xfs: Set allowed quota types
[deliverable/linux.git] / include / linux / fs.h
CommitLineData
1da177e4
LT
1#ifndef _LINUX_FS_H
2#define _LINUX_FS_H
3
1da177e4
LT
4
5#include <linux/linkage.h>
6#include <linux/wait.h>
1da177e4
LT
7#include <linux/kdev_t.h>
8#include <linux/dcache.h>
3f8206d4 9#include <linux/path.h>
1da177e4
LT
10#include <linux/stat.h>
11#include <linux/cache.h>
1da177e4 12#include <linux/list.h>
bc3b14cb 13#include <linux/list_lru.h>
4f5e65a1 14#include <linux/llist.h>
1da177e4 15#include <linux/radix-tree.h>
6b2dbba8 16#include <linux/rbtree.h>
1da177e4 17#include <linux/init.h>
914e2637 18#include <linux/pid.h>
187f1882 19#include <linux/bug.h>
1b1dcc1b 20#include <linux/mutex.h>
3bd858ab 21#include <linux/capability.h>
6188e10d 22#include <linux/semaphore.h>
c4b929b8 23#include <linux/fiemap.h>
ceb5bdc2 24#include <linux/rculist_bl.h>
07b8ce1e 25#include <linux/atomic.h>
83aeeada 26#include <linux/shrinker.h>
c1aab02d 27#include <linux/migrate_mode.h>
92361636 28#include <linux/uidgid.h>
5accdf82 29#include <linux/lockdep.h>
c2b1ad80 30#include <linux/percpu-rwsem.h>
08cce05c 31#include <linux/blk_types.h>
1da177e4 32
1da177e4 33#include <asm/byteorder.h>
607ca46e 34#include <uapi/linux/fs.h>
1da177e4 35
a5694255 36struct export_operations;
a885c8c4 37struct hd_geometry;
1da177e4
LT
38struct iovec;
39struct nameidata;
92198f7e 40struct kiocb;
57cc7215 41struct kobject;
1da177e4
LT
42struct pipe_inode_info;
43struct poll_table_struct;
44struct kstatfs;
45struct vm_area_struct;
46struct vfsmount;
745ca247 47struct cred;
a509bc1a 48struct swap_info_struct;
55985dd7 49struct seq_file;
7b7a8665 50struct workqueue_struct;
3b93f911 51struct iov_iter;
1da177e4 52
74bf17cf 53extern void __init inode_init(void);
1da177e4 54extern void __init inode_init_early(void);
1da177e4
LT
55extern void __init files_init(unsigned long);
56
dded4f4d 57extern struct files_stat_struct files_stat;
518de9b3 58extern unsigned long get_max_files(void);
dded4f4d
JE
59extern int sysctl_nr_open;
60extern struct inodes_stat_t inodes_stat;
61extern int leases_enable, lease_break_time;
800179c9
KC
62extern int sysctl_protected_symlinks;
63extern int sysctl_protected_hardlinks;
dded4f4d 64
1da177e4
LT
65struct buffer_head;
66typedef int (get_block_t)(struct inode *inode, sector_t iblock,
67 struct buffer_head *bh_result, int create);
92198f7e 68typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
7b7a8665 69 ssize_t bytes, void *private);
1da177e4 70
bbc1096a
DH
71#define MAY_EXEC 0x00000001
72#define MAY_WRITE 0x00000002
73#define MAY_READ 0x00000004
74#define MAY_APPEND 0x00000008
75#define MAY_ACCESS 0x00000010
76#define MAY_OPEN 0x00000020
77#define MAY_CHDIR 0x00000040
78/* called from RCU mode, don't block */
79#define MAY_NOT_BLOCK 0x00000080
80
81/*
82 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
83 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
84 */
85
86/* file is open for reading */
87#define FMODE_READ ((__force fmode_t)0x1)
88/* file is open for writing */
89#define FMODE_WRITE ((__force fmode_t)0x2)
90/* file is seekable */
91#define FMODE_LSEEK ((__force fmode_t)0x4)
92/* file can be accessed using pread */
93#define FMODE_PREAD ((__force fmode_t)0x8)
94/* file can be accessed using pwrite */
95#define FMODE_PWRITE ((__force fmode_t)0x10)
96/* File is opened for execution with sys_execve / sys_uselib */
97#define FMODE_EXEC ((__force fmode_t)0x20)
98/* File is opened with O_NDELAY (only set for block devices) */
99#define FMODE_NDELAY ((__force fmode_t)0x40)
100/* File is opened with O_EXCL (only set for block devices) */
101#define FMODE_EXCL ((__force fmode_t)0x80)
102/* File is opened using open(.., 3, ..) and is writeable only for ioctls
103 (specialy hack for floppy.c) */
104#define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
105/* 32bit hashes as llseek() offset (for directories) */
106#define FMODE_32BITHASH ((__force fmode_t)0x200)
107/* 64bit hashes as llseek() offset (for directories) */
108#define FMODE_64BITHASH ((__force fmode_t)0x400)
109
110/*
111 * Don't update ctime and mtime.
112 *
113 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
114 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
115 */
116#define FMODE_NOCMTIME ((__force fmode_t)0x800)
117
118/* Expect random access pattern */
119#define FMODE_RANDOM ((__force fmode_t)0x1000)
120
121/* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
122#define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
123
124/* File is opened with O_PATH; almost nothing can be done with it */
125#define FMODE_PATH ((__force fmode_t)0x4000)
126
9c225f26
LT
127/* File needs atomic accesses to f_pos */
128#define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
83f936c7
AV
129/* Write access to underlying fs */
130#define FMODE_WRITER ((__force fmode_t)0x10000)
7f7f25e8
AV
131/* Has read method(s) */
132#define FMODE_CAN_READ ((__force fmode_t)0x20000)
133/* Has write method(s) */
134#define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
9c225f26 135
bbc1096a
DH
136/* File was opened by fanotify and shouldn't generate fanotify events */
137#define FMODE_NONOTIFY ((__force fmode_t)0x1000000)
138
139/*
140 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
141 * that indicates that they should check the contents of the iovec are
142 * valid, but not check the memory that the iovec elements
143 * points too.
144 */
145#define CHECK_IOVEC_ONLY -1
146
08cce05c
DH
147/*
148 * The below are the various read and write types that we support. Some of
149 * them include behavioral modifiers that send information down to the
150 * block layer and IO scheduler. Terminology:
151 *
152 * The block layer uses device plugging to defer IO a little bit, in
153 * the hope that we will see more IO very shortly. This increases
154 * coalescing of adjacent IO and thus reduces the number of IOs we
155 * have to send to the device. It also allows for better queuing,
156 * if the IO isn't mergeable. If the caller is going to be waiting
157 * for the IO, then he must ensure that the device is unplugged so
158 * that the IO is dispatched to the driver.
159 *
160 * All IO is handled async in Linux. This is fine for background
161 * writes, but for reads or writes that someone waits for completion
162 * on, we want to notify the block layer and IO scheduler so that they
163 * know about it. That allows them to make better scheduling
164 * decisions. So when the below references 'sync' and 'async', it
165 * is referencing this priority hint.
166 *
167 * With that in mind, the available types are:
168 *
169 * READ A normal read operation. Device will be plugged.
170 * READ_SYNC A synchronous read. Device is not plugged, caller can
171 * immediately wait on this read without caring about
172 * unplugging.
173 * READA Used for read-ahead operations. Lower priority, and the
174 * block layer could (in theory) choose to ignore this
175 * request if it runs into resource problems.
176 * WRITE A normal async write. Device will be plugged.
177 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
178 * the hint that someone will be waiting on this IO
179 * shortly. The write equivalent of READ_SYNC.
180 * WRITE_ODIRECT Special case write for O_DIRECT only.
181 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
182 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
183 * non-volatile media on completion.
184 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
185 * by a cache flush and data is guaranteed to be on
186 * non-volatile media on completion.
187 *
188 */
189#define RW_MASK REQ_WRITE
190#define RWA_MASK REQ_RAHEAD
191
192#define READ 0
193#define WRITE RW_MASK
194#define READA RWA_MASK
08cce05c
DH
195
196#define READ_SYNC (READ | REQ_SYNC)
197#define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
198#define WRITE_ODIRECT (WRITE | REQ_SYNC)
199#define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
200#define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
201#define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
202
1da177e4
LT
203/*
204 * Attribute flags. These should be or-ed together to figure out what
205 * has been changed!
206 */
9767d749
MS
207#define ATTR_MODE (1 << 0)
208#define ATTR_UID (1 << 1)
209#define ATTR_GID (1 << 2)
210#define ATTR_SIZE (1 << 3)
211#define ATTR_ATIME (1 << 4)
212#define ATTR_MTIME (1 << 5)
213#define ATTR_CTIME (1 << 6)
214#define ATTR_ATIME_SET (1 << 7)
215#define ATTR_MTIME_SET (1 << 8)
216#define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
217#define ATTR_ATTR_FLAG (1 << 10)
218#define ATTR_KILL_SUID (1 << 11)
219#define ATTR_KILL_SGID (1 << 12)
220#define ATTR_FILE (1 << 13)
221#define ATTR_KILL_PRIV (1 << 14)
222#define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
223#define ATTR_TIMES_SET (1 << 16)
1da177e4 224
787fb6bc
MS
225/*
226 * Whiteout is represented by a char device. The following constants define the
227 * mode and device number to use.
228 */
229#define WHITEOUT_MODE 0
230#define WHITEOUT_DEV 0
231
1da177e4
LT
232/*
233 * This is the Inode Attributes structure, used for notify_change(). It
234 * uses the above definitions as flags, to know which values have changed.
235 * Also, in this manner, a Filesystem can look at only the values it cares
236 * about. Basically, these are the attributes that the VFS layer can
237 * request to change from the FS layer.
238 *
239 * Derek Atkins <warlord@MIT.EDU> 94-10-20
240 */
241struct iattr {
242 unsigned int ia_valid;
243 umode_t ia_mode;
92361636
EB
244 kuid_t ia_uid;
245 kgid_t ia_gid;
1da177e4
LT
246 loff_t ia_size;
247 struct timespec ia_atime;
248 struct timespec ia_mtime;
249 struct timespec ia_ctime;
cc4e69de
MS
250
251 /*
25985edc 252 * Not an attribute, but an auxiliary info for filesystems wanting to
cc4e69de
MS
253 * implement an ftruncate() like method. NOTE: filesystem should
254 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
255 */
256 struct file *ia_file;
1da177e4
LT
257};
258
1da177e4
LT
259/*
260 * Includes for diskquotas.
261 */
262#include <linux/quota.h>
263
69c433ed
MS
264/*
265 * Maximum number of layers of fs stack. Needs to be limited to
266 * prevent kernel stack overflow
267 */
268#define FILESYSTEM_MAX_STACK_DEPTH 2
269
994fc28c
ZB
270/**
271 * enum positive_aop_returns - aop return codes with specific semantics
272 *
273 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
274 * completed, that the page is still locked, and
275 * should be considered active. The VM uses this hint
276 * to return the page to the active list -- it won't
277 * be a candidate for writeback again in the near
278 * future. Other callers must be careful to unlock
279 * the page if they get this return. Returned by
280 * writepage();
281 *
282 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
283 * unlocked it and the page might have been truncated.
284 * The caller should back up to acquiring a new page and
285 * trying again. The aop will be taking reasonable
286 * precautions not to livelock. If the caller held a page
287 * reference, it should drop it before retrying. Returned
55144768 288 * by readpage().
994fc28c
ZB
289 *
290 * address_space_operation functions return these large constants to indicate
291 * special semantics to the caller. These are much larger than the bytes in a
292 * page to allow for functions that return the number of bytes operated on in a
293 * given page.
294 */
295
296enum positive_aop_returns {
297 AOP_WRITEPAGE_ACTIVATE = 0x80000,
298 AOP_TRUNCATED_PAGE = 0x80001,
299};
300
afddba49 301#define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
89e10787 302#define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
54566b2c
NP
303#define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
304 * helper code (eg buffer layer)
305 * to clear GFP_FS from alloc */
afddba49 306
1da177e4
LT
307/*
308 * oh the beauties of C type declarations.
309 */
310struct page;
311struct address_space;
312struct writeback_control;
1da177e4 313
8ab22b9a
HH
314/*
315 * "descriptor" for what we're up to with a read.
316 * This allows us to use the same read code yet
317 * have multiple different users of the data that
318 * we read from a file.
319 *
320 * The simplest case just copies the data to user
321 * mode.
322 */
323typedef struct {
324 size_t written;
325 size_t count;
326 union {
327 char __user *buf;
328 void *data;
329 } arg;
330 int error;
331} read_descriptor_t;
332
333typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
334 unsigned long, unsigned long);
2f718ffc 335
1da177e4
LT
336struct address_space_operations {
337 int (*writepage)(struct page *page, struct writeback_control *wbc);
338 int (*readpage)(struct file *, struct page *);
1da177e4
LT
339
340 /* Write back some dirty pages from this mapping. */
341 int (*writepages)(struct address_space *, struct writeback_control *);
342
4741c9fd 343 /* Set a page dirty. Return true if this dirtied it */
1da177e4
LT
344 int (*set_page_dirty)(struct page *page);
345
346 int (*readpages)(struct file *filp, struct address_space *mapping,
347 struct list_head *pages, unsigned nr_pages);
348
afddba49
NP
349 int (*write_begin)(struct file *, struct address_space *mapping,
350 loff_t pos, unsigned len, unsigned flags,
351 struct page **pagep, void **fsdata);
352 int (*write_end)(struct file *, struct address_space *mapping,
353 loff_t pos, unsigned len, unsigned copied,
354 struct page *page, void *fsdata);
355
1da177e4
LT
356 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
357 sector_t (*bmap)(struct address_space *, sector_t);
d47992f8 358 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
27496a8c 359 int (*releasepage) (struct page *, gfp_t);
6072d13c 360 void (*freepage)(struct page *);
d8d3d94b 361 ssize_t (*direct_IO)(int, struct kiocb *, struct iov_iter *iter, loff_t offset);
70688e4d
NP
362 int (*get_xip_mem)(struct address_space *, pgoff_t, int,
363 void **, unsigned long *);
b969c4ab 364 /*
ef277c73
TC
365 * migrate the contents of a page to the specified target. If
366 * migrate_mode is MIGRATE_ASYNC, it must not block.
b969c4ab 367 */
2d1db3b1 368 int (*migratepage) (struct address_space *,
a6bc32b8 369 struct page *, struct page *, enum migrate_mode);
e3db7691 370 int (*launder_page) (struct page *);
c186afb4 371 int (*is_partially_uptodate) (struct page *, unsigned long,
8ab22b9a 372 unsigned long);
b4597226 373 void (*is_dirty_writeback) (struct page *, bool *, bool *);
25718736 374 int (*error_remove_page)(struct address_space *, struct page *);
62c230bc
MG
375
376 /* swapfile support */
a509bc1a
MG
377 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
378 sector_t *span);
379 void (*swap_deactivate)(struct file *file);
1da177e4
LT
380};
381
7dcda1c9
JA
382extern const struct address_space_operations empty_aops;
383
afddba49
NP
384/*
385 * pagecache_write_begin/pagecache_write_end must be used by general code
386 * to write into the pagecache.
387 */
388int pagecache_write_begin(struct file *, struct address_space *mapping,
389 loff_t pos, unsigned len, unsigned flags,
390 struct page **pagep, void **fsdata);
391
392int pagecache_write_end(struct file *, struct address_space *mapping,
393 loff_t pos, unsigned len, unsigned copied,
394 struct page *page, void *fsdata);
395
1da177e4
LT
396struct backing_dev_info;
397struct address_space {
398 struct inode *host; /* owner: inode, block_device */
399 struct radix_tree_root page_tree; /* radix tree of all pages */
19fd6231 400 spinlock_t tree_lock; /* and lock protecting it */
4bb5f5d9 401 atomic_t i_mmap_writable;/* count VM_SHARED mappings */
6b2dbba8 402 struct rb_root i_mmap; /* tree of private and shared mappings */
1da177e4 403 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
3d48ae45 404 struct mutex i_mmap_mutex; /* protect tree, count, list */
08142579 405 /* Protected by tree_lock together with the radix tree */
1da177e4 406 unsigned long nrpages; /* number of total pages */
91b0abe3 407 unsigned long nrshadows; /* number of shadow entries */
1da177e4 408 pgoff_t writeback_index;/* writeback starts here */
f5e54d6e 409 const struct address_space_operations *a_ops; /* methods */
1da177e4
LT
410 unsigned long flags; /* error bits/gfp mask */
411 struct backing_dev_info *backing_dev_info; /* device readahead, etc */
412 spinlock_t private_lock; /* for use by the address_space */
413 struct list_head private_list; /* ditto */
252aa6f5 414 void *private_data; /* ditto */
1da177e4
LT
415} __attribute__((aligned(sizeof(long))));
416 /*
417 * On most architectures that alignment is already the case; but
25985edc 418 * must be enforced here for CRIS, to let the least significant bit
1da177e4
LT
419 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
420 */
87192a2a 421struct request_queue;
1da177e4
LT
422
423struct block_device {
424 dev_t bd_dev; /* not a kdev_t - it's a search key */
f3ccfcda 425 int bd_openers;
1da177e4 426 struct inode * bd_inode; /* will die */
87d8fe1e 427 struct super_block * bd_super;
c039e313 428 struct mutex bd_mutex; /* open/close mutex */
1da177e4 429 struct list_head bd_inodes;
6b4517a7 430 void * bd_claiming;
1da177e4
LT
431 void * bd_holder;
432 int bd_holders;
77ea887e 433 bool bd_write_holder;
641dc636 434#ifdef CONFIG_SYSFS
49731baa 435 struct list_head bd_holder_disks;
641dc636 436#endif
1da177e4
LT
437 struct block_device * bd_contains;
438 unsigned bd_block_size;
439 struct hd_struct * bd_part;
440 /* number of times partitions within this device have been opened. */
441 unsigned bd_part_count;
442 int bd_invalidated;
443 struct gendisk * bd_disk;
87192a2a 444 struct request_queue * bd_queue;
1da177e4 445 struct list_head bd_list;
1da177e4
LT
446 /*
447 * Private data. You must have bd_claim'ed the block_device
448 * to use this. NOTE: bd_claim allows an owner to claim
449 * the same device multiple times, the owner must take special
450 * care to not mess up bd_private for that case.
451 */
452 unsigned long bd_private;
fcccf502
TS
453
454 /* The counter of freeze processes */
455 int bd_fsfreeze_count;
456 /* Mutex for freeze */
457 struct mutex bd_fsfreeze_mutex;
1da177e4
LT
458};
459
460/*
461 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
462 * radix trees
463 */
464#define PAGECACHE_TAG_DIRTY 0
465#define PAGECACHE_TAG_WRITEBACK 1
f446daae 466#define PAGECACHE_TAG_TOWRITE 2
1da177e4
LT
467
468int mapping_tagged(struct address_space *mapping, int tag);
469
470/*
471 * Might pages of this file be mapped into userspace?
472 */
473static inline int mapping_mapped(struct address_space *mapping)
474{
6b2dbba8 475 return !RB_EMPTY_ROOT(&mapping->i_mmap) ||
1da177e4
LT
476 !list_empty(&mapping->i_mmap_nonlinear);
477}
478
479/*
480 * Might pages of this file have been modified in userspace?
481 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
482 * marks vma as VM_SHARED if it is shared, and the file was opened for
483 * writing i.e. vma may be mprotected writable even if now readonly.
4bb5f5d9
DH
484 *
485 * If i_mmap_writable is negative, no new writable mappings are allowed. You
486 * can only deny writable mappings, if none exists right now.
1da177e4
LT
487 */
488static inline int mapping_writably_mapped(struct address_space *mapping)
489{
4bb5f5d9
DH
490 return atomic_read(&mapping->i_mmap_writable) > 0;
491}
492
493static inline int mapping_map_writable(struct address_space *mapping)
494{
495 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
496 0 : -EPERM;
497}
498
499static inline void mapping_unmap_writable(struct address_space *mapping)
500{
501 atomic_dec(&mapping->i_mmap_writable);
502}
503
504static inline int mapping_deny_writable(struct address_space *mapping)
505{
506 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
507 0 : -EBUSY;
508}
509
510static inline void mapping_allow_writable(struct address_space *mapping)
511{
512 atomic_inc(&mapping->i_mmap_writable);
1da177e4
LT
513}
514
515/*
516 * Use sequence counter to get consistent i_size on 32-bit processors.
517 */
518#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
519#include <linux/seqlock.h>
520#define __NEED_I_SIZE_ORDERED
521#define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
522#else
523#define i_size_ordered_init(inode) do { } while (0)
524#endif
525
f19d4a8f
AV
526struct posix_acl;
527#define ACL_NOT_CACHED ((void *)(-1))
528
3ddcd056
LT
529#define IOP_FASTPERM 0x0001
530#define IOP_LOOKUP 0x0002
531#define IOP_NOFOLLOW 0x0004
532
533/*
534 * Keep mostly read-only and often accessed (especially for
535 * the RCU path lookup and 'stat' data) fields at the beginning
536 * of the 'struct inode'
537 */
1da177e4 538struct inode {
44a7d7a8 539 umode_t i_mode;
3ddcd056 540 unsigned short i_opflags;
92361636
EB
541 kuid_t i_uid;
542 kgid_t i_gid;
3ddcd056
LT
543 unsigned int i_flags;
544
545#ifdef CONFIG_FS_POSIX_ACL
546 struct posix_acl *i_acl;
547 struct posix_acl *i_default_acl;
548#endif
549
44a7d7a8
NP
550 const struct inode_operations *i_op;
551 struct super_block *i_sb;
3ddcd056 552 struct address_space *i_mapping;
44a7d7a8 553
13e12d14
LT
554#ifdef CONFIG_SECURITY
555 void *i_security;
556#endif
44a7d7a8 557
3ddcd056
LT
558 /* Stat data, not accessed from path walking */
559 unsigned long i_ino;
a78ef704
MS
560 /*
561 * Filesystems may only read i_nlink directly. They shall use the
562 * following functions for modification:
563 *
564 * (set|clear|inc|drop)_nlink
565 * inode_(inc|dec)_link_count
566 */
567 union {
568 const unsigned int i_nlink;
569 unsigned int __i_nlink;
570 };
3ddcd056 571 dev_t i_rdev;
2f9d3df8 572 loff_t i_size;
3ddcd056
LT
573 struct timespec i_atime;
574 struct timespec i_mtime;
575 struct timespec i_ctime;
6cdbb0ef
TT
576 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
577 unsigned short i_bytes;
2f9d3df8 578 unsigned int i_blkbits;
3ddcd056
LT
579 blkcnt_t i_blocks;
580
581#ifdef __NEED_I_SIZE_ORDERED
582 seqcount_t i_size_seqcount;
583#endif
584
585 /* Misc */
586 unsigned long i_state;
3ddcd056 587 struct mutex i_mutex;
13e12d14 588
44a7d7a8
NP
589 unsigned long dirtied_when; /* jiffies of first dirtying */
590
1da177e4 591 struct hlist_node i_hash;
7ccf19a8
NP
592 struct list_head i_wb_list; /* backing dev IO list */
593 struct list_head i_lru; /* inode LRU list */
1da177e4 594 struct list_head i_sb_list;
fa0d7e3d 595 union {
b3d9b7a3 596 struct hlist_head i_dentry;
fa0d7e3d
NP
597 struct rcu_head i_rcu;
598 };
7a224228 599 u64 i_version;
2f9d3df8 600 atomic_t i_count;
bd5fe6c5 601 atomic_t i_dio_count;
6cdbb0ef 602 atomic_t i_writecount;
d984ea60
MZ
603#ifdef CONFIG_IMA
604 atomic_t i_readcount; /* struct files open RO */
605#endif
99ac48f5 606 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
1da177e4 607 struct file_lock *i_flock;
1da177e4
LT
608 struct address_space i_data;
609#ifdef CONFIG_QUOTA
610 struct dquot *i_dquot[MAXQUOTAS];
611#endif
1da177e4 612 struct list_head i_devices;
4c154168
TT
613 union {
614 struct pipe_inode_info *i_pipe;
eaf796e7 615 struct block_device *i_bdev;
577c4eb0 616 struct cdev *i_cdev;
4c154168 617 };
1da177e4
LT
618
619 __u32 i_generation;
620
3be25f49
EP
621#ifdef CONFIG_FSNOTIFY
622 __u32 i_fsnotify_mask; /* all events this inode cares about */
e61ce867 623 struct hlist_head i_fsnotify_marks;
0eeca283
RL
624#endif
625
8e18e294 626 void *i_private; /* fs or device private pointer */
1da177e4
LT
627};
628
1d3382cb
AV
629static inline int inode_unhashed(struct inode *inode)
630{
631 return hlist_unhashed(&inode->i_hash);
632}
633
f2eace23
IM
634/*
635 * inode->i_mutex nesting subclasses for the lock validator:
636 *
637 * 0: the object of the current VFS operation
638 * 1: parent
639 * 2: child/target
40bd22c9
BF
640 * 3: xattr
641 * 4: second non-directory
d1b72cc6
MS
642 * 5: second parent (when locking independent directories in rename)
643 *
644 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
40bd22c9 645 * non-directories at once.
f2eace23
IM
646 *
647 * The locking order between these classes is
d1b72cc6 648 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
f2eace23
IM
649 */
650enum inode_i_mutex_lock_class
651{
652 I_MUTEX_NORMAL,
653 I_MUTEX_PARENT,
654 I_MUTEX_CHILD,
4df46240 655 I_MUTEX_XATTR,
d1b72cc6
MS
656 I_MUTEX_NONDIR2,
657 I_MUTEX_PARENT2,
f2eace23
IM
658};
659
375e289e
BF
660void lock_two_nondirectories(struct inode *, struct inode*);
661void unlock_two_nondirectories(struct inode *, struct inode*);
662
1da177e4
LT
663/*
664 * NOTE: in a 32bit arch with a preemptable kernel and
665 * an UP compile the i_size_read/write must be atomic
666 * with respect to the local cpu (unlike with preempt disabled),
667 * but they don't need to be atomic with respect to other cpus like in
668 * true SMP (so they need either to either locally disable irq around
669 * the read or for example on x86 they can be still implemented as a
670 * cmpxchg8b without the need of the lock prefix). For SMP compiles
671 * and 64bit archs it makes no difference if preempt is enabled or not.
672 */
48ed214d 673static inline loff_t i_size_read(const struct inode *inode)
1da177e4
LT
674{
675#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
676 loff_t i_size;
677 unsigned int seq;
678
679 do {
680 seq = read_seqcount_begin(&inode->i_size_seqcount);
681 i_size = inode->i_size;
682 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
683 return i_size;
684#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
685 loff_t i_size;
686
687 preempt_disable();
688 i_size = inode->i_size;
689 preempt_enable();
690 return i_size;
691#else
692 return inode->i_size;
693#endif
694}
695
7762f5a0
MS
696/*
697 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
698 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
699 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
700 */
1da177e4
LT
701static inline void i_size_write(struct inode *inode, loff_t i_size)
702{
703#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
74e3d1e1 704 preempt_disable();
1da177e4
LT
705 write_seqcount_begin(&inode->i_size_seqcount);
706 inode->i_size = i_size;
707 write_seqcount_end(&inode->i_size_seqcount);
74e3d1e1 708 preempt_enable();
1da177e4
LT
709#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
710 preempt_disable();
711 inode->i_size = i_size;
712 preempt_enable();
713#else
714 inode->i_size = i_size;
715#endif
716}
717
92361636
EB
718/* Helper functions so that in most cases filesystems will
719 * not need to deal directly with kuid_t and kgid_t and can
720 * instead deal with the raw numeric values that are stored
721 * in the filesystem.
722 */
723static inline uid_t i_uid_read(const struct inode *inode)
724{
725 return from_kuid(&init_user_ns, inode->i_uid);
726}
727
728static inline gid_t i_gid_read(const struct inode *inode)
729{
730 return from_kgid(&init_user_ns, inode->i_gid);
731}
732
733static inline void i_uid_write(struct inode *inode, uid_t uid)
734{
735 inode->i_uid = make_kuid(&init_user_ns, uid);
736}
737
738static inline void i_gid_write(struct inode *inode, gid_t gid)
739{
740 inode->i_gid = make_kgid(&init_user_ns, gid);
741}
742
48ed214d 743static inline unsigned iminor(const struct inode *inode)
1da177e4
LT
744{
745 return MINOR(inode->i_rdev);
746}
747
48ed214d 748static inline unsigned imajor(const struct inode *inode)
1da177e4
LT
749{
750 return MAJOR(inode->i_rdev);
751}
752
753extern struct block_device *I_BDEV(struct inode *inode);
754
755struct fown_struct {
756 rwlock_t lock; /* protects pid, uid, euid fields */
609d7fa9
EB
757 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
758 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
92361636 759 kuid_t uid, euid; /* uid/euid of process setting the owner */
1da177e4
LT
760 int signum; /* posix.1b rt signal to be delivered on IO */
761};
762
763/*
764 * Track a single file's readahead state
765 */
766struct file_ra_state {
937085aa
FW
767 pgoff_t start; /* where readahead started */
768 unsigned int size; /* # of readahead pages */
769 unsigned int async_size; /* do asynchronous readahead when
f9acc8c7 770 there are only # of pages ahead */
5ce1110b 771
937085aa 772 unsigned int ra_pages; /* Maximum readahead window */
1ebf26a9 773 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
f4e6b498 774 loff_t prev_pos; /* Cache last read() position */
1da177e4 775};
1da177e4 776
5ce1110b
FW
777/*
778 * Check if @index falls in the readahead windows.
779 */
780static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
781{
f9acc8c7
FW
782 return (index >= ra->start &&
783 index < ra->start + ra->size);
5ce1110b
FW
784}
785
1da177e4 786struct file {
2f512016 787 union {
4f5e65a1 788 struct llist_node fu_llist;
2f512016
ED
789 struct rcu_head fu_rcuhead;
790 } f_u;
0f7fc9e4
JJS
791 struct path f_path;
792#define f_dentry f_path.dentry
dd37978c 793 struct inode *f_inode; /* cached value */
99ac48f5 794 const struct file_operations *f_op;
ef3d0fd2
AK
795
796 /*
9c225f26 797 * Protects f_ep_links, f_flags.
ef3d0fd2
AK
798 * Must not be taken from IRQ context.
799 */
800 spinlock_t f_lock;
516e0cc5 801 atomic_long_t f_count;
1da177e4 802 unsigned int f_flags;
aeb5d727 803 fmode_t f_mode;
9c225f26 804 struct mutex f_pos_lock;
1da177e4
LT
805 loff_t f_pos;
806 struct fown_struct f_owner;
d76b0d9b 807 const struct cred *f_cred;
1da177e4
LT
808 struct file_ra_state f_ra;
809
2b47c361 810 u64 f_version;
50462062 811#ifdef CONFIG_SECURITY
1da177e4 812 void *f_security;
50462062 813#endif
1da177e4
LT
814 /* needed for tty driver, and maybe others */
815 void *private_data;
816
817#ifdef CONFIG_EPOLL
818 /* Used by fs/eventpoll.c to link all the hooks to this file */
819 struct list_head f_ep_links;
28d82dc1 820 struct list_head f_tfile_llink;
1da177e4
LT
821#endif /* #ifdef CONFIG_EPOLL */
822 struct address_space *f_mapping;
bd2a31d5 823} __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
1da177e4 824
990d6c2d
AK
825struct file_handle {
826 __u32 handle_bytes;
827 int handle_type;
828 /* file identifier */
829 unsigned char f_handle[0];
830};
831
cb0942b8
AV
832static inline struct file *get_file(struct file *f)
833{
834 atomic_long_inc(&f->f_count);
835 return f;
836}
d7065da0 837#define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
516e0cc5 838#define file_count(x) atomic_long_read(&(x)->f_count)
1da177e4
LT
839
840#define MAX_NON_LFS ((1UL<<31) - 1)
841
842/* Page cache limit. The filesystems should put that into their s_maxbytes
843 limits, otherwise bad things can happen in VM. */
844#if BITS_PER_LONG==32
2bd2c194 845#define MAX_LFS_FILESIZE (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
1da177e4 846#elif BITS_PER_LONG==64
614c321f 847#define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
1da177e4
LT
848#endif
849
850#define FL_POSIX 1
851#define FL_FLOCK 2
617588d5 852#define FL_DELEG 4 /* NFSv4 delegation */
1da177e4 853#define FL_ACCESS 8 /* not trying to lock, just looking */
f475ae95 854#define FL_EXISTS 16 /* when unlocking, test for existence */
1da177e4 855#define FL_LEASE 32 /* lease held on this file */
75e1fcc0 856#define FL_CLOSE 64 /* unlock on close */
1da177e4 857#define FL_SLEEP 128 /* A blocking lock */
778fc546
BF
858#define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
859#define FL_UNLOCK_PENDING 512 /* Lease is being broken */
cff2fce5 860#define FL_OFDLCK 1024 /* lock is "owned" by struct file */
1da177e4 861
bde74e4b
MS
862/*
863 * Special return value from posix_lock_file() and vfs_lock_file() for
864 * asynchronous locking.
865 */
866#define FILE_LOCK_DEFERRED 1
867
7ca76311 868/* legacy typedef, should eventually be removed */
17fa388d 869typedef void *fl_owner_t;
1da177e4
LT
870
871struct file_lock_operations {
1da177e4
LT
872 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
873 void (*fl_release_private)(struct file_lock *);
874};
875
876struct lock_manager_operations {
8fb47a4f 877 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
3999e493 878 unsigned long (*lm_owner_key)(struct file_lock *);
5c97d7b1
KM
879 void (*lm_get_owner)(struct file_lock *, struct file_lock *);
880 void (*lm_put_owner)(struct file_lock *);
8fb47a4f 881 void (*lm_notify)(struct file_lock *); /* unblock callback */
d0449b90 882 int (*lm_grant)(struct file_lock *, int);
4d01b7f5 883 bool (*lm_break)(struct file_lock *);
c45198ed 884 int (*lm_change)(struct file_lock **, int, struct list_head *);
1c7dd2ff 885 void (*lm_setup)(struct file_lock *, void **);
1da177e4
LT
886};
887
af558e33
BF
888struct lock_manager {
889 struct list_head list;
890};
891
5ccb0066
SK
892struct net;
893void locks_start_grace(struct net *, struct lock_manager *);
af558e33 894void locks_end_grace(struct lock_manager *);
5ccb0066 895int locks_in_grace(struct net *);
af558e33 896
1da177e4
LT
897/* that will die - we need it for nfs_lock_info */
898#include <linux/nfs_fs_i.h>
899
1cb36012
JL
900/*
901 * struct file_lock represents a generic "file lock". It's used to represent
902 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
903 * note that the same struct is used to represent both a request for a lock and
904 * the lock itself, but the same object is never used for both.
905 *
906 * FIXME: should we create a separate "struct lock_request" to help distinguish
907 * these two uses?
908 *
909 * The i_flock list is ordered by:
910 *
911 * 1) lock type -- FL_LEASEs first, then FL_FLOCK, and finally FL_POSIX
912 * 2) lock owner
913 * 3) lock range start
914 * 4) lock range end
915 *
916 * Obviously, the last two criteria only matter for POSIX locks.
917 */
1da177e4
LT
918struct file_lock {
919 struct file_lock *fl_next; /* singly linked list for this inode */
139ca04e 920 struct hlist_node fl_link; /* node in global lists */
1da177e4
LT
921 struct list_head fl_block; /* circular list of blocked processes */
922 fl_owner_t fl_owner;
710b7216 923 unsigned int fl_flags;
afc1246f 924 unsigned char fl_type;
1da177e4 925 unsigned int fl_pid;
7012b02a 926 int fl_link_cpu; /* what cpu's list is this on? */
ab1f1611 927 struct pid *fl_nspid;
1da177e4
LT
928 wait_queue_head_t fl_wait;
929 struct file *fl_file;
1da177e4
LT
930 loff_t fl_start;
931 loff_t fl_end;
932
933 struct fasync_struct * fl_fasync; /* for lease break notifications */
778fc546
BF
934 /* for lease breaks: */
935 unsigned long fl_break_time;
936 unsigned long fl_downgrade_time;
1da177e4 937
6aed6285 938 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
7b021967 939 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1da177e4
LT
940 union {
941 struct nfs_lock_info nfs_fl;
8d0a8a9d 942 struct nfs4_lock_info nfs4_fl;
e8d6c554
DH
943 struct {
944 struct list_head link; /* link in AFS vnode's pending_locks list */
945 int state; /* state of grant or error if -ve */
946 } afs;
1da177e4
LT
947 } fl_u;
948};
949
950/* The following constant reflects the upper bound of the file/locking space */
951#ifndef OFFSET_MAX
952#define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
953#define OFFSET_MAX INT_LIMIT(loff_t)
954#define OFFT_OFFSET_MAX INT_LIMIT(off_t)
955#endif
956
1da177e4
LT
957#include <linux/fcntl.h>
958
bfcd17a6
TP
959extern void send_sigio(struct fown_struct *fown, int fd, int band);
960
bfcd17a6 961#ifdef CONFIG_FILE_LOCKING
c1e62b8f 962extern int fcntl_getlk(struct file *, unsigned int, struct flock __user *);
c293621b
PS
963extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
964 struct flock __user *);
1da177e4
LT
965
966#if BITS_PER_LONG == 32
c1e62b8f 967extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 __user *);
c293621b
PS
968extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
969 struct flock64 __user *);
1da177e4
LT
970#endif
971
1da177e4
LT
972extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
973extern int fcntl_getlease(struct file *filp);
974
975/* fs/locks.c */
05fa3135 976void locks_free_lock(struct file_lock *fl);
1da177e4 977extern void locks_init_lock(struct file_lock *);
c5b1f0d9 978extern struct file_lock * locks_alloc_lock(void);
1da177e4 979extern void locks_copy_lock(struct file_lock *, struct file_lock *);
3fe0fff1 980extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1da177e4 981extern void locks_remove_posix(struct file *, fl_owner_t);
78ed8a13 982extern void locks_remove_file(struct file *);
a9e61e25 983extern void locks_release_private(struct file_lock *);
6d34ac19 984extern void posix_test_lock(struct file *, struct file_lock *);
150b3934 985extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1da177e4 986extern int posix_lock_file_wait(struct file *, struct file_lock *);
f891a29f 987extern int posix_unblock_lock(struct file_lock *);
3ee17abd 988extern int vfs_test_lock(struct file *, struct file_lock *);
150b3934 989extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
9b9d2ab4 990extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1da177e4 991extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
df4e8d2c 992extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1da177e4 993extern void lease_get_mtime(struct inode *, struct timespec *time);
e6f5c789
JL
994extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
995extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
c45198ed 996extern int lease_modify(struct file_lock **, int, struct list_head *);
bfcd17a6 997#else /* !CONFIG_FILE_LOCKING */
c1e62b8f
JL
998static inline int fcntl_getlk(struct file *file, unsigned int cmd,
999 struct flock __user *user)
c2aca5e5
SW
1000{
1001 return -EINVAL;
1002}
1003
1004static inline int fcntl_setlk(unsigned int fd, struct file *file,
1005 unsigned int cmd, struct flock __user *user)
1006{
1007 return -EACCES;
1008}
1009
bfcd17a6 1010#if BITS_PER_LONG == 32
c1e62b8f
JL
1011static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1012 struct flock64 __user *user)
c2aca5e5
SW
1013{
1014 return -EINVAL;
1015}
1016
1017static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1018 unsigned int cmd, struct flock64 __user *user)
1019{
1020 return -EACCES;
1021}
bfcd17a6 1022#endif
c2aca5e5
SW
1023static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1024{
1b2b32dc 1025 return -EINVAL;
c2aca5e5
SW
1026}
1027
1028static inline int fcntl_getlease(struct file *filp)
1029{
1b2b32dc 1030 return F_UNLCK;
c2aca5e5
SW
1031}
1032
1033static inline void locks_init_lock(struct file_lock *fl)
1034{
1035 return;
1036}
1037
3fe0fff1 1038static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
c2aca5e5
SW
1039{
1040 return;
1041}
1042
1043static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1044{
1045 return;
1046}
1047
1048static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1049{
1050 return;
1051}
1052
78ed8a13 1053static inline void locks_remove_file(struct file *filp)
c2aca5e5
SW
1054{
1055 return;
1056}
1057
1058static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1059{
1060 return;
1061}
1062
1063static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1064 struct file_lock *conflock)
1065{
1066 return -ENOLCK;
1067}
1068
1069static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1070{
1071 return -ENOLCK;
1072}
1073
f891a29f 1074static inline int posix_unblock_lock(struct file_lock *waiter)
c2aca5e5
SW
1075{
1076 return -ENOENT;
1077}
1078
1079static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1080{
1081 return 0;
1082}
1083
1084static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1085 struct file_lock *fl, struct file_lock *conf)
1086{
1087 return -ENOLCK;
1088}
1089
1090static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1091{
1092 return 0;
1093}
1094
1095static inline int flock_lock_file_wait(struct file *filp,
1096 struct file_lock *request)
1097{
1098 return -ENOLCK;
1099}
1100
df4e8d2c 1101static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
c2aca5e5
SW
1102{
1103 return 0;
1104}
1105
1106static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1107{
1108 return;
1109}
1110
1111static inline int generic_setlease(struct file *filp, long arg,
e6f5c789 1112 struct file_lock **flp, void **priv)
c2aca5e5
SW
1113{
1114 return -EINVAL;
1115}
1116
1117static inline int vfs_setlease(struct file *filp, long arg,
e6f5c789 1118 struct file_lock **lease, void **priv)
c2aca5e5
SW
1119{
1120 return -EINVAL;
1121}
1122
c45198ed
JL
1123static inline int lease_modify(struct file_lock **before, int arg,
1124 struct list_head *dispose)
c2aca5e5
SW
1125{
1126 return -EINVAL;
1127}
bfcd17a6
TP
1128#endif /* !CONFIG_FILE_LOCKING */
1129
1da177e4
LT
1130
1131struct fasync_struct {
989a2979
ED
1132 spinlock_t fa_lock;
1133 int magic;
1134 int fa_fd;
1135 struct fasync_struct *fa_next; /* singly linked list */
1136 struct file *fa_file;
1137 struct rcu_head fa_rcu;
1da177e4
LT
1138};
1139
1140#define FASYNC_MAGIC 0x4601
1141
1142/* SMP safe fasync helpers: */
1143extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
f7347ce4
LT
1144extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1145extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1146extern struct fasync_struct *fasync_alloc(void);
1147extern void fasync_free(struct fasync_struct *);
1148
1da177e4
LT
1149/* can be called from interrupts */
1150extern void kill_fasync(struct fasync_struct **, int, int);
1da177e4 1151
e0b93edd
JL
1152extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1153extern void f_setown(struct file *filp, unsigned long arg, int force);
1da177e4 1154extern void f_delown(struct file *filp);
609d7fa9 1155extern pid_t f_getown(struct file *filp);
1da177e4
LT
1156extern int send_sigurg(struct fown_struct *fown);
1157
5accdf82
JK
1158struct mm_struct;
1159
1da177e4
LT
1160/*
1161 * Umount options
1162 */
1163
1164#define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1165#define MNT_DETACH 0x00000002 /* Just detach from the tree */
1166#define MNT_EXPIRE 0x00000004 /* Mark for expiry */
db1f05bb
MS
1167#define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1168#define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1da177e4
LT
1169
1170extern struct list_head super_blocks;
1171extern spinlock_t sb_lock;
1172
5accdf82
JK
1173/* Possible states of 'frozen' field */
1174enum {
1175 SB_UNFROZEN = 0, /* FS is unfrozen */
1176 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
5accdf82
JK
1177 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1178 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1179 * internal threads if needed) */
1180 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1181};
1182
1183#define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1184
1185struct sb_writers {
1186 /* Counters for counting writers at each level */
1187 struct percpu_counter counter[SB_FREEZE_LEVELS];
1188 wait_queue_head_t wait; /* queue for waiting for
1189 writers / faults to finish */
1190 int frozen; /* Is sb frozen? */
1191 wait_queue_head_t wait_unfrozen; /* queue for waiting for
1192 sb to be thawed */
1193#ifdef CONFIG_DEBUG_LOCK_ALLOC
1194 struct lockdep_map lock_map[SB_FREEZE_LEVELS];
1195#endif
1196};
1197
1da177e4
LT
1198struct super_block {
1199 struct list_head s_list; /* Keep this first */
1200 dev_t s_dev; /* search index; _not_ kdev_t */
270ba5f7
RK
1201 unsigned char s_blocksize_bits;
1202 unsigned long s_blocksize;
42cb56ae 1203 loff_t s_maxbytes; /* Max file size */
1da177e4 1204 struct file_system_type *s_type;
ee9b6d61 1205 const struct super_operations *s_op;
61e225dc 1206 const struct dquot_operations *dq_op;
0d54b217 1207 const struct quotactl_ops *s_qcop;
39655164 1208 const struct export_operations *s_export_op;
1da177e4
LT
1209 unsigned long s_flags;
1210 unsigned long s_magic;
1211 struct dentry *s_root;
1212 struct rw_semaphore s_umount;
1da177e4 1213 int s_count;
1da177e4 1214 atomic_t s_active;
50462062 1215#ifdef CONFIG_SECURITY
1da177e4 1216 void *s_security;
50462062 1217#endif
bb435453 1218 const struct xattr_handler **s_xattr;
1da177e4
LT
1219
1220 struct list_head s_inodes; /* all inodes */
ceb5bdc2 1221 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
39f7c4db 1222 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1da177e4 1223 struct block_device *s_bdev;
32a88aa1 1224 struct backing_dev_info *s_bdi;
acaebfd8 1225 struct mtd_info *s_mtd;
a5166169 1226 struct hlist_node s_instances;
2c5f648a 1227 unsigned int s_quota_types; /* Bitmask of supported quota types */
1da177e4
LT
1228 struct quota_info s_dquot; /* Diskquota specific options */
1229
5accdf82 1230 struct sb_writers s_writers;
1da177e4
LT
1231
1232 char s_id[32]; /* Informational name */
0bba0169 1233 u8 s_uuid[16]; /* UUID */
1da177e4
LT
1234
1235 void *s_fs_info; /* Filesystem private info */
8de52778 1236 unsigned int s_max_links;
30c40d2c 1237 fmode_t s_mode;
1da177e4 1238
270ba5f7
RK
1239 /* Granularity of c/m/atime in ns.
1240 Cannot be worse than a second */
1241 u32 s_time_gran;
1242
1da177e4
LT
1243 /*
1244 * The next field is for VFS *only*. No filesystems have any business
1245 * even looking at it. You had been warned.
1246 */
a11f3a05 1247 struct mutex s_vfs_rename_mutex; /* Kludge */
1da177e4 1248
79c0b2df
MS
1249 /*
1250 * Filesystem subtype. If non-empty the filesystem type field
1251 * in /proc/mounts will be "type.subtype"
1252 */
1253 char *s_subtype;
b3b304a2
MS
1254
1255 /*
1256 * Saved mount options for lazy filesystems using
1257 * generic_show_options()
1258 */
4d2deb40 1259 char __rcu *s_options;
c8aebb0c 1260 const struct dentry_operations *s_d_op; /* default d_op for dentries */
9fdfdcf1
DM
1261
1262 /*
1263 * Saved pool identifier for cleancache (-1 means none)
1264 */
1265 int cleancache_poolid;
b0d40c92
DC
1266
1267 struct shrinker s_shrink; /* per-sb shrinker handle */
4ed5e82f 1268
7ada4db8
MS
1269 /* Number of inodes with nlink == 0 but still referenced */
1270 atomic_long_t s_remove_count;
1271
4ed5e82f
MS
1272 /* Being remounted read-only */
1273 int s_readonly_remount;
7b7a8665
CH
1274
1275 /* AIO completions deferred from interrupt context */
1276 struct workqueue_struct *s_dio_done_wq;
215752fc 1277 struct hlist_head s_pins;
f6041567
DC
1278
1279 /*
1280 * Keep the lru lists last in the structure so they always sit on their
1281 * own individual cachelines.
1282 */
1283 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1284 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
e2fec7c3 1285 struct rcu_head rcu;
69c433ed
MS
1286
1287 /*
1288 * Indicates how deep in a filesystem stack this SB is
1289 */
1290 int s_stack_depth;
1da177e4
LT
1291};
1292
1293extern struct timespec current_fs_time(struct super_block *sb);
1294
1295/*
1296 * Snapshotting support.
1297 */
1da177e4 1298
5accdf82
JK
1299void __sb_end_write(struct super_block *sb, int level);
1300int __sb_start_write(struct super_block *sb, int level, bool wait);
1301
1302/**
1303 * sb_end_write - drop write access to a superblock
1304 * @sb: the super we wrote to
1305 *
1306 * Decrement number of writers to the filesystem. Wake up possible waiters
1307 * wanting to freeze the filesystem.
1308 */
1309static inline void sb_end_write(struct super_block *sb)
1310{
1311 __sb_end_write(sb, SB_FREEZE_WRITE);
1312}
1313
1314/**
1315 * sb_end_pagefault - drop write access to a superblock from a page fault
1316 * @sb: the super we wrote to
1317 *
1318 * Decrement number of processes handling write page fault to the filesystem.
1319 * Wake up possible waiters wanting to freeze the filesystem.
1320 */
1321static inline void sb_end_pagefault(struct super_block *sb)
1322{
1323 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1324}
1325
1326/**
1327 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1328 * @sb: the super we wrote to
1329 *
1330 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1331 * waiters wanting to freeze the filesystem.
1332 */
1333static inline void sb_end_intwrite(struct super_block *sb)
1334{
1335 __sb_end_write(sb, SB_FREEZE_FS);
1336}
1337
1338/**
1339 * sb_start_write - get write access to a superblock
1340 * @sb: the super we write to
1341 *
1342 * When a process wants to write data or metadata to a file system (i.e. dirty
1343 * a page or an inode), it should embed the operation in a sb_start_write() -
1344 * sb_end_write() pair to get exclusion against file system freezing. This
1345 * function increments number of writers preventing freezing. If the file
1346 * system is already frozen, the function waits until the file system is
1347 * thawed.
1348 *
1349 * Since freeze protection behaves as a lock, users have to preserve
1350 * ordering of freeze protection and other filesystem locks. Generally,
1351 * freeze protection should be the outermost lock. In particular, we have:
1352 *
1353 * sb_start_write
1354 * -> i_mutex (write path, truncate, directory ops, ...)
1355 * -> s_umount (freeze_super, thaw_super)
1356 */
1357static inline void sb_start_write(struct super_block *sb)
1358{
1359 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1360}
1361
1362static inline int sb_start_write_trylock(struct super_block *sb)
1363{
1364 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1365}
1366
1367/**
1368 * sb_start_pagefault - get write access to a superblock from a page fault
1369 * @sb: the super we write to
1370 *
1371 * When a process starts handling write page fault, it should embed the
1372 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1373 * exclusion against file system freezing. This is needed since the page fault
1374 * is going to dirty a page. This function increments number of running page
1375 * faults preventing freezing. If the file system is already frozen, the
1376 * function waits until the file system is thawed.
1377 *
1378 * Since page fault freeze protection behaves as a lock, users have to preserve
1379 * ordering of freeze protection and other filesystem locks. It is advised to
1380 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1381 * handling code implies lock dependency:
1382 *
1383 * mmap_sem
1384 * -> sb_start_pagefault
1385 */
1386static inline void sb_start_pagefault(struct super_block *sb)
1387{
1388 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1389}
1390
1391/*
1392 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1393 * @sb: the super we write to
1394 *
1395 * This is the third level of protection against filesystem freezing. It is
1396 * free for use by a filesystem. The only requirement is that it must rank
1397 * below sb_start_pagefault.
1398 *
1399 * For example filesystem can call sb_start_intwrite() when starting a
1400 * transaction which somewhat eases handling of freezing for internal sources
1401 * of filesystem changes (internal fs threads, discarding preallocation on file
1402 * close, etc.).
1403 */
1404static inline void sb_start_intwrite(struct super_block *sb)
1405{
1406 __sb_start_write(sb, SB_FREEZE_FS, true);
1407}
1da177e4 1408
1da177e4 1409
2e149670 1410extern bool inode_owner_or_capable(const struct inode *inode);
3bd858ab 1411
1da177e4
LT
1412/*
1413 * VFS helper functions..
1414 */
312b63fb 1415extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
18bb1db3 1416extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1a67aafb 1417extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
db2e747b 1418extern int vfs_symlink(struct inode *, struct dentry *, const char *);
146a8595 1419extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1da177e4 1420extern int vfs_rmdir(struct inode *, struct dentry *);
b21996e3 1421extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
520c8b16 1422extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
787fb6bc 1423extern int vfs_whiteout(struct inode *, struct dentry *);
1da177e4
LT
1424
1425/*
1426 * VFS dentry helper functions.
1427 */
1428extern void dentry_unhash(struct dentry *dentry);
1429
8c744fb8
CH
1430/*
1431 * VFS file helper functions.
1432 */
a1bd120d 1433extern void inode_init_owner(struct inode *inode, const struct inode *dir,
62bb1091 1434 umode_t mode);
c4b929b8
MF
1435/*
1436 * VFS FS_IOC_FIEMAP helper definitions.
1437 */
1438struct fiemap_extent_info {
1439 unsigned int fi_flags; /* Flags as passed from user */
1440 unsigned int fi_extents_mapped; /* Number of mapped extents */
1441 unsigned int fi_extents_max; /* Size of fiemap_extent array */
ecf5632d
NK
1442 struct fiemap_extent __user *fi_extents_start; /* Start of
1443 fiemap_extent array */
c4b929b8
MF
1444};
1445int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1446 u64 phys, u64 len, u32 flags);
1447int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1448
1da177e4
LT
1449/*
1450 * File types
1451 *
1452 * NOTE! These match bits 12..15 of stat.st_mode
1453 * (ie "(i_mode >> 12) & 15").
1454 */
1455#define DT_UNKNOWN 0
1456#define DT_FIFO 1
1457#define DT_CHR 2
1458#define DT_DIR 4
1459#define DT_BLK 6
1460#define DT_REG 8
1461#define DT_LNK 10
1462#define DT_SOCK 12
1463#define DT_WHT 14
1464
1da177e4
LT
1465/*
1466 * This is the "filldir" function type, used by readdir() to let
1467 * the kernel specify what kind of dirent layout it wants to have.
1468 * This allows the kernel to read directories into kernel space or
1469 * to have different dirent layouts depending on the binary type.
1470 */
afefdbb2 1471typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
5c0ba4e0 1472struct dir_context {
ac6614b7 1473 const filldir_t actor;
bb6f619b 1474 loff_t pos;
5c0ba4e0 1475};
bb6f619b 1476
08f85851 1477struct block_device_operations;
1da177e4 1478
1da177e4
LT
1479/* These macros are for out of kernel modules to test that
1480 * the kernel supports the unlocked_ioctl and compat_ioctl
1481 * fields in struct file_operations. */
1482#define HAVE_COMPAT_IOCTL 1
1483#define HAVE_UNLOCKED_IOCTL 1
1484
293bc982
AV
1485struct iov_iter;
1486
1da177e4
LT
1487struct file_operations {
1488 struct module *owner;
1489 loff_t (*llseek) (struct file *, loff_t, int);
1490 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1da177e4 1491 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
027445c3
BP
1492 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1493 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
293bc982
AV
1494 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1495 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
bb6f619b 1496 int (*iterate) (struct file *, struct dir_context *);
1da177e4 1497 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1da177e4
LT
1498 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1499 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1500 int (*mmap) (struct file *, struct vm_area_struct *);
1501 int (*open) (struct inode *, struct file *);
75e1fcc0 1502 int (*flush) (struct file *, fl_owner_t id);
1da177e4 1503 int (*release) (struct inode *, struct file *);
02c24a82 1504 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1da177e4
LT
1505 int (*aio_fsync) (struct kiocb *, int datasync);
1506 int (*fasync) (int, struct file *, int);
1507 int (*lock) (struct file *, int, struct file_lock *);
1da177e4
LT
1508 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1509 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1510 int (*check_flags)(int);
1da177e4 1511 int (*flock) (struct file *, int, struct file_lock *);
cbb7e577
JA
1512 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1513 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
e6f5c789 1514 int (*setlease)(struct file *, long, struct file_lock **, void **);
2fe17c10
CH
1515 long (*fallocate)(struct file *file, int mode, loff_t offset,
1516 loff_t len);
55985dd7 1517 int (*show_fdinfo)(struct seq_file *m, struct file *f);
1da177e4
LT
1518};
1519
1520struct inode_operations {
00cd8dd3 1521 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
44a7d7a8 1522 void * (*follow_link) (struct dentry *, struct nameidata *);
10556cb2 1523 int (*permission) (struct inode *, int);
4e34e719 1524 struct posix_acl * (*get_acl)(struct inode *, int);
44a7d7a8
NP
1525
1526 int (*readlink) (struct dentry *, char __user *,int);
1527 void (*put_link) (struct dentry *, struct nameidata *, void *);
1528
ebfc3b49 1529 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1da177e4
LT
1530 int (*link) (struct dentry *,struct inode *,struct dentry *);
1531 int (*unlink) (struct inode *,struct dentry *);
1532 int (*symlink) (struct inode *,struct dentry *,const char *);
18bb1db3 1533 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1da177e4 1534 int (*rmdir) (struct inode *,struct dentry *);
1a67aafb 1535 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1da177e4
LT
1536 int (*rename) (struct inode *, struct dentry *,
1537 struct inode *, struct dentry *);
520c8b16
MS
1538 int (*rename2) (struct inode *, struct dentry *,
1539 struct inode *, struct dentry *, unsigned int);
1da177e4
LT
1540 int (*setattr) (struct dentry *, struct iattr *);
1541 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1542 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1543 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1544 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1545 int (*removexattr) (struct dentry *, const char *);
c4b929b8
MF
1546 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1547 u64 len);
c3b2da31 1548 int (*update_time)(struct inode *, struct timespec *, int);
d9585277 1549 int (*atomic_open)(struct inode *, struct dentry *,
30d90494 1550 struct file *, unsigned open_flag,
d9585277 1551 umode_t create_mode, int *opened);
60545d0d 1552 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
893d46e4 1553 int (*set_acl)(struct inode *, struct posix_acl *, int);
4aa7c634
MS
1554
1555 /* WARNING: probably going away soon, do not use! */
1556 int (*dentry_open)(struct dentry *, struct file *, const struct cred *);
44a7d7a8 1557} ____cacheline_aligned;
1da177e4 1558
eed4e51f 1559ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
fcf63409
CY
1560 unsigned long nr_segs, unsigned long fast_segs,
1561 struct iovec *fast_pointer,
ac34ebb3 1562 struct iovec **ret_pointer);
eed4e51f 1563
1da177e4
LT
1564extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1565extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1566extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1567 unsigned long, loff_t *);
1568extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1569 unsigned long, loff_t *);
1570
1da177e4
LT
1571struct super_operations {
1572 struct inode *(*alloc_inode)(struct super_block *sb);
1573 void (*destroy_inode)(struct inode *);
1574
aa385729 1575 void (*dirty_inode) (struct inode *, int flags);
a9185b41 1576 int (*write_inode) (struct inode *, struct writeback_control *wbc);
45321ac5 1577 int (*drop_inode) (struct inode *);
be7ce416 1578 void (*evict_inode) (struct inode *);
1da177e4 1579 void (*put_super) (struct super_block *);
1da177e4 1580 int (*sync_fs)(struct super_block *sb, int wait);
c4be0c1d
TS
1581 int (*freeze_fs) (struct super_block *);
1582 int (*unfreeze_fs) (struct super_block *);
726c3342 1583 int (*statfs) (struct dentry *, struct kstatfs *);
1da177e4 1584 int (*remount_fs) (struct super_block *, int *, char *);
42faad99 1585 void (*umount_begin) (struct super_block *);
1da177e4 1586
34c80b1d 1587 int (*show_options)(struct seq_file *, struct dentry *);
d861c630 1588 int (*show_devname)(struct seq_file *, struct dentry *);
a6322de6 1589 int (*show_path)(struct seq_file *, struct dentry *);
64132379 1590 int (*show_stats)(struct seq_file *, struct dentry *);
0e51a720 1591#ifdef CONFIG_QUOTA
1da177e4
LT
1592 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1593 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
0e51a720 1594#endif
87d8fe1e 1595 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
9b17c623
DC
1596 long (*nr_cached_objects)(struct super_block *, int);
1597 long (*free_cached_objects)(struct super_block *, long, int);
1da177e4
LT
1598};
1599
bbc1096a
DH
1600/*
1601 * Inode flags - they have no relation to superblock flags now
1602 */
1603#define S_SYNC 1 /* Writes are synced at once */
1604#define S_NOATIME 2 /* Do not update access times */
1605#define S_APPEND 4 /* Append-only file */
1606#define S_IMMUTABLE 8 /* Immutable file */
1607#define S_DEAD 16 /* removed, but still open directory */
1608#define S_NOQUOTA 32 /* Inode is not counted to quota */
1609#define S_DIRSYNC 64 /* Directory modifications are synchronous */
1610#define S_NOCMTIME 128 /* Do not update file c/mtime */
1611#define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1612#define S_PRIVATE 512 /* Inode is fs-internal */
1613#define S_IMA 1024 /* Inode has an associated IMA struct */
1614#define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1615#define S_NOSEC 4096 /* no suid or xattr security attributes */
1616
1617/*
1618 * Note that nosuid etc flags are inode-specific: setting some file-system
1619 * flags just means all the inodes inherit those flags by default. It might be
1620 * possible to override it selectively if you really wanted to with some
1621 * ioctl() that is not currently implemented.
1622 *
1623 * Exception: MS_RDONLY is always applied to the entire file system.
1624 *
1625 * Unfortunately, it is possible to change a filesystems flags with it mounted
1626 * with files in use. This means that all of the inodes will not have their
1627 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1628 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1629 */
1630#define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1631
1632#define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1633#define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1634 ((inode)->i_flags & S_SYNC))
1635#define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1636 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1637#define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1638#define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1639#define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1640
1641#define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1642#define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1643#define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1644#define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1645
1646#define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1647#define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1648#define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1649#define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1650#define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1651#define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1652#define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1653
787fb6bc
MS
1654#define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1655 (inode)->i_rdev == WHITEOUT_DEV)
1656
1c0eeaf5 1657/*
250df6ed 1658 * Inode state bits. Protected by inode->i_lock
1c0eeaf5
JE
1659 *
1660 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1661 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1662 *
1663 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1664 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1665 * various stages of removing an inode.
1666 *
eaff8079 1667 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1c0eeaf5 1668 *
e7ca2d41
JE
1669 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1670 * fdatasync(). i_atime is the usual cause.
000cb48e
JK
1671 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1672 * these changes separately from I_DIRTY_SYNC so that we
1673 * don't have to write inode on fdatasync() when only
1674 * mtime has changed in it.
1c0eeaf5 1675 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
eaff8079
CH
1676 * I_NEW Serves as both a mutex and completion notification.
1677 * New inodes set I_NEW. If two processes both create
1678 * the same inode, one of them will release its inode and
1679 * wait for I_NEW to be released before returning.
1680 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1681 * also cause waiting on I_NEW, without I_NEW actually
1682 * being set. find_inode() uses this to prevent returning
1683 * nearly-dead inodes.
1c0eeaf5
JE
1684 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1685 * is zero. I_FREEING must be set when I_WILL_FREE is
1686 * cleared.
1687 * I_FREEING Set when inode is about to be freed but still has dirty
1688 * pages or buffers attached or the inode itself is still
1689 * dirty.
dbd5768f
JK
1690 * I_CLEAR Added by clear_inode(). In this state the inode is
1691 * clean and can be destroyed. Inode keeps I_FREEING.
1c0eeaf5
JE
1692 *
1693 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1694 * prohibited for many purposes. iget() must wait for
1695 * the inode to be completely released, then create it
1696 * anew. Other functions will just ignore such inodes,
eaff8079 1697 * if appropriate. I_NEW is used for waiting.
1c0eeaf5 1698 *
169ebd90
JK
1699 * I_SYNC Writeback of inode is running. The bit is set during
1700 * data writeback, and cleared with a wakeup on the bit
1701 * address once it is done. The bit is also used to pin
1702 * the inode in memory for flusher thread.
1c0eeaf5 1703 *
bd5fe6c5
CH
1704 * I_REFERENCED Marks the inode as recently references on the LRU list.
1705 *
1706 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1707 *
1c0eeaf5 1708 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1c0eeaf5 1709 */
9e38d86f
NP
1710#define I_DIRTY_SYNC (1 << 0)
1711#define I_DIRTY_DATASYNC (1 << 1)
1712#define I_DIRTY_PAGES (1 << 2)
eaff8079
CH
1713#define __I_NEW 3
1714#define I_NEW (1 << __I_NEW)
9e38d86f
NP
1715#define I_WILL_FREE (1 << 4)
1716#define I_FREEING (1 << 5)
1717#define I_CLEAR (1 << 6)
eaff8079 1718#define __I_SYNC 7
1c0eeaf5 1719#define I_SYNC (1 << __I_SYNC)
9e38d86f 1720#define I_REFERENCED (1 << 8)
bd5fe6c5
CH
1721#define __I_DIO_WAKEUP 9
1722#define I_DIO_WAKEUP (1 << I_DIO_WAKEUP)
f4e0c30c 1723#define I_LINKABLE (1 << 10)
1da177e4
LT
1724
1725#define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1726
1727extern void __mark_inode_dirty(struct inode *, int);
1728static inline void mark_inode_dirty(struct inode *inode)
1729{
1730 __mark_inode_dirty(inode, I_DIRTY);
1731}
1732
1733static inline void mark_inode_dirty_sync(struct inode *inode)
1734{
1735 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1736}
1737
7ada4db8
MS
1738extern void inc_nlink(struct inode *inode);
1739extern void drop_nlink(struct inode *inode);
1740extern void clear_nlink(struct inode *inode);
1741extern void set_nlink(struct inode *inode, unsigned int nlink);
d8c76e6f
DH
1742
1743static inline void inode_inc_link_count(struct inode *inode)
1744{
1745 inc_nlink(inode);
66d21730
AD
1746 mark_inode_dirty(inode);
1747}
1748
9a53c3a7
DH
1749static inline void inode_dec_link_count(struct inode *inode)
1750{
1751 drop_nlink(inode);
66d21730
AD
1752 mark_inode_dirty(inode);
1753}
1754
25ec56b5
JNC
1755/**
1756 * inode_inc_iversion - increments i_version
1757 * @inode: inode that need to be updated
1758 *
1759 * Every time the inode is modified, the i_version field will be incremented.
1760 * The filesystem has to be mounted with i_version flag
1761 */
1762
1763static inline void inode_inc_iversion(struct inode *inode)
1764{
1765 spin_lock(&inode->i_lock);
1766 inode->i_version++;
1767 spin_unlock(&inode->i_lock);
1768}
1769
c3b2da31
JB
1770enum file_time_flags {
1771 S_ATIME = 1,
1772 S_MTIME = 2,
1773 S_CTIME = 4,
1774 S_VERSION = 8,
1775};
1776
badcf2b7 1777extern void touch_atime(const struct path *);
1da177e4
LT
1778static inline void file_accessed(struct file *file)
1779{
1780 if (!(file->f_flags & O_NOATIME))
68ac1234 1781 touch_atime(&file->f_path);
1da177e4
LT
1782}
1783
1784int sync_inode(struct inode *inode, struct writeback_control *wbc);
c3765016 1785int sync_inode_metadata(struct inode *inode, int wait);
1da177e4 1786
1da177e4
LT
1787struct file_system_type {
1788 const char *name;
1789 int fs_flags;
bbc1096a
DH
1790#define FS_REQUIRES_DEV 1
1791#define FS_BINARY_MOUNTDATA 2
1792#define FS_HAS_SUBTYPE 4
0c55cfc4
EB
1793#define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
1794#define FS_USERNS_DEV_MOUNT 16 /* A userns mount does not imply MNT_NODEV */
bbc1096a 1795#define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
c96e41e9
AV
1796 struct dentry *(*mount) (struct file_system_type *, int,
1797 const char *, void *);
1da177e4
LT
1798 void (*kill_sb) (struct super_block *);
1799 struct module *owner;
1800 struct file_system_type * next;
a5166169 1801 struct hlist_head fs_supers;
d475fd42 1802
cf516249 1803 struct lock_class_key s_lock_key;
897c6ff9 1804 struct lock_class_key s_umount_key;
51ee049e 1805 struct lock_class_key s_vfs_rename_key;
5accdf82 1806 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
d475fd42
PZ
1807
1808 struct lock_class_key i_lock_key;
1809 struct lock_class_key i_mutex_key;
14358e6d 1810 struct lock_class_key i_mutex_dir_key;
1da177e4
LT
1811};
1812
7f78e035
EB
1813#define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
1814
ceefda69
AV
1815extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1816 void *data, int (*fill_super)(struct super_block *, void *, int));
152a0836
AV
1817extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1818 int flags, const char *dev_name, void *data,
1819 int (*fill_super)(struct super_block *, void *, int));
fc14f2fe
AV
1820extern struct dentry *mount_single(struct file_system_type *fs_type,
1821 int flags, void *data,
1822 int (*fill_super)(struct super_block *, void *, int));
3c26ff6e
AV
1823extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1824 int flags, void *data,
1825 int (*fill_super)(struct super_block *, void *, int));
ea441d11 1826extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
1da177e4
LT
1827void generic_shutdown_super(struct super_block *sb);
1828void kill_block_super(struct super_block *sb);
1829void kill_anon_super(struct super_block *sb);
1830void kill_litter_super(struct super_block *sb);
1831void deactivate_super(struct super_block *sb);
74dbbdd7 1832void deactivate_locked_super(struct super_block *sb);
1da177e4 1833int set_anon_super(struct super_block *s, void *data);
0ee5dc67
AV
1834int get_anon_bdev(dev_t *);
1835void free_anon_bdev(dev_t);
1da177e4
LT
1836struct super_block *sget(struct file_system_type *type,
1837 int (*test)(struct super_block *,void *),
1838 int (*set)(struct super_block *,void *),
9249e17f 1839 int flags, void *data);
51139ada 1840extern struct dentry *mount_pseudo(struct file_system_type *, char *,
c74a1cbb
AV
1841 const struct super_operations *ops,
1842 const struct dentry_operations *dops,
1843 unsigned long);
1da177e4
LT
1844
1845/* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1846#define fops_get(fops) \
1847 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1848#define fops_put(fops) \
1849 do { if (fops) module_put((fops)->owner); } while(0)
e84f9e57
AV
1850/*
1851 * This one is to be used *ONLY* from ->open() instances.
1852 * fops must be non-NULL, pinned down *and* module dependencies
1853 * should be sufficient to pin the caller down as well.
1854 */
1855#define replace_fops(f, fops) \
1856 do { \
1857 struct file *__file = (f); \
1858 fops_put(__file->f_op); \
1859 BUG_ON(!(__file->f_op = (fops))); \
1860 } while(0)
1da177e4
LT
1861
1862extern int register_filesystem(struct file_system_type *);
1863extern int unregister_filesystem(struct file_system_type *);
8bf9725c
PE
1864extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1865#define kern_mount(type) kern_mount_data(type, NULL)
423e0ab0 1866extern void kern_unmount(struct vfsmount *mnt);
1da177e4
LT
1867extern int may_umount_tree(struct vfsmount *);
1868extern int may_umount(struct vfsmount *);
5e6123f3
SL
1869extern long do_mount(const char *, const char __user *,
1870 const char *, unsigned long, void *);
589ff870 1871extern struct vfsmount *collect_mounts(struct path *);
8aec0809 1872extern void drop_collected_mounts(struct vfsmount *);
1f707137
AV
1873extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1874 struct vfsmount *);
ebabe9a9 1875extern int vfs_statfs(struct path *, struct kstatfs *);
c8b91acc
AV
1876extern int user_statfs(const char __user *, struct kstatfs *);
1877extern int fd_statfs(int, struct kstatfs *);
cf31e70d 1878extern int vfs_ustat(dev_t, struct kstatfs *);
18e9e510
JB
1879extern int freeze_super(struct super_block *super);
1880extern int thaw_super(struct super_block *super);
02125a82 1881extern bool our_mnt(struct vfsmount *mnt);
e51db735 1882extern bool fs_fully_visible(struct file_system_type *);
1da177e4 1883
ce3b0f8d
AV
1884extern int current_umask(void);
1885
5a14696c
BF
1886extern void ihold(struct inode * inode);
1887extern void iput(struct inode *);
1888
1fa97e8b 1889static inline struct inode *file_inode(const struct file *f)
d7a06983
JL
1890{
1891 return f->f_inode;
1892}
1893
f87fd4c2 1894/* /sys/fs */
00d26666 1895extern struct kobject *fs_kobj;
f87fd4c2 1896
435f49a5 1897#define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
bfcd17a6 1898
1da177e4
LT
1899#define FLOCK_VERIFY_READ 1
1900#define FLOCK_VERIFY_WRITE 2
1901
bfcd17a6 1902#ifdef CONFIG_FILE_LOCKING
d7a06983 1903extern int locks_mandatory_locked(struct file *);
1da177e4
LT
1904extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1905
1906/*
1907 * Candidates for mandatory locking have the setgid bit set
1908 * but no group execute bit - an otherwise meaningless combination.
1909 */
a16877ca
PE
1910
1911static inline int __mandatory_lock(struct inode *ino)
1912{
1913 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1914}
1915
1916/*
1917 * ... and these candidates should be on MS_MANDLOCK mounted fs,
1918 * otherwise these will be advisory locks
1919 */
1920
1921static inline int mandatory_lock(struct inode *ino)
1922{
1923 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1924}
1da177e4 1925
d7a06983 1926static inline int locks_verify_locked(struct file *file)
1da177e4 1927{
d7a06983
JL
1928 if (mandatory_lock(file_inode(file)))
1929 return locks_mandatory_locked(file);
1da177e4
LT
1930 return 0;
1931}
1932
1da177e4
LT
1933static inline int locks_verify_truncate(struct inode *inode,
1934 struct file *filp,
1935 loff_t size)
1936{
a16877ca 1937 if (inode->i_flock && mandatory_lock(inode))
1da177e4
LT
1938 return locks_mandatory_area(
1939 FLOCK_VERIFY_WRITE, inode, filp,
1940 size < inode->i_size ? size : inode->i_size,
1941 (size < inode->i_size ? inode->i_size - size
1942 : size - inode->i_size)
1943 );
1944 return 0;
1945}
1946
1947static inline int break_lease(struct inode *inode, unsigned int mode)
1948{
24cbe784
JL
1949 /*
1950 * Since this check is lockless, we must ensure that any refcounts
1951 * taken are done before checking inode->i_flock. Otherwise, we could
1952 * end up racing with tasks trying to set a new lease on this file.
1953 */
1954 smp_mb();
1da177e4 1955 if (inode->i_flock)
df4e8d2c 1956 return __break_lease(inode, mode, FL_LEASE);
1da177e4
LT
1957 return 0;
1958}
df4e8d2c
BF
1959
1960static inline int break_deleg(struct inode *inode, unsigned int mode)
1961{
962bd40b
JL
1962 /*
1963 * Since this check is lockless, we must ensure that any refcounts
1964 * taken are done before checking inode->i_flock. Otherwise, we could
1965 * end up racing with tasks trying to set a new lease on this file.
1966 */
1967 smp_mb();
df4e8d2c
BF
1968 if (inode->i_flock)
1969 return __break_lease(inode, mode, FL_DELEG);
1da177e4
LT
1970 return 0;
1971}
df4e8d2c 1972
5a14696c
BF
1973static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
1974{
1975 int ret;
1976
1977 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
1978 if (ret == -EWOULDBLOCK && delegated_inode) {
1979 *delegated_inode = inode;
1980 ihold(inode);
1981 }
1982 return ret;
1983}
1984
1985static inline int break_deleg_wait(struct inode **delegated_inode)
1986{
1987 int ret;
1988
1989 ret = break_deleg(*delegated_inode, O_WRONLY);
1990 iput(*delegated_inode);
1991 *delegated_inode = NULL;
1992 return ret;
1993}
1994
bfcd17a6 1995#else /* !CONFIG_FILE_LOCKING */
d7a06983 1996static inline int locks_mandatory_locked(struct file *file)
af5df566
SW
1997{
1998 return 0;
1999}
2000
2001static inline int locks_mandatory_area(int rw, struct inode *inode,
2002 struct file *filp, loff_t offset,
2003 size_t count)
2004{
2005 return 0;
2006}
2007
2008static inline int __mandatory_lock(struct inode *inode)
2009{
2010 return 0;
2011}
2012
2013static inline int mandatory_lock(struct inode *inode)
2014{
2015 return 0;
2016}
2017
d7a06983 2018static inline int locks_verify_locked(struct file *file)
af5df566
SW
2019{
2020 return 0;
2021}
2022
2023static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2024 size_t size)
2025{
2026 return 0;
2027}
2028
2029static inline int break_lease(struct inode *inode, unsigned int mode)
2030{
2031 return 0;
2032}
2033
df4e8d2c
BF
2034static inline int break_deleg(struct inode *inode, unsigned int mode)
2035{
2036 return 0;
2037}
5a14696c
BF
2038
2039static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2040{
2041 return 0;
2042}
2043
2044static inline int break_deleg_wait(struct inode **delegated_inode)
2045{
2046 BUG();
2047 return 0;
2048}
2049
bfcd17a6 2050#endif /* CONFIG_FILE_LOCKING */
1da177e4
LT
2051
2052/* fs/open.c */
adb5c247 2053struct audit_names;
91a27b2a 2054struct filename {
adb5c247
JL
2055 const char *name; /* pointer to actual string */
2056 const __user char *uptr; /* original userland pointer */
2057 struct audit_names *aname;
7950e385 2058 bool separate; /* should "name" be freed? */
91a27b2a 2059};
1da177e4 2060
a02de960 2061extern long vfs_truncate(struct path *, loff_t);
4a30131e
N
2062extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2063 struct file *filp);
3e63cbb1
AJ
2064extern int do_fallocate(struct file *file, int mode, loff_t offset,
2065 loff_t len);
8e8a1407 2066extern long do_sys_open(int dfd, const char __user *filename, int flags,
a218d0fd 2067 umode_t mode);
669abf4e 2068extern struct file *file_open_name(struct filename *, int, umode_t);
a218d0fd 2069extern struct file *filp_open(const char *, int, umode_t);
73d049a4
AV
2070extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2071 const char *, int);
4aa7c634 2072extern int vfs_open(const struct path *, struct file *, const struct cred *);
765927b2 2073extern struct file * dentry_open(const struct path *, int, const struct cred *);
1da177e4 2074extern int filp_close(struct file *, fl_owner_t id);
91a27b2a
JL
2075
2076extern struct filename *getname(const char __user *);
c4ad8f98 2077extern struct filename *getname_kernel(const char *);
91a27b2a 2078
47237687
AV
2079enum {
2080 FILE_CREATED = 1,
2081 FILE_OPENED = 2
2082};
30d90494
AV
2083extern int finish_open(struct file *file, struct dentry *dentry,
2084 int (*open)(struct inode *, struct file *),
2085 int *opened);
e45198a6 2086extern int finish_no_open(struct file *file, struct dentry *dentry);
1da177e4 2087
3e63cbb1
AJ
2088/* fs/ioctl.c */
2089
2090extern int ioctl_preallocate(struct file *filp, void __user *argp);
2091
1da177e4
LT
2092/* fs/dcache.c */
2093extern void __init vfs_caches_init_early(void);
2094extern void __init vfs_caches_init(unsigned long);
2095
b86c089b
CL
2096extern struct kmem_cache *names_cachep;
2097
91a27b2a
JL
2098extern void final_putname(struct filename *name);
2099
a608ca21 2100#define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
3446a8aa 2101#define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
1da177e4 2102#ifndef CONFIG_AUDITSYSCALL
91a27b2a 2103#define putname(name) final_putname(name)
1da177e4 2104#else
91a27b2a 2105extern void putname(struct filename *name);
1da177e4
LT
2106#endif
2107
9361401e 2108#ifdef CONFIG_BLOCK
1da177e4 2109extern int register_blkdev(unsigned int, const char *);
f4480240 2110extern void unregister_blkdev(unsigned int, const char *);
1da177e4 2111extern struct block_device *bdget(dev_t);
dddac6a7 2112extern struct block_device *bdgrab(struct block_device *bdev);
1da177e4
LT
2113extern void bd_set_size(struct block_device *, loff_t size);
2114extern void bd_forget(struct inode *inode);
2115extern void bdput(struct block_device *);
585d3bc0 2116extern void invalidate_bdev(struct block_device *);
5c0d6b60 2117extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
585d3bc0 2118extern int sync_blockdev(struct block_device *bdev);
ff01bb48 2119extern void kill_bdev(struct block_device *);
585d3bc0 2120extern struct super_block *freeze_bdev(struct block_device *);
c2d75438 2121extern void emergency_thaw_all(void);
585d3bc0
NP
2122extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2123extern int fsync_bdev(struct block_device *);
a8855990 2124extern int sb_is_blkdev_sb(struct super_block *sb);
9361401e
DH
2125#else
2126static inline void bd_forget(struct inode *inode) {}
47e4491b 2127static inline int sync_blockdev(struct block_device *bdev) { return 0; }
ff01bb48 2128static inline void kill_bdev(struct block_device *bdev) {}
47e4491b
AV
2129static inline void invalidate_bdev(struct block_device *bdev) {}
2130
2131static inline struct super_block *freeze_bdev(struct block_device *sb)
2132{
2133 return NULL;
2134}
2135
2136static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2137{
2138 return 0;
2139}
5c0d6b60
JK
2140
2141static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2142{
2143}
a8855990
JK
2144
2145static inline int sb_is_blkdev_sb(struct super_block *sb)
2146{
2147 return 0;
2148}
9361401e 2149#endif
60b0680f 2150extern int sync_filesystem(struct super_block *);
9361401e 2151extern const struct file_operations def_blk_fops;
4b6f5d20
AV
2152extern const struct file_operations def_chr_fops;
2153extern const struct file_operations bad_sock_fops;
9361401e 2154#ifdef CONFIG_BLOCK
1da177e4 2155extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
56b26add 2156extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
1da177e4 2157extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
e525fd89 2158extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
d4d77629
TH
2159extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2160 void *holder);
2161extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2162 void *holder);
4385bab1 2163extern void blkdev_put(struct block_device *bdev, fmode_t mode);
641dc636 2164#ifdef CONFIG_SYSFS
e09b457b 2165extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
49731baa
TH
2166extern void bd_unlink_disk_holder(struct block_device *bdev,
2167 struct gendisk *disk);
641dc636 2168#else
e09b457b
TH
2169static inline int bd_link_disk_holder(struct block_device *bdev,
2170 struct gendisk *disk)
2171{
2172 return 0;
2173}
49731baa
TH
2174static inline void bd_unlink_disk_holder(struct block_device *bdev,
2175 struct gendisk *disk)
2176{
2177}
641dc636 2178#endif
9361401e 2179#endif
1da177e4
LT
2180
2181/* fs/char_dev.c */
68eef3b4 2182#define CHRDEV_MAJOR_HASH_SIZE 255
1da177e4
LT
2183extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2184extern int register_chrdev_region(dev_t, unsigned, const char *);
1905b1bf
TH
2185extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2186 unsigned int count, const char *name,
2187 const struct file_operations *fops);
2188extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2189 unsigned int count, const char *name);
1da177e4 2190extern void unregister_chrdev_region(dev_t, unsigned);
68eef3b4 2191extern void chrdev_show(struct seq_file *,off_t);
1da177e4 2192
1905b1bf
TH
2193static inline int register_chrdev(unsigned int major, const char *name,
2194 const struct file_operations *fops)
2195{
2196 return __register_chrdev(major, 0, 256, name, fops);
2197}
2198
2199static inline void unregister_chrdev(unsigned int major, const char *name)
2200{
2201 __unregister_chrdev(major, 0, 256, name);
2202}
2203
1da177e4
LT
2204/* fs/block_dev.c */
2205#define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
1f014290 2206#define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
9361401e
DH
2207
2208#ifdef CONFIG_BLOCK
2209#define BLKDEV_MAJOR_HASH_SIZE 255
1da177e4
LT
2210extern const char *__bdevname(dev_t, char *buffer);
2211extern const char *bdevname(struct block_device *bdev, char *buffer);
2212extern struct block_device *lookup_bdev(const char *);
68eef3b4 2213extern void blkdev_show(struct seq_file *,off_t);
30c40d2c 2214
9361401e
DH
2215#else
2216#define BLKDEV_MAJOR_HASH_SIZE 0
2217#endif
1da177e4
LT
2218
2219extern void init_special_inode(struct inode *, umode_t, dev_t);
2220
2221/* Invalid inode operations -- fs/bad_inode.c */
2222extern void make_bad_inode(struct inode *);
2223extern int is_bad_inode(struct inode *);
2224
9361401e 2225#ifdef CONFIG_BLOCK
1da177e4
LT
2226/*
2227 * return READ, READA, or WRITE
2228 */
2229#define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2230
2231/*
2232 * return data direction, READ or WRITE
2233 */
2234#define bio_data_dir(bio) ((bio)->bi_rw & 1)
2235
c3279d14
AP
2236extern void check_disk_size_change(struct gendisk *disk,
2237 struct block_device *bdev);
0c002c2f 2238extern int revalidate_disk(struct gendisk *);
1da177e4 2239extern int check_disk_change(struct block_device *);
93b270f7 2240extern int __invalidate_device(struct block_device *, bool);
1da177e4 2241extern int invalidate_partition(struct gendisk *, int);
9361401e 2242#endif
1da177e4
LT
2243unsigned long invalidate_mapping_pages(struct address_space *mapping,
2244 pgoff_t start, pgoff_t end);
54bc4855 2245
1da177e4
LT
2246static inline void invalidate_remote_inode(struct inode *inode)
2247{
2248 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2249 S_ISLNK(inode->i_mode))
fc0ecff6 2250 invalidate_mapping_pages(inode->i_mapping, 0, -1);
1da177e4
LT
2251}
2252extern int invalidate_inode_pages2(struct address_space *mapping);
2253extern int invalidate_inode_pages2_range(struct address_space *mapping,
2254 pgoff_t start, pgoff_t end);
2255extern int write_inode_now(struct inode *, int);
2256extern int filemap_fdatawrite(struct address_space *);
2257extern int filemap_flush(struct address_space *);
2258extern int filemap_fdatawait(struct address_space *);
d3bccb6f
JK
2259extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2260 loff_t lend);
1da177e4
LT
2261extern int filemap_write_and_wait(struct address_space *mapping);
2262extern int filemap_write_and_wait_range(struct address_space *mapping,
2263 loff_t lstart, loff_t lend);
ebcf28e1
AM
2264extern int __filemap_fdatawrite_range(struct address_space *mapping,
2265 loff_t start, loff_t end, int sync_mode);
f4c0a0fd
JK
2266extern int filemap_fdatawrite_range(struct address_space *mapping,
2267 loff_t start, loff_t end);
ebcf28e1 2268
8018ab05
CH
2269extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2270 int datasync);
2271extern int vfs_fsync(struct file *file, int datasync);
d311d79d
AV
2272static inline int generic_write_sync(struct file *file, loff_t pos, loff_t count)
2273{
2274 if (!(file->f_flags & O_DSYNC) && !IS_SYNC(file->f_mapping->host))
2275 return 0;
2276 return vfs_fsync_range(file, pos, pos + count - 1,
2277 (file->f_flags & __O_SYNC) ? 0 : 1);
2278}
1da177e4
LT
2279extern void emergency_sync(void);
2280extern void emergency_remount(void);
9361401e 2281#ifdef CONFIG_BLOCK
1da177e4 2282extern sector_t bmap(struct inode *, sector_t);
9361401e 2283#endif
27ac0ffe 2284extern int notify_change(struct dentry *, struct iattr *, struct inode **);
f419a2e3 2285extern int inode_permission(struct inode *, int);
bd5d0856 2286extern int __inode_permission(struct inode *, int);
2830ba7f 2287extern int generic_permission(struct inode *, int);
cbdf35bc 2288extern int __check_sticky(struct inode *dir, struct inode *inode);
1da177e4 2289
f696a365
MS
2290static inline bool execute_ok(struct inode *inode)
2291{
2292 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2293}
2294
8d71db4f
AV
2295static inline void file_start_write(struct file *file)
2296{
2297 if (!S_ISREG(file_inode(file)->i_mode))
2298 return;
2299 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2300}
2301
5ae98f15
JK
2302static inline bool file_start_write_trylock(struct file *file)
2303{
2304 if (!S_ISREG(file_inode(file)->i_mode))
2305 return true;
2306 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2307}
2308
8d71db4f
AV
2309static inline void file_end_write(struct file *file)
2310{
2311 if (!S_ISREG(file_inode(file)->i_mode))
2312 return;
2313 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2314}
2315
07b8ce1e
AV
2316/*
2317 * get_write_access() gets write permission for a file.
2318 * put_write_access() releases this write permission.
2319 * This is used for regular files.
2320 * We cannot support write (and maybe mmap read-write shared) accesses and
2321 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2322 * can have the following values:
2323 * 0: no writers, no VM_DENYWRITE mappings
2324 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2325 * > 0: (i_writecount) users are writing to the file.
2326 *
2327 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2328 * except for the cases where we don't hold i_writecount yet. Then we need to
2329 * use {get,deny}_write_access() - these functions check the sign and refuse
2330 * to do the change if sign is wrong.
2331 */
2332static inline int get_write_access(struct inode *inode)
2333{
2334 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2335}
2336static inline int deny_write_access(struct file *file)
2337{
496ad9aa 2338 struct inode *inode = file_inode(file);
07b8ce1e
AV
2339 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2340}
1da177e4
LT
2341static inline void put_write_access(struct inode * inode)
2342{
2343 atomic_dec(&inode->i_writecount);
2344}
2345static inline void allow_write_access(struct file *file)
2346{
2347 if (file)
496ad9aa 2348 atomic_inc(&file_inode(file)->i_writecount);
1da177e4 2349}
f1fe29b4
DH
2350static inline bool inode_is_open_for_write(const struct inode *inode)
2351{
2352 return atomic_read(&inode->i_writecount) > 0;
2353}
2354
a5c96ebf
MZ
2355#ifdef CONFIG_IMA
2356static inline void i_readcount_dec(struct inode *inode)
2357{
2358 BUG_ON(!atomic_read(&inode->i_readcount));
2359 atomic_dec(&inode->i_readcount);
2360}
2361static inline void i_readcount_inc(struct inode *inode)
2362{
2363 atomic_inc(&inode->i_readcount);
2364}
2365#else
2366static inline void i_readcount_dec(struct inode *inode)
2367{
2368 return;
2369}
2370static inline void i_readcount_inc(struct inode *inode)
2371{
2372 return;
2373}
2374#endif
ed8cae8b 2375extern int do_pipe_flags(int *, int);
1da177e4 2376
6777d773 2377extern int kernel_read(struct file *, loff_t, char *, unsigned long);
7bb307e8 2378extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
ed44724b 2379extern ssize_t __kernel_write(struct file *, const char *, size_t, loff_t *);
1da177e4
LT
2380extern struct file * open_exec(const char *);
2381
2382/* fs/dcache.c -- generic fs support functions */
2383extern int is_subdir(struct dentry *, struct dentry *);
2096f759 2384extern int path_is_under(struct path *, struct path *);
1da177e4
LT
2385
2386#include <linux/err.h>
2387
2388/* needed for stackable file system support */
965c8e59 2389extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
1da177e4 2390
965c8e59 2391extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
1da177e4 2392
54e34621 2393extern int inode_init_always(struct super_block *, struct inode *);
1da177e4 2394extern void inode_init_once(struct inode *);
2aa15890 2395extern void address_space_init_once(struct address_space *mapping);
1da177e4
LT
2396extern struct inode * igrab(struct inode *);
2397extern ino_t iunique(struct super_block *, ino_t);
2398extern int inode_needs_sync(struct inode *inode);
45321ac5 2399extern int generic_delete_inode(struct inode *inode);
9bcb4b73
AV
2400static inline int generic_drop_inode(struct inode *inode)
2401{
2402 return !inode->i_nlink || inode_unhashed(inode);
2403}
1da177e4 2404
88bd5121
AA
2405extern struct inode *ilookup5_nowait(struct super_block *sb,
2406 unsigned long hashval, int (*test)(struct inode *, void *),
2407 void *data);
1da177e4
LT
2408extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2409 int (*test)(struct inode *, void *), void *data);
2410extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2411
2412extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2413extern struct inode * iget_locked(struct super_block *, unsigned long);
261bca86
AV
2414extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2415extern int insert_inode_locked(struct inode *);
e096d0c7
JB
2416#ifdef CONFIG_DEBUG_LOCK_ALLOC
2417extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2418#else
2419static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2420#endif
1da177e4 2421extern void unlock_new_inode(struct inode *);
85fe4025 2422extern unsigned int get_next_ino(void);
1da177e4 2423
1da177e4 2424extern void __iget(struct inode * inode);
b46980fe 2425extern void iget_failed(struct inode *);
dbd5768f 2426extern void clear_inode(struct inode *);
2e00c97e 2427extern void __destroy_inode(struct inode *);
a209dfc7
ED
2428extern struct inode *new_inode_pseudo(struct super_block *sb);
2429extern struct inode *new_inode(struct super_block *sb);
ff0c7d15 2430extern void free_inode_nonrcu(struct inode *inode);
01de85e0 2431extern int should_remove_suid(struct dentry *);
2f1936b8 2432extern int file_remove_suid(struct file *);
1da177e4
LT
2433
2434extern void __insert_inode_hash(struct inode *, unsigned long hashval);
646ec461
CH
2435static inline void insert_inode_hash(struct inode *inode)
2436{
1da177e4
LT
2437 __insert_inode_hash(inode, inode->i_ino);
2438}
f2ee7abf
ED
2439
2440extern void __remove_inode_hash(struct inode *);
2441static inline void remove_inode_hash(struct inode *inode)
2442{
2443 if (!inode_unhashed(inode))
2444 __remove_inode_hash(inode);
2445}
2446
646ec461 2447extern void inode_sb_list_add(struct inode *inode);
1da177e4 2448
9361401e 2449#ifdef CONFIG_BLOCK
1da177e4
LT
2450extern void submit_bio(int, struct bio *);
2451extern int bdev_read_only(struct block_device *);
9361401e 2452#endif
1da177e4
LT
2453extern int set_blocksize(struct block_device *, int);
2454extern int sb_set_blocksize(struct super_block *, int);
2455extern int sb_min_blocksize(struct super_block *, int);
2456
2457extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2458extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
0b173bc4
KK
2459extern int generic_file_remap_pages(struct vm_area_struct *, unsigned long addr,
2460 unsigned long size, pgoff_t pgoff);
1da177e4 2461int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
ed978a81 2462extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
8174202b 2463extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
8174202b 2464extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
0c949334 2465extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *, loff_t);
3b93f911 2466extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
1da177e4
LT
2467extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2468extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
293bc982
AV
2469extern ssize_t new_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2470extern ssize_t new_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
88e6faef 2471
eef99380 2472/* fs/block_dev.c */
b2de525f 2473extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
1456c0a8 2474extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
02c24a82
JB
2475extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2476 int datasync);
ff01bb48 2477extern void block_sync_page(struct page *page);
eef99380 2478
88e6faef 2479/* fs/splice.c */
cbb7e577 2480extern ssize_t generic_file_splice_read(struct file *, loff_t *,
88e6faef 2481 struct pipe_inode_info *, size_t, unsigned int);
6818173b
MS
2482extern ssize_t default_file_splice_read(struct file *, loff_t *,
2483 struct pipe_inode_info *, size_t, unsigned int);
8d020765 2484extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
cbb7e577 2485 struct file *, loff_t *, size_t, unsigned int);
88e6faef 2486extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
cbb7e577 2487 struct file *out, loff_t *, size_t len, unsigned int flags);
1c118596
MS
2488extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2489 loff_t *opos, size_t len, unsigned int flags);
2490
88e6faef 2491
1da177e4
LT
2492extern void
2493file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
965c8e59
AM
2494extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2495extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
46a1c2c7 2496extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
965c8e59 2497extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
5760495a 2498extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
965c8e59 2499 int whence, loff_t maxsize, loff_t eof);
1bf9d14d
AV
2500extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2501 int whence, loff_t size);
1da177e4
LT
2502extern int generic_file_open(struct inode * inode, struct file * filp);
2503extern int nonseekable_open(struct inode * inode, struct file * filp);
2504
ceffc078 2505#ifdef CONFIG_FS_XIP
eb6fe0c3
CO
2506extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2507 loff_t *ppos);
ceffc078 2508extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
eb6fe0c3
CO
2509extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2510 size_t len, loff_t *ppos);
ceffc078 2511extern int xip_truncate_page(struct address_space *mapping, loff_t from);
6d79125b
CO
2512#else
2513static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2514{
2515 return 0;
2516}
ceffc078
CO
2517#endif
2518
9361401e 2519#ifdef CONFIG_BLOCK
facd07b0
JB
2520typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2521 loff_t file_offset);
1da177e4
LT
2522
2523enum {
1e431f5c
CH
2524 /* need locking between buffered and direct access */
2525 DIO_LOCKING = 0x01,
2526
2527 /* filesystem does not support filling holes */
2528 DIO_SKIP_HOLES = 0x02,
60392573
CH
2529
2530 /* filesystem can handle aio writes beyond i_size */
2531 DIO_ASYNC_EXTEND = 0x04,
1da177e4
LT
2532};
2533
eafdc7d1
CH
2534void dio_end_io(struct bio *bio, int error);
2535
2536ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
31b14039
AV
2537 struct block_device *bdev, struct iov_iter *iter, loff_t offset,
2538 get_block_t get_block, dio_iodone_t end_io,
eafdc7d1 2539 dio_submit_t submit_io, int flags);
7bb46a67 2540
1da177e4 2541static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
31b14039
AV
2542 struct inode *inode, struct iov_iter *iter, loff_t offset,
2543 get_block_t get_block)
1da177e4 2544{
31b14039
AV
2545 return __blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iter,
2546 offset, get_block, NULL, NULL,
1e431f5c 2547 DIO_LOCKING | DIO_SKIP_HOLES);
1da177e4 2548}
9361401e 2549#endif
1da177e4 2550
1d59d61f
TM
2551void inode_dio_wait(struct inode *inode);
2552void inode_dio_done(struct inode *inode);
2553
5f16f322
TT
2554extern void inode_set_flags(struct inode *inode, unsigned int flags,
2555 unsigned int mask);
2556
4b6f5d20 2557extern const struct file_operations generic_ro_fops;
1da177e4
LT
2558
2559#define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2560
5d826c84 2561extern int readlink_copy(char __user *, int, const char *);
1da177e4 2562extern int page_readlink(struct dentry *, char __user *, int);
cc314eef
LT
2563extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2564extern void page_put_link(struct dentry *, struct nameidata *, void *);
0adb25d2 2565extern int __page_symlink(struct inode *inode, const char *symname, int len,
54566b2c 2566 int nofs);
1da177e4 2567extern int page_symlink(struct inode *inode, const char *symname, int len);
c5ef1c42 2568extern const struct inode_operations page_symlink_inode_operations;
87dc800b 2569extern void kfree_put_link(struct dentry *, struct nameidata *, void *);
1da177e4
LT
2570extern int generic_readlink(struct dentry *, char __user *, int);
2571extern void generic_fillattr(struct inode *, struct kstat *);
b7a6ec52 2572int vfs_getattr_nosec(struct path *path, struct kstat *stat);
3dadecce 2573extern int vfs_getattr(struct path *, struct kstat *);
b462707e 2574void __inode_add_bytes(struct inode *inode, loff_t bytes);
1da177e4 2575void inode_add_bytes(struct inode *inode, loff_t bytes);
1c8924eb 2576void __inode_sub_bytes(struct inode *inode, loff_t bytes);
1da177e4
LT
2577void inode_sub_bytes(struct inode *inode, loff_t bytes);
2578loff_t inode_get_bytes(struct inode *inode);
2579void inode_set_bytes(struct inode *inode, loff_t bytes);
2580
2581extern int vfs_readdir(struct file *, filldir_t, void *);
5c0ba4e0 2582extern int iterate_dir(struct file *, struct dir_context *);
1da177e4 2583
c7887325
DH
2584extern int vfs_stat(const char __user *, struct kstat *);
2585extern int vfs_lstat(const char __user *, struct kstat *);
1da177e4 2586extern int vfs_fstat(unsigned int, struct kstat *);
c7887325 2587extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
1da177e4 2588
deb21db7
EZ
2589extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2590 unsigned long arg);
e9079cce 2591extern int __generic_block_fiemap(struct inode *inode,
3a3076f4
JB
2592 struct fiemap_extent_info *fieinfo,
2593 loff_t start, loff_t len,
2594 get_block_t *get_block);
68c9d702
JB
2595extern int generic_block_fiemap(struct inode *inode,
2596 struct fiemap_extent_info *fieinfo, u64 start,
2597 u64 len, get_block_t *get_block);
1da177e4 2598
c18479fe
MS
2599extern void get_filesystem(struct file_system_type *fs);
2600extern void put_filesystem(struct file_system_type *fs);
1da177e4
LT
2601extern struct file_system_type *get_fs_type(const char *name);
2602extern struct super_block *get_super(struct block_device *);
6b6dc836 2603extern struct super_block *get_super_thawed(struct block_device *);
4504230a 2604extern struct super_block *get_active_super(struct block_device *bdev);
1da177e4 2605extern void drop_super(struct super_block *sb);
01a05b33 2606extern void iterate_supers(void (*)(struct super_block *, void *), void *);
43e15cdb
AV
2607extern void iterate_supers_type(struct file_system_type *,
2608 void (*)(struct super_block *, void *), void *);
1da177e4
LT
2609
2610extern int dcache_dir_open(struct inode *, struct file *);
2611extern int dcache_dir_close(struct inode *, struct file *);
2612extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
5f99f4e7 2613extern int dcache_readdir(struct file *, struct dir_context *);
7bb46a67 2614extern int simple_setattr(struct dentry *, struct iattr *);
1da177e4 2615extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
726c3342 2616extern int simple_statfs(struct dentry *, struct kstatfs *);
20955e89 2617extern int simple_open(struct inode *inode, struct file *file);
1da177e4
LT
2618extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2619extern int simple_unlink(struct inode *, struct dentry *);
2620extern int simple_rmdir(struct inode *, struct dentry *);
2621extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
02c24a82 2622extern int noop_fsync(struct file *, loff_t, loff_t, int);
1da177e4
LT
2623extern int simple_empty(struct dentry *);
2624extern int simple_readpage(struct file *file, struct page *page);
afddba49
NP
2625extern int simple_write_begin(struct file *file, struct address_space *mapping,
2626 loff_t pos, unsigned len, unsigned flags,
2627 struct page **pagep, void **fsdata);
2628extern int simple_write_end(struct file *file, struct address_space *mapping,
2629 loff_t pos, unsigned len, unsigned copied,
2630 struct page *page, void *fsdata);
b26d4cd3 2631extern int always_delete_dentry(const struct dentry *);
6987843f 2632extern struct inode *alloc_anon_inode(struct super_block *);
e6f5c789 2633extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
b26d4cd3 2634extern const struct dentry_operations simple_dentry_operations;
1da177e4 2635
00cd8dd3 2636extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
1da177e4 2637extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
4b6f5d20 2638extern const struct file_operations simple_dir_operations;
c5ef1c42 2639extern const struct inode_operations simple_dir_inode_operations;
4b6f5d20 2640struct tree_descr { char *name; const struct file_operations *ops; int mode; };
1da177e4 2641struct dentry *d_alloc_name(struct dentry *, const char *);
7d683a09 2642extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
1f5ce9e9 2643extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
1da177e4
LT
2644extern void simple_release_fs(struct vfsmount **mount, int *count);
2645
93b07113
AM
2646extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2647 loff_t *ppos, const void *from, size_t available);
6a727b43
JS
2648extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2649 const void __user *from, size_t count);
1da177e4 2650
ac13a829 2651extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
02c24a82 2652extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
d5aacad5 2653
30ca22c7
PL
2654extern int generic_check_addressable(unsigned, u64);
2655
e965f963 2656#ifdef CONFIG_MIGRATION
2d1db3b1 2657extern int buffer_migrate_page(struct address_space *,
a6bc32b8
MG
2658 struct page *, struct page *,
2659 enum migrate_mode);
e965f963
CL
2660#else
2661#define buffer_migrate_page NULL
2662#endif
2663
25d9e2d1 2664extern int inode_change_ok(const struct inode *, struct iattr *);
2665extern int inode_newsize_ok(const struct inode *, loff_t offset);
6a1a90ad 2666extern void setattr_copy(struct inode *inode, const struct iattr *attr);
1da177e4 2667
c3b2da31 2668extern int file_update_time(struct file *file);
1da177e4 2669
34c80b1d 2670extern int generic_show_options(struct seq_file *m, struct dentry *root);
b3b304a2 2671extern void save_mount_options(struct super_block *sb, char *options);
2a32cebd 2672extern void replace_mount_options(struct super_block *sb, char *options);
b3b304a2 2673
1da177e4
LT
2674static inline ino_t parent_ino(struct dentry *dentry)
2675{
2676 ino_t res;
2677
b5c84bf6
NP
2678 /*
2679 * Don't strictly need d_lock here? If the parent ino could change
2680 * then surely we'd have a deeper race in the caller?
2681 */
1da177e4
LT
2682 spin_lock(&dentry->d_lock);
2683 res = dentry->d_parent->d_inode->i_ino;
2684 spin_unlock(&dentry->d_lock);
2685 return res;
2686}
2687
1da177e4
LT
2688/* Transaction based IO helpers */
2689
2690/*
2691 * An argresp is stored in an allocated page and holds the
2692 * size of the argument or response, along with its content
2693 */
2694struct simple_transaction_argresp {
2695 ssize_t size;
2696 char data[0];
2697};
2698
2699#define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2700
2701char *simple_transaction_get(struct file *file, const char __user *buf,
2702 size_t size);
2703ssize_t simple_transaction_read(struct file *file, char __user *buf,
2704 size_t size, loff_t *pos);
2705int simple_transaction_release(struct inode *inode, struct file *file);
2706
76791ab2 2707void simple_transaction_set(struct file *file, size_t n);
1da177e4 2708
acaefc25
AB
2709/*
2710 * simple attribute files
2711 *
2712 * These attributes behave similar to those in sysfs:
2713 *
2714 * Writing to an attribute immediately sets a value, an open file can be
2715 * written to multiple times.
2716 *
2717 * Reading from an attribute creates a buffer from the value that might get
2718 * read with multiple read calls. When the attribute has been read
2719 * completely, no further read calls are possible until the file is opened
2720 * again.
2721 *
2722 * All attributes contain a text representation of a numeric value
2723 * that are accessed with the get() and set() functions.
2724 */
2725#define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2726static int __fops ## _open(struct inode *inode, struct file *file) \
2727{ \
2728 __simple_attr_check_format(__fmt, 0ull); \
2729 return simple_attr_open(inode, file, __get, __set, __fmt); \
2730} \
828c0950 2731static const struct file_operations __fops = { \
acaefc25
AB
2732 .owner = THIS_MODULE, \
2733 .open = __fops ## _open, \
74bedc4d 2734 .release = simple_attr_release, \
acaefc25
AB
2735 .read = simple_attr_read, \
2736 .write = simple_attr_write, \
1ec5584e 2737 .llseek = generic_file_llseek, \
68be3029 2738}
acaefc25 2739
b9075fa9
JP
2740static inline __printf(1, 2)
2741void __simple_attr_check_format(const char *fmt, ...)
acaefc25
AB
2742{
2743 /* don't do anything, just let the compiler check the arguments; */
2744}
2745
2746int simple_attr_open(struct inode *inode, struct file *file,
8b88b099 2747 int (*get)(void *, u64 *), int (*set)(void *, u64),
acaefc25 2748 const char *fmt);
74bedc4d 2749int simple_attr_release(struct inode *inode, struct file *file);
acaefc25
AB
2750ssize_t simple_attr_read(struct file *file, char __user *buf,
2751 size_t len, loff_t *ppos);
2752ssize_t simple_attr_write(struct file *file, const char __user *buf,
2753 size_t len, loff_t *ppos);
2754
4be28540 2755struct ctl_table;
8d65af78 2756int proc_nr_files(struct ctl_table *table, int write,
62239ac2 2757 void __user *buffer, size_t *lenp, loff_t *ppos);
312d3ca8
CH
2758int proc_nr_dentry(struct ctl_table *table, int write,
2759 void __user *buffer, size_t *lenp, loff_t *ppos);
cffbc8aa
DC
2760int proc_nr_inodes(struct ctl_table *table, int write,
2761 void __user *buffer, size_t *lenp, loff_t *ppos);
38e23c95 2762int __init get_filesystem_list(char *buf);
62239ac2 2763
3cd90ea4 2764#define __FMODE_EXEC ((__force int) FMODE_EXEC)
1a44bc8c
NK
2765#define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
2766
6d125529 2767#define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
ecf081d1 2768#define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
1a44bc8c 2769 (flag & __FMODE_NONOTIFY)))
5300990c 2770
8d334acd 2771static inline int is_sxid(umode_t mode)
69b45732
AK
2772{
2773 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
2774}
2775
cbdf35bc
MS
2776static inline int check_sticky(struct inode *dir, struct inode *inode)
2777{
2778 if (!(dir->i_mode & S_ISVTX))
2779 return 0;
2780
2781 return __check_sticky(dir, inode);
2782}
2783
69b45732
AK
2784static inline void inode_has_no_xattr(struct inode *inode)
2785{
9e1f1de0 2786 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
69b45732
AK
2787 inode->i_flags |= S_NOSEC;
2788}
2789
5f99f4e7
AV
2790static inline bool dir_emit(struct dir_context *ctx,
2791 const char *name, int namelen,
2792 u64 ino, unsigned type)
2793{
2794 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
2795}
2796static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
2797{
2798 return ctx->actor(ctx, ".", 1, ctx->pos,
2799 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
2800}
2801static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
2802{
2803 return ctx->actor(ctx, "..", 2, ctx->pos,
2804 parent_ino(file->f_path.dentry), DT_DIR) == 0;
2805}
2806static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
2807{
2808 if (ctx->pos == 0) {
2809 if (!dir_emit_dot(file, ctx))
2810 return false;
2811 ctx->pos = 1;
2812 }
2813 if (ctx->pos == 1) {
2814 if (!dir_emit_dotdot(file, ctx))
2815 return false;
2816 ctx->pos = 2;
2817 }
2818 return true;
2819}
5ded75ec
AV
2820static inline bool dir_relax(struct inode *inode)
2821{
2822 mutex_unlock(&inode->i_mutex);
2823 mutex_lock(&inode->i_mutex);
2824 return !IS_DEADDIR(inode);
2825}
5f99f4e7 2826
1da177e4 2827#endif /* _LINUX_FS_H */
This page took 1.403118 seconds and 5 git commands to generate.