipe_flags(int *, int);
2299
2300 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2301 extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
2302 extern struct file * open_exec(const char *);
2303
2304 /* fs/dcache.c -- generic fs support functions */
2305 extern int is_subdir(struct dentry *, struct dentry *);
2306 extern int path_is_under(struct path *, struct path *);
2307 extern ino_t find_inode_number(struct dentry *, struct qstr *);
2308
2309 #include <linux/err.h>
2310
2311 /* needed for stackable file system support */
2312 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2313
2314 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2315
2316 extern int inode_init_always(struct super_block *, struct inode *);
2317 extern void inode_init_once(struct inode *);
2318 extern void address_space_init_once(struct address_space *mapping);
2319 extern void ihold(struct inode * inode);
2320 extern void iput(struct inode *);
2321 extern struct inode * igrab(struct inode *);
2322 extern ino_t iunique(struct super_block *, ino_t);
2323 extern int inode_needs_sync(struct inode *inode);
2324 extern int generic_delete_inode(struct inode *inode);
2325 static inline int generic_drop_inode(struct inode *inode)
2326 {
2327 return !inode->i_nlink || inode_unhashed(inode);
2328 }
2329
2330 extern struct inode *ilookup5_nowait(struct super_block *sb,
2331 unsigned long hashval, int (*test)(struct inode *, void *),
2332 void *data);
2333 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2334 int (*test)(struct inode *, void *), void *data);
2335 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2336
2337 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2338 extern struct inode * iget_locked(struct super_block *, unsigned long);
2339 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2340 extern int insert_inode_locked(struct inode *);
2341 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2342 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2343 #else
2344 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2345 #endif
2346 extern void unlock_new_inode(struct inode *);
2347 extern unsigned int get_next_ino(void);
2348
2349 extern void __iget(struct inode * inode);
2350 extern void iget_failed(struct inode *);
2351 extern void clear_inode(struct inode *);
2352 extern void __destroy_inode(struct inode *);
2353 extern struct inode *new_inode_pseudo(struct super_block *sb);
2354 extern struct inode *new_inode(struct super_block *sb);
2355 extern void free_inode_nonrcu(struct inode *inode);
2356 extern int should_remove_suid(struct dentry *);
2357 extern int file_remove_suid(struct file *);
2358
2359 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2360 static inline void insert_inode_hash(struct inode *inode)
2361 {
2362 __insert_inode_hash(inode, inode->i_ino);
2363 }
2364
2365 extern void __remove_inode_hash(struct inode *);
2366 static inline void remove_inode_hash(struct inode *inode)
2367 {
2368 if (!inode_unhashed(inode))
2369 __remove_inode_hash(inode);
2370 }
2371
2372 extern void inode_sb_list_add(struct inode *inode);
2373
2374 #ifdef CONFIG_BLOCK
2375 extern void submit_bio(int, struct bio *);
2376 extern int bdev_read_only(struct block_device *);
2377 #endif
2378 extern int set_blocksize(struct block_device *, int);
2379 extern int sb_set_blocksize(struct super_block *, int);
2380 extern int sb_min_blocksize(struct super_block *, int);
2381
2382 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2383 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2384 extern int generic_file_remap_pages(struct vm_area_struct *, unsigned long addr,
2385 unsigned long size, pgoff_t pgoff);
2386 extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
2387 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
2388 extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2389 extern ssize_t __generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long,
2390 loff_t *);
2391 extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2392 extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
2393 unsigned long *, loff_t, loff_t *, size_t, size_t);
2394 extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
2395 unsigned long, loff_t, loff_t *, size_t, ssize_t);
2396 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2397 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2398 extern int generic_segment_checks(const struct iovec *iov,
2399 unsigned long *nr_segs, size_t *count, int access_flags);
2400
2401 /* fs/block_dev.c */
2402 extern ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov,
2403 unsigned long nr_segs, loff_t pos);
2404 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2405 int datasync);
2406 extern void block_sync_page(struct page *page);
2407
2408 /* fs/splice.c */
2409 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2410 struct pipe_inode_info *, size_t, unsigned int);
2411 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2412 struct pipe_inode_info *, size_t, unsigned int);
2413 extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
2414 struct file *, loff_t *, size_t, unsigned int);
2415 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2416 struct file *out, loff_t *, size_t len, unsigned int flags);
2417
2418 extern void
2419 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2420 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2421 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2422 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2423 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2424 int whence, loff_t maxsize, loff_t eof);
2425 extern int generic_file_open(struct inode * inode, struct file * filp);
2426 extern int nonseekable_open(struct inode * inode, struct file * filp);
2427
2428 #ifdef CONFIG_FS_XIP
2429 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2430 loff_t *ppos);
2431 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
2432 extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2433 size_t len, loff_t *ppos);
2434 extern int xip_truncate_page(struct address_space *mapping, loff_t from);
2435 #else
2436 static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2437 {
2438 return 0;
2439 }
2440 #endif
2441
2442 #ifdef CONFIG_BLOCK
2443 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2444 loff_t file_offset);
2445
2446 enum {
2447 /* need locking between buffered and direct access */
2448 DIO_LOCKING = 0x01,
2449
2450 /* filesystem does not support filling holes */
2451 DIO_SKIP_HOLES = 0x02,
2452 };
2453
2454 void dio_end_io(struct bio *bio, int error);
2455
2456 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2457 struct block_device *bdev, const struct iovec *iov, loff_t offset,
2458 unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
2459 dio_submit_t submit_io, int flags);
2460
2461 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
2462 struct inode *inode, const struct iovec *iov, loff_t offset,
2463 unsigned long nr_segs, get_block_t get_block)
2464 {
2465 return __blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov,
2466 offset, nr_segs, get_block, NULL, NULL,
2467 DIO_LOCKING | DIO_SKIP_HOLES);
2468 }
2469 #endif
2470
2471 void inode_dio_wait(struct inode *inode);
2472 void inode_dio_done(struct inode *inode);
2473
2474 extern const struct file_operations generic_ro_fops;
2475
2476 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2477
2478 extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
2479 extern int vfs_follow_link(struct nameidata *, const char *);
2480 extern int page_readlink(struct dentry *, char __user *, int);
2481 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2482 extern void page_put_link(struct dentry *, struct nameidata *, void *);
2483 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2484 int nofs);
2485 extern int page_symlink(struct inode *inode, const char *symname, int len);
2486 extern const struct inode_operations page_symlink_inode_operations;
2487 extern int generic_readlink(struct dentry *, char __user *, int);
2488 extern void generic_fillattr(struct inode *, struct kstat *);
2489 extern int vfs_getattr(struct path *, struct kstat *);
2490 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2491 void inode_add_bytes(struct inode *inode, loff_t bytes);
2492 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
2493 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2494 loff_t inode_get_bytes(struct inode *inode);
2495 void inode_set_bytes(struct inode *inode, loff_t bytes);
2496
2497 extern int vfs_readdir(struct file *, filldir_t, void *);
2498
2499 extern int vfs_stat(const char __user *, struct kstat *);
2500 extern int vfs_lstat(const char __user *, struct kstat *);
2501 extern int vfs_fstat(unsigned int, struct kstat *);
2502 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2503
2504 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2505 unsigned long arg);
2506 extern int __generic_block_fiemap(struct inode *inode,
2507 struct fiemap_extent_info *fieinfo,
2508 loff_t start, loff_t len,
2509 get_block_t *get_block);
2510 extern int generic_block_fiemap(struct inode *inode,
2511 struct fiemap_extent_info *fieinfo, u64 start,
2512 u64 len, get_block_t *get_block);
2513
2514 extern void get_filesystem(struct file_system_type *fs);
2515 extern void put_filesystem(struct file_system_type *fs);
2516 extern struct file_system_type *get_fs_type(const char *name);
2517 extern struct super_block *get_super(struct block_device *);
2518 extern struct super_block *get_super_thawed(struct block_device *);
2519 extern struct super_block *get_active_super(struct block_device *bdev);
2520 extern void drop_super(struct super_block *sb);
2521 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2522 extern void iterate_supers_type(struct file_system_type *,
2523 void (*)(struct super_block *, void *), void *);
2524
2525 extern int dcache_dir_open(struct inode *, struct file *);
2526 extern int dcache_dir_close(struct inode *, struct file *);
2527 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2528 extern int dcache_readdir(struct file *, void *, filldir_t);
2529 extern int simple_setattr(struct dentry *, struct iattr *);
2530 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2531 extern int simple_statfs(struct dentry *, struct kstatfs *);
2532 extern int simple_open(struct inode *inode, struct file *file);
2533 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2534 extern int simple_unlink(struct inode *, struct dentry *);
2535 extern int simple_rmdir(struct inode *, struct dentry *);
2536 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2537 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2538 extern int simple_empty(struct dentry *);
2539 extern int simple_readpage(struct file *file, struct page *page);
2540 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2541 loff_t pos, unsigned len, unsigned flags,
2542 struct page **pagep, void **fsdata);
2543 extern int simple_write_end(struct file *file, struct address_space *mapping,
2544 loff_t pos, unsigned len, unsigned copied,
2545 struct page *page, void *fsdata);
2546
2547 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
2548 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2549 extern const struct file_operations simple_dir_operations;
2550 extern const struct inode_operations simple_dir_inode_operations;
2551 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2552 struct dentry *d_alloc_name(struct dentry *, const char *);
2553 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2554 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2555 extern void simple_release_fs(struct vfsmount **mount, int *count);
2556
2557 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2558 loff_t *ppos, const void *from, size_t available);
2559 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2560 const void __user *from, size_t count);
2561
2562 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2563
2564 extern int generic_check_addressable(unsigned, u64);
2565
2566 #ifdef CONFIG_MIGRATION
2567 extern int buffer_migrate_page(struct address_space *,
2568 struct page *, struct page *,
2569 enum migrate_mode);
2570 #else
2571 #define buffer_migrate_page NULL
2572 #endif
2573
2574 extern int inode_change_ok(const struct inode *, struct iattr *);
2575 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2576 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2577
2578 extern int file_update_time(struct file *file);
2579
2580 extern int generic_show_options(struct seq_file *m, struct dentry *root);
2581 extern void save_mount_options(struct super_block *sb, char *options);
2582 extern void replace_mount_options(struct super_block *sb, char *options);
2583
2584 static inline ino_t parent_ino(struct dentry *dentry)
2585 {
2586 ino_t res;
2587
2588 /*
2589 * Don't strictly need d_lock here? If the parent ino could change
2590 * then surely we'd have a deeper race in the caller?
2591 */
2592 spin_lock(&dentry->d_lock);
2593 res = dentry->d_parent->d_inode->i_ino;
2594 spin_unlock(&dentry->d_lock);
2595 return res;
2596 }
2597
2598 /* Transaction based IO helpers */
2599
2600 /*
2601 * An argresp is stored in an allocated page and holds the
2602 * size of the argument or response, along with its content
2603 */
2604 struct simple_transaction_argresp {
2605 ssize_t size;
2606 char data[0];
2607 };
2608
2609 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2610
2611 char *simple_transaction_get(struct file *file, const char __user *buf,
2612 size_t size);
2613 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2614 size_t size, loff_t *pos);
2615 int simple_transaction_release(struct inode *inode, struct file *file);
2616
2617 void simple_transaction_set(struct file *file, size_t n);
2618
2619 /*
2620 * simple attribute files
2621 *
2622 * These attributes behave similar to those in sysfs:
2623 *
2624 * Writing to an attribute immediately sets a value, an open file can be
2625 * written to multiple times.
2626 *
2627 * Reading from an attribute creates a buffer from the value that might get
2628 * read with multiple read calls. When the attribute has been read
2629 * completely, no further read calls are possible until the file is opened
2630 * again.
2631 *
2632 * All attributes contain a text representation of a numeric value
2633 * that are accessed with the get() and set() functions.
2634 */
2635 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2636 static int __fops ## _open(struct inode *inode, struct file *file) \
2637 { \
2638 __simple_attr_check_format(__fmt, 0ull); \
2639 return simple_attr_open(inode, file, __get, __set, __fmt); \
2640 } \
2641 static const struct file_operations __fops = { \
2642 .owner = THIS_MODULE, \
2643 .open = __fops ## _open, \
2644 .release = simple_attr_release, \
2645 .read = simple_attr_read, \
2646 .write = simple_attr_write, \
2647 .llseek = generic_file_llseek, \
2648 };
2649
2650 static inline __printf(1, 2)
2651 void __simple_attr_check_format(const char *fmt, ...)
2652 {
2653 /* don't do anything, just let the compiler check the arguments; */
2654 }
2655
2656 int simple_attr_open(struct inode *inode, struct file *file,
2657 int (*get)(void *, u64 *), int (*set)(void *, u64),
2658 const char *fmt);
2659 int simple_attr_release(struct inode *inode, struct file *file);
2660 ssize_t simple_attr_read(struct file *file, char __user *buf,
2661 size_t len, loff_t *ppos);
2662 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2663 size_t len, loff_t *ppos);
2664
2665 struct ctl_table;
2666 int proc_nr_files(struct ctl_table *table, int write,
2667 void __user *buffer, size_t *lenp, loff_t *ppos);
2668 int proc_nr_dentry(struct ctl_table *table, int write,
2669 void __user *buffer, size_t *lenp, loff_t *ppos);
2670 int proc_nr_inodes(struct ctl_table *table, int write,
2671 void __user *buffer, size_t *lenp, loff_t *ppos);
2672 int __init get_filesystem_list(char *buf);
2673
2674 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
2675 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
2676
2677 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2678 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2679 (flag & __FMODE_NONOTIFY)))
2680
2681 static inline int is_sxid(umode_t mode)
2682 {
2683 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
2684 }
2685
2686 static inline void inode_has_no_xattr(struct inode *inode)
2687 {
2688 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
2689 inode->i_flags |= S_NOSEC;
2690 }
2691
2692 #endif /* _LINUX_FS_H */