locks: report l_pid as -1 for FL_FILE_PVT locks
[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;
1da177e4 51
74bf17cf 52extern void __init inode_init(void);
1da177e4 53extern void __init inode_init_early(void);
1da177e4
LT
54extern void __init files_init(unsigned long);
55
dded4f4d 56extern struct files_stat_struct files_stat;
518de9b3 57extern unsigned long get_max_files(void);
dded4f4d
JE
58extern int sysctl_nr_open;
59extern struct inodes_stat_t inodes_stat;
60extern int leases_enable, lease_break_time;
800179c9
KC
61extern int sysctl_protected_symlinks;
62extern int sysctl_protected_hardlinks;
dded4f4d 63
1da177e4
LT
64struct buffer_head;
65typedef int (get_block_t)(struct inode *inode, sector_t iblock,
66 struct buffer_head *bh_result, int create);
92198f7e 67typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
7b7a8665 68 ssize_t bytes, void *private);
1da177e4 69
bbc1096a
DH
70#define MAY_EXEC 0x00000001
71#define MAY_WRITE 0x00000002
72#define MAY_READ 0x00000004
73#define MAY_APPEND 0x00000008
74#define MAY_ACCESS 0x00000010
75#define MAY_OPEN 0x00000020
76#define MAY_CHDIR 0x00000040
77/* called from RCU mode, don't block */
78#define MAY_NOT_BLOCK 0x00000080
79
80/*
81 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
82 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
83 */
84
85/* file is open for reading */
86#define FMODE_READ ((__force fmode_t)0x1)
87/* file is open for writing */
88#define FMODE_WRITE ((__force fmode_t)0x2)
89/* file is seekable */
90#define FMODE_LSEEK ((__force fmode_t)0x4)
91/* file can be accessed using pread */
92#define FMODE_PREAD ((__force fmode_t)0x8)
93/* file can be accessed using pwrite */
94#define FMODE_PWRITE ((__force fmode_t)0x10)
95/* File is opened for execution with sys_execve / sys_uselib */
96#define FMODE_EXEC ((__force fmode_t)0x20)
97/* File is opened with O_NDELAY (only set for block devices) */
98#define FMODE_NDELAY ((__force fmode_t)0x40)
99/* File is opened with O_EXCL (only set for block devices) */
100#define FMODE_EXCL ((__force fmode_t)0x80)
101/* File is opened using open(.., 3, ..) and is writeable only for ioctls
102 (specialy hack for floppy.c) */
103#define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
104/* 32bit hashes as llseek() offset (for directories) */
105#define FMODE_32BITHASH ((__force fmode_t)0x200)
106/* 64bit hashes as llseek() offset (for directories) */
107#define FMODE_64BITHASH ((__force fmode_t)0x400)
108
109/*
110 * Don't update ctime and mtime.
111 *
112 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
113 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
114 */
115#define FMODE_NOCMTIME ((__force fmode_t)0x800)
116
117/* Expect random access pattern */
118#define FMODE_RANDOM ((__force fmode_t)0x1000)
119
120/* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
121#define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
122
123/* File is opened with O_PATH; almost nothing can be done with it */
124#define FMODE_PATH ((__force fmode_t)0x4000)
125
126/* File was opened by fanotify and shouldn't generate fanotify events */
127#define FMODE_NONOTIFY ((__force fmode_t)0x1000000)
128
129/*
130 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
131 * that indicates that they should check the contents of the iovec are
132 * valid, but not check the memory that the iovec elements
133 * points too.
134 */
135#define CHECK_IOVEC_ONLY -1
136
08cce05c
DH
137/*
138 * The below are the various read and write types that we support. Some of
139 * them include behavioral modifiers that send information down to the
140 * block layer and IO scheduler. Terminology:
141 *
142 * The block layer uses device plugging to defer IO a little bit, in
143 * the hope that we will see more IO very shortly. This increases
144 * coalescing of adjacent IO and thus reduces the number of IOs we
145 * have to send to the device. It also allows for better queuing,
146 * if the IO isn't mergeable. If the caller is going to be waiting
147 * for the IO, then he must ensure that the device is unplugged so
148 * that the IO is dispatched to the driver.
149 *
150 * All IO is handled async in Linux. This is fine for background
151 * writes, but for reads or writes that someone waits for completion
152 * on, we want to notify the block layer and IO scheduler so that they
153 * know about it. That allows them to make better scheduling
154 * decisions. So when the below references 'sync' and 'async', it
155 * is referencing this priority hint.
156 *
157 * With that in mind, the available types are:
158 *
159 * READ A normal read operation. Device will be plugged.
160 * READ_SYNC A synchronous read. Device is not plugged, caller can
161 * immediately wait on this read without caring about
162 * unplugging.
163 * READA Used for read-ahead operations. Lower priority, and the
164 * block layer could (in theory) choose to ignore this
165 * request if it runs into resource problems.
166 * WRITE A normal async write. Device will be plugged.
167 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
168 * the hint that someone will be waiting on this IO
169 * shortly. The write equivalent of READ_SYNC.
170 * WRITE_ODIRECT Special case write for O_DIRECT only.
171 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
172 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
173 * non-volatile media on completion.
174 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
175 * by a cache flush and data is guaranteed to be on
176 * non-volatile media on completion.
177 *
178 */
179#define RW_MASK REQ_WRITE
180#define RWA_MASK REQ_RAHEAD
181
182#define READ 0
183#define WRITE RW_MASK
184#define READA RWA_MASK
185#define KERNEL_READ (READ|REQ_KERNEL)
186#define KERNEL_WRITE (WRITE|REQ_KERNEL)
187
188#define READ_SYNC (READ | REQ_SYNC)
189#define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
190#define WRITE_ODIRECT (WRITE | REQ_SYNC)
191#define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
192#define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
193#define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
194
1da177e4
LT
195/*
196 * Attribute flags. These should be or-ed together to figure out what
197 * has been changed!
198 */
9767d749
MS
199#define ATTR_MODE (1 << 0)
200#define ATTR_UID (1 << 1)
201#define ATTR_GID (1 << 2)
202#define ATTR_SIZE (1 << 3)
203#define ATTR_ATIME (1 << 4)
204#define ATTR_MTIME (1 << 5)
205#define ATTR_CTIME (1 << 6)
206#define ATTR_ATIME_SET (1 << 7)
207#define ATTR_MTIME_SET (1 << 8)
208#define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
209#define ATTR_ATTR_FLAG (1 << 10)
210#define ATTR_KILL_SUID (1 << 11)
211#define ATTR_KILL_SGID (1 << 12)
212#define ATTR_FILE (1 << 13)
213#define ATTR_KILL_PRIV (1 << 14)
214#define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
215#define ATTR_TIMES_SET (1 << 16)
1da177e4
LT
216
217/*
218 * This is the Inode Attributes structure, used for notify_change(). It
219 * uses the above definitions as flags, to know which values have changed.
220 * Also, in this manner, a Filesystem can look at only the values it cares
221 * about. Basically, these are the attributes that the VFS layer can
222 * request to change from the FS layer.
223 *
224 * Derek Atkins <warlord@MIT.EDU> 94-10-20
225 */
226struct iattr {
227 unsigned int ia_valid;
228 umode_t ia_mode;
92361636
EB
229 kuid_t ia_uid;
230 kgid_t ia_gid;
1da177e4
LT
231 loff_t ia_size;
232 struct timespec ia_atime;
233 struct timespec ia_mtime;
234 struct timespec ia_ctime;
cc4e69de
MS
235
236 /*
25985edc 237 * Not an attribute, but an auxiliary info for filesystems wanting to
cc4e69de
MS
238 * implement an ftruncate() like method. NOTE: filesystem should
239 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
240 */
241 struct file *ia_file;
1da177e4
LT
242};
243
1da177e4
LT
244/*
245 * Includes for diskquotas.
246 */
247#include <linux/quota.h>
248
994fc28c
ZB
249/**
250 * enum positive_aop_returns - aop return codes with specific semantics
251 *
252 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
253 * completed, that the page is still locked, and
254 * should be considered active. The VM uses this hint
255 * to return the page to the active list -- it won't
256 * be a candidate for writeback again in the near
257 * future. Other callers must be careful to unlock
258 * the page if they get this return. Returned by
259 * writepage();
260 *
261 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
262 * unlocked it and the page might have been truncated.
263 * The caller should back up to acquiring a new page and
264 * trying again. The aop will be taking reasonable
265 * precautions not to livelock. If the caller held a page
266 * reference, it should drop it before retrying. Returned
55144768 267 * by readpage().
994fc28c
ZB
268 *
269 * address_space_operation functions return these large constants to indicate
270 * special semantics to the caller. These are much larger than the bytes in a
271 * page to allow for functions that return the number of bytes operated on in a
272 * given page.
273 */
274
275enum positive_aop_returns {
276 AOP_WRITEPAGE_ACTIVATE = 0x80000,
277 AOP_TRUNCATED_PAGE = 0x80001,
278};
279
afddba49 280#define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
89e10787 281#define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
54566b2c
NP
282#define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
283 * helper code (eg buffer layer)
284 * to clear GFP_FS from alloc */
afddba49 285
1da177e4
LT
286/*
287 * oh the beauties of C type declarations.
288 */
289struct page;
290struct address_space;
291struct writeback_control;
1da177e4 292
2f718ffc
NP
293struct iov_iter {
294 const struct iovec *iov;
295 unsigned long nr_segs;
296 size_t iov_offset;
297 size_t count;
298};
299
300size_t iov_iter_copy_from_user_atomic(struct page *page,
301 struct iov_iter *i, unsigned long offset, size_t bytes);
302size_t iov_iter_copy_from_user(struct page *page,
303 struct iov_iter *i, unsigned long offset, size_t bytes);
304void iov_iter_advance(struct iov_iter *i, size_t bytes);
afddba49 305int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes);
d28574e0 306size_t iov_iter_single_seg_count(const struct iov_iter *i);
2f718ffc
NP
307
308static inline void iov_iter_init(struct iov_iter *i,
309 const struct iovec *iov, unsigned long nr_segs,
310 size_t count, size_t written)
311{
312 i->iov = iov;
313 i->nr_segs = nr_segs;
314 i->iov_offset = 0;
315 i->count = count + written;
316
317 iov_iter_advance(i, written);
318}
319
320static inline size_t iov_iter_count(struct iov_iter *i)
321{
322 return i->count;
323}
324
8ab22b9a
HH
325/*
326 * "descriptor" for what we're up to with a read.
327 * This allows us to use the same read code yet
328 * have multiple different users of the data that
329 * we read from a file.
330 *
331 * The simplest case just copies the data to user
332 * mode.
333 */
334typedef struct {
335 size_t written;
336 size_t count;
337 union {
338 char __user *buf;
339 void *data;
340 } arg;
341 int error;
342} read_descriptor_t;
343
344typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
345 unsigned long, unsigned long);
2f718ffc 346
1da177e4
LT
347struct address_space_operations {
348 int (*writepage)(struct page *page, struct writeback_control *wbc);
349 int (*readpage)(struct file *, struct page *);
1da177e4
LT
350
351 /* Write back some dirty pages from this mapping. */
352 int (*writepages)(struct address_space *, struct writeback_control *);
353
4741c9fd 354 /* Set a page dirty. Return true if this dirtied it */
1da177e4
LT
355 int (*set_page_dirty)(struct page *page);
356
357 int (*readpages)(struct file *filp, struct address_space *mapping,
358 struct list_head *pages, unsigned nr_pages);
359
afddba49
NP
360 int (*write_begin)(struct file *, struct address_space *mapping,
361 loff_t pos, unsigned len, unsigned flags,
362 struct page **pagep, void **fsdata);
363 int (*write_end)(struct file *, struct address_space *mapping,
364 loff_t pos, unsigned len, unsigned copied,
365 struct page *page, void *fsdata);
366
1da177e4
LT
367 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
368 sector_t (*bmap)(struct address_space *, sector_t);
d47992f8 369 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
27496a8c 370 int (*releasepage) (struct page *, gfp_t);
6072d13c 371 void (*freepage)(struct page *);
1da177e4
LT
372 ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
373 loff_t offset, unsigned long nr_segs);
70688e4d
NP
374 int (*get_xip_mem)(struct address_space *, pgoff_t, int,
375 void **, unsigned long *);
b969c4ab 376 /*
ef277c73
TC
377 * migrate the contents of a page to the specified target. If
378 * migrate_mode is MIGRATE_ASYNC, it must not block.
b969c4ab 379 */
2d1db3b1 380 int (*migratepage) (struct address_space *,
a6bc32b8 381 struct page *, struct page *, enum migrate_mode);
e3db7691 382 int (*launder_page) (struct page *);
8ab22b9a
HH
383 int (*is_partially_uptodate) (struct page *, read_descriptor_t *,
384 unsigned long);
b4597226 385 void (*is_dirty_writeback) (struct page *, bool *, bool *);
25718736 386 int (*error_remove_page)(struct address_space *, struct page *);
62c230bc
MG
387
388 /* swapfile support */
a509bc1a
MG
389 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
390 sector_t *span);
391 void (*swap_deactivate)(struct file *file);
1da177e4
LT
392};
393
7dcda1c9
JA
394extern const struct address_space_operations empty_aops;
395
afddba49
NP
396/*
397 * pagecache_write_begin/pagecache_write_end must be used by general code
398 * to write into the pagecache.
399 */
400int pagecache_write_begin(struct file *, struct address_space *mapping,
401 loff_t pos, unsigned len, unsigned flags,
402 struct page **pagep, void **fsdata);
403
404int pagecache_write_end(struct file *, struct address_space *mapping,
405 loff_t pos, unsigned len, unsigned copied,
406 struct page *page, void *fsdata);
407
1da177e4
LT
408struct backing_dev_info;
409struct address_space {
410 struct inode *host; /* owner: inode, block_device */
411 struct radix_tree_root page_tree; /* radix tree of all pages */
19fd6231 412 spinlock_t tree_lock; /* and lock protecting it */
1da177e4 413 unsigned int i_mmap_writable;/* count VM_SHARED mappings */
6b2dbba8 414 struct rb_root i_mmap; /* tree of private and shared mappings */
1da177e4 415 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
3d48ae45 416 struct mutex i_mmap_mutex; /* protect tree, count, list */
08142579 417 /* Protected by tree_lock together with the radix tree */
1da177e4
LT
418 unsigned long nrpages; /* number of total pages */
419 pgoff_t writeback_index;/* writeback starts here */
f5e54d6e 420 const struct address_space_operations *a_ops; /* methods */
1da177e4
LT
421 unsigned long flags; /* error bits/gfp mask */
422 struct backing_dev_info *backing_dev_info; /* device readahead, etc */
423 spinlock_t private_lock; /* for use by the address_space */
424 struct list_head private_list; /* ditto */
252aa6f5 425 void *private_data; /* ditto */
1da177e4
LT
426} __attribute__((aligned(sizeof(long))));
427 /*
428 * On most architectures that alignment is already the case; but
25985edc 429 * must be enforced here for CRIS, to let the least significant bit
1da177e4
LT
430 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
431 */
87192a2a 432struct request_queue;
1da177e4
LT
433
434struct block_device {
435 dev_t bd_dev; /* not a kdev_t - it's a search key */
f3ccfcda 436 int bd_openers;
1da177e4 437 struct inode * bd_inode; /* will die */
87d8fe1e 438 struct super_block * bd_super;
c039e313 439 struct mutex bd_mutex; /* open/close mutex */
1da177e4 440 struct list_head bd_inodes;
6b4517a7 441 void * bd_claiming;
1da177e4
LT
442 void * bd_holder;
443 int bd_holders;
77ea887e 444 bool bd_write_holder;
641dc636 445#ifdef CONFIG_SYSFS
49731baa 446 struct list_head bd_holder_disks;
641dc636 447#endif
1da177e4
LT
448 struct block_device * bd_contains;
449 unsigned bd_block_size;
450 struct hd_struct * bd_part;
451 /* number of times partitions within this device have been opened. */
452 unsigned bd_part_count;
453 int bd_invalidated;
454 struct gendisk * bd_disk;
87192a2a 455 struct request_queue * bd_queue;
1da177e4 456 struct list_head bd_list;
1da177e4
LT
457 /*
458 * Private data. You must have bd_claim'ed the block_device
459 * to use this. NOTE: bd_claim allows an owner to claim
460 * the same device multiple times, the owner must take special
461 * care to not mess up bd_private for that case.
462 */
463 unsigned long bd_private;
fcccf502
TS
464
465 /* The counter of freeze processes */
466 int bd_fsfreeze_count;
467 /* Mutex for freeze */
468 struct mutex bd_fsfreeze_mutex;
1da177e4
LT
469};
470
471/*
472 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
473 * radix trees
474 */
475#define PAGECACHE_TAG_DIRTY 0
476#define PAGECACHE_TAG_WRITEBACK 1
f446daae 477#define PAGECACHE_TAG_TOWRITE 2
1da177e4
LT
478
479int mapping_tagged(struct address_space *mapping, int tag);
480
481/*
482 * Might pages of this file be mapped into userspace?
483 */
484static inline int mapping_mapped(struct address_space *mapping)
485{
6b2dbba8 486 return !RB_EMPTY_ROOT(&mapping->i_mmap) ||
1da177e4
LT
487 !list_empty(&mapping->i_mmap_nonlinear);
488}
489
490/*
491 * Might pages of this file have been modified in userspace?
492 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
493 * marks vma as VM_SHARED if it is shared, and the file was opened for
494 * writing i.e. vma may be mprotected writable even if now readonly.
495 */
496static inline int mapping_writably_mapped(struct address_space *mapping)
497{
498 return mapping->i_mmap_writable != 0;
499}
500
501/*
502 * Use sequence counter to get consistent i_size on 32-bit processors.
503 */
504#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
505#include <linux/seqlock.h>
506#define __NEED_I_SIZE_ORDERED
507#define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
508#else
509#define i_size_ordered_init(inode) do { } while (0)
510#endif
511
f19d4a8f
AV
512struct posix_acl;
513#define ACL_NOT_CACHED ((void *)(-1))
514
3ddcd056
LT
515#define IOP_FASTPERM 0x0001
516#define IOP_LOOKUP 0x0002
517#define IOP_NOFOLLOW 0x0004
518
519/*
520 * Keep mostly read-only and often accessed (especially for
521 * the RCU path lookup and 'stat' data) fields at the beginning
522 * of the 'struct inode'
523 */
1da177e4 524struct inode {
44a7d7a8 525 umode_t i_mode;
3ddcd056 526 unsigned short i_opflags;
92361636
EB
527 kuid_t i_uid;
528 kgid_t i_gid;
3ddcd056
LT
529 unsigned int i_flags;
530
531#ifdef CONFIG_FS_POSIX_ACL
532 struct posix_acl *i_acl;
533 struct posix_acl *i_default_acl;
534#endif
535
44a7d7a8
NP
536 const struct inode_operations *i_op;
537 struct super_block *i_sb;
3ddcd056 538 struct address_space *i_mapping;
44a7d7a8 539
13e12d14
LT
540#ifdef CONFIG_SECURITY
541 void *i_security;
542#endif
44a7d7a8 543
3ddcd056
LT
544 /* Stat data, not accessed from path walking */
545 unsigned long i_ino;
a78ef704
MS
546 /*
547 * Filesystems may only read i_nlink directly. They shall use the
548 * following functions for modification:
549 *
550 * (set|clear|inc|drop)_nlink
551 * inode_(inc|dec)_link_count
552 */
553 union {
554 const unsigned int i_nlink;
555 unsigned int __i_nlink;
556 };
3ddcd056 557 dev_t i_rdev;
2f9d3df8 558 loff_t i_size;
3ddcd056
LT
559 struct timespec i_atime;
560 struct timespec i_mtime;
561 struct timespec i_ctime;
6cdbb0ef
TT
562 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
563 unsigned short i_bytes;
2f9d3df8 564 unsigned int i_blkbits;
3ddcd056
LT
565 blkcnt_t i_blocks;
566
567#ifdef __NEED_I_SIZE_ORDERED
568 seqcount_t i_size_seqcount;
569#endif
570
571 /* Misc */
572 unsigned long i_state;
3ddcd056 573 struct mutex i_mutex;
13e12d14 574
44a7d7a8
NP
575 unsigned long dirtied_when; /* jiffies of first dirtying */
576
1da177e4 577 struct hlist_node i_hash;
7ccf19a8
NP
578 struct list_head i_wb_list; /* backing dev IO list */
579 struct list_head i_lru; /* inode LRU list */
1da177e4 580 struct list_head i_sb_list;
fa0d7e3d 581 union {
b3d9b7a3 582 struct hlist_head i_dentry;
fa0d7e3d
NP
583 struct rcu_head i_rcu;
584 };
7a224228 585 u64 i_version;
2f9d3df8 586 atomic_t i_count;
bd5fe6c5 587 atomic_t i_dio_count;
6cdbb0ef 588 atomic_t i_writecount;
99ac48f5 589 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
1da177e4 590 struct file_lock *i_flock;
1da177e4
LT
591 struct address_space i_data;
592#ifdef CONFIG_QUOTA
593 struct dquot *i_dquot[MAXQUOTAS];
594#endif
1da177e4 595 struct list_head i_devices;
4c154168
TT
596 union {
597 struct pipe_inode_info *i_pipe;
eaf796e7 598 struct block_device *i_bdev;
577c4eb0 599 struct cdev *i_cdev;
4c154168 600 };
1da177e4
LT
601
602 __u32 i_generation;
603
3be25f49
EP
604#ifdef CONFIG_FSNOTIFY
605 __u32 i_fsnotify_mask; /* all events this inode cares about */
e61ce867 606 struct hlist_head i_fsnotify_marks;
0eeca283
RL
607#endif
608
a178d202 609#ifdef CONFIG_IMA
a68a27b6 610 atomic_t i_readcount; /* struct files open RO */
a178d202 611#endif
8e18e294 612 void *i_private; /* fs or device private pointer */
1da177e4
LT
613};
614
1d3382cb
AV
615static inline int inode_unhashed(struct inode *inode)
616{
617 return hlist_unhashed(&inode->i_hash);
618}
619
f2eace23
IM
620/*
621 * inode->i_mutex nesting subclasses for the lock validator:
622 *
623 * 0: the object of the current VFS operation
624 * 1: parent
625 * 2: child/target
40bd22c9
BF
626 * 3: xattr
627 * 4: second non-directory
628 * The last is for certain operations (such as rename) which lock two
629 * non-directories at once.
f2eace23
IM
630 *
631 * The locking order between these classes is
40bd22c9 632 * parent -> child -> normal -> xattr -> second non-directory
f2eace23
IM
633 */
634enum inode_i_mutex_lock_class
635{
636 I_MUTEX_NORMAL,
637 I_MUTEX_PARENT,
638 I_MUTEX_CHILD,
4df46240 639 I_MUTEX_XATTR,
40bd22c9 640 I_MUTEX_NONDIR2
f2eace23
IM
641};
642
375e289e
BF
643void lock_two_nondirectories(struct inode *, struct inode*);
644void unlock_two_nondirectories(struct inode *, struct inode*);
645
1da177e4
LT
646/*
647 * NOTE: in a 32bit arch with a preemptable kernel and
648 * an UP compile the i_size_read/write must be atomic
649 * with respect to the local cpu (unlike with preempt disabled),
650 * but they don't need to be atomic with respect to other cpus like in
651 * true SMP (so they need either to either locally disable irq around
652 * the read or for example on x86 they can be still implemented as a
653 * cmpxchg8b without the need of the lock prefix). For SMP compiles
654 * and 64bit archs it makes no difference if preempt is enabled or not.
655 */
48ed214d 656static inline loff_t i_size_read(const struct inode *inode)
1da177e4
LT
657{
658#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
659 loff_t i_size;
660 unsigned int seq;
661
662 do {
663 seq = read_seqcount_begin(&inode->i_size_seqcount);
664 i_size = inode->i_size;
665 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
666 return i_size;
667#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
668 loff_t i_size;
669
670 preempt_disable();
671 i_size = inode->i_size;
672 preempt_enable();
673 return i_size;
674#else
675 return inode->i_size;
676#endif
677}
678
7762f5a0
MS
679/*
680 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
681 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
682 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
683 */
1da177e4
LT
684static inline void i_size_write(struct inode *inode, loff_t i_size)
685{
686#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
74e3d1e1 687 preempt_disable();
1da177e4
LT
688 write_seqcount_begin(&inode->i_size_seqcount);
689 inode->i_size = i_size;
690 write_seqcount_end(&inode->i_size_seqcount);
74e3d1e1 691 preempt_enable();
1da177e4
LT
692#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
693 preempt_disable();
694 inode->i_size = i_size;
695 preempt_enable();
696#else
697 inode->i_size = i_size;
698#endif
699}
700
92361636
EB
701/* Helper functions so that in most cases filesystems will
702 * not need to deal directly with kuid_t and kgid_t and can
703 * instead deal with the raw numeric values that are stored
704 * in the filesystem.
705 */
706static inline uid_t i_uid_read(const struct inode *inode)
707{
708 return from_kuid(&init_user_ns, inode->i_uid);
709}
710
711static inline gid_t i_gid_read(const struct inode *inode)
712{
713 return from_kgid(&init_user_ns, inode->i_gid);
714}
715
716static inline void i_uid_write(struct inode *inode, uid_t uid)
717{
718 inode->i_uid = make_kuid(&init_user_ns, uid);
719}
720
721static inline void i_gid_write(struct inode *inode, gid_t gid)
722{
723 inode->i_gid = make_kgid(&init_user_ns, gid);
724}
725
48ed214d 726static inline unsigned iminor(const struct inode *inode)
1da177e4
LT
727{
728 return MINOR(inode->i_rdev);
729}
730
48ed214d 731static inline unsigned imajor(const struct inode *inode)
1da177e4
LT
732{
733 return MAJOR(inode->i_rdev);
734}
735
736extern struct block_device *I_BDEV(struct inode *inode);
737
738struct fown_struct {
739 rwlock_t lock; /* protects pid, uid, euid fields */
609d7fa9
EB
740 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
741 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
92361636 742 kuid_t uid, euid; /* uid/euid of process setting the owner */
1da177e4
LT
743 int signum; /* posix.1b rt signal to be delivered on IO */
744};
745
746/*
747 * Track a single file's readahead state
748 */
749struct file_ra_state {
937085aa
FW
750 pgoff_t start; /* where readahead started */
751 unsigned int size; /* # of readahead pages */
752 unsigned int async_size; /* do asynchronous readahead when
f9acc8c7 753 there are only # of pages ahead */
5ce1110b 754
937085aa 755 unsigned int ra_pages; /* Maximum readahead window */
1ebf26a9 756 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
f4e6b498 757 loff_t prev_pos; /* Cache last read() position */
1da177e4 758};
1da177e4 759
5ce1110b
FW
760/*
761 * Check if @index falls in the readahead windows.
762 */
763static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
764{
f9acc8c7
FW
765 return (index >= ra->start &&
766 index < ra->start + ra->size);
5ce1110b
FW
767}
768
ad775f5a
DH
769#define FILE_MNT_WRITE_TAKEN 1
770#define FILE_MNT_WRITE_RELEASED 2
771
1da177e4 772struct file {
2f512016 773 union {
4f5e65a1 774 struct llist_node fu_llist;
2f512016
ED
775 struct rcu_head fu_rcuhead;
776 } f_u;
0f7fc9e4
JJS
777 struct path f_path;
778#define f_dentry f_path.dentry
dd37978c 779 struct inode *f_inode; /* cached value */
99ac48f5 780 const struct file_operations *f_op;
ef3d0fd2
AK
781
782 /*
783 * Protects f_ep_links, f_flags, f_pos vs i_size in lseek SEEK_CUR.
784 * Must not be taken from IRQ context.
785 */
786 spinlock_t f_lock;
516e0cc5 787 atomic_long_t f_count;
1da177e4 788 unsigned int f_flags;
aeb5d727 789 fmode_t f_mode;
1da177e4
LT
790 loff_t f_pos;
791 struct fown_struct f_owner;
d76b0d9b 792 const struct cred *f_cred;
1da177e4
LT
793 struct file_ra_state f_ra;
794
2b47c361 795 u64 f_version;
50462062 796#ifdef CONFIG_SECURITY
1da177e4 797 void *f_security;
50462062 798#endif
1da177e4
LT
799 /* needed for tty driver, and maybe others */
800 void *private_data;
801
802#ifdef CONFIG_EPOLL
803 /* Used by fs/eventpoll.c to link all the hooks to this file */
804 struct list_head f_ep_links;
28d82dc1 805 struct list_head f_tfile_llink;
1da177e4
LT
806#endif /* #ifdef CONFIG_EPOLL */
807 struct address_space *f_mapping;
ad775f5a
DH
808#ifdef CONFIG_DEBUG_WRITECOUNT
809 unsigned long f_mnt_write_state;
810#endif
1da177e4 811};
1da177e4 812
990d6c2d
AK
813struct file_handle {
814 __u32 handle_bytes;
815 int handle_type;
816 /* file identifier */
817 unsigned char f_handle[0];
818};
819
cb0942b8
AV
820static inline struct file *get_file(struct file *f)
821{
822 atomic_long_inc(&f->f_count);
823 return f;
824}
d7065da0 825#define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
516e0cc5 826#define file_count(x) atomic_long_read(&(x)->f_count)
1da177e4 827
ad775f5a
DH
828#ifdef CONFIG_DEBUG_WRITECOUNT
829static inline void file_take_write(struct file *f)
830{
831 WARN_ON(f->f_mnt_write_state != 0);
832 f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN;
833}
834static inline void file_release_write(struct file *f)
835{
836 f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED;
837}
838static inline void file_reset_write(struct file *f)
839{
840 f->f_mnt_write_state = 0;
841}
842static inline void file_check_state(struct file *f)
843{
844 /*
845 * At this point, either both or neither of these bits
846 * should be set.
847 */
848 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN);
849 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED);
850}
851static inline int file_check_writeable(struct file *f)
852{
853 if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN)
854 return 0;
855 printk(KERN_WARNING "writeable file with no "
856 "mnt_want_write()\n");
857 WARN_ON(1);
858 return -EINVAL;
859}
860#else /* !CONFIG_DEBUG_WRITECOUNT */
861static inline void file_take_write(struct file *filp) {}
862static inline void file_release_write(struct file *filp) {}
863static inline void file_reset_write(struct file *filp) {}
864static inline void file_check_state(struct file *filp) {}
865static inline int file_check_writeable(struct file *filp)
866{
867 return 0;
868}
869#endif /* CONFIG_DEBUG_WRITECOUNT */
870
1da177e4
LT
871#define MAX_NON_LFS ((1UL<<31) - 1)
872
873/* Page cache limit. The filesystems should put that into their s_maxbytes
874 limits, otherwise bad things can happen in VM. */
875#if BITS_PER_LONG==32
2bd2c194 876#define MAX_LFS_FILESIZE (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
1da177e4 877#elif BITS_PER_LONG==64
614c321f 878#define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
1da177e4
LT
879#endif
880
881#define FL_POSIX 1
882#define FL_FLOCK 2
617588d5 883#define FL_DELEG 4 /* NFSv4 delegation */
1da177e4 884#define FL_ACCESS 8 /* not trying to lock, just looking */
f475ae95 885#define FL_EXISTS 16 /* when unlocking, test for existence */
1da177e4 886#define FL_LEASE 32 /* lease held on this file */
75e1fcc0 887#define FL_CLOSE 64 /* unlock on close */
1da177e4 888#define FL_SLEEP 128 /* A blocking lock */
778fc546
BF
889#define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
890#define FL_UNLOCK_PENDING 512 /* Lease is being broken */
c918d42a 891#define FL_FILE_PVT 1024 /* lock is private to the file */
1da177e4 892
bde74e4b
MS
893/*
894 * Special return value from posix_lock_file() and vfs_lock_file() for
895 * asynchronous locking.
896 */
897#define FILE_LOCK_DEFERRED 1
898
1da177e4
LT
899/*
900 * The POSIX file lock owner is determined by
901 * the "struct files_struct" in the thread group
902 * (or NULL for no owner - BSD locks).
903 *
904 * Lockd stuffs a "host" pointer into this.
905 */
906typedef struct files_struct *fl_owner_t;
907
908struct file_lock_operations {
1da177e4
LT
909 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
910 void (*fl_release_private)(struct file_lock *);
911};
912
913struct lock_manager_operations {
8fb47a4f 914 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
3999e493 915 unsigned long (*lm_owner_key)(struct file_lock *);
8fb47a4f
BF
916 void (*lm_notify)(struct file_lock *); /* unblock callback */
917 int (*lm_grant)(struct file_lock *, struct file_lock *, int);
8fb47a4f
BF
918 void (*lm_break)(struct file_lock *);
919 int (*lm_change)(struct file_lock **, int);
1da177e4
LT
920};
921
af558e33
BF
922struct lock_manager {
923 struct list_head list;
924};
925
5ccb0066
SK
926struct net;
927void locks_start_grace(struct net *, struct lock_manager *);
af558e33 928void locks_end_grace(struct lock_manager *);
5ccb0066 929int locks_in_grace(struct net *);
af558e33 930
1da177e4
LT
931/* that will die - we need it for nfs_lock_info */
932#include <linux/nfs_fs_i.h>
933
1cb36012
JL
934/*
935 * struct file_lock represents a generic "file lock". It's used to represent
936 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
937 * note that the same struct is used to represent both a request for a lock and
938 * the lock itself, but the same object is never used for both.
939 *
940 * FIXME: should we create a separate "struct lock_request" to help distinguish
941 * these two uses?
942 *
943 * The i_flock list is ordered by:
944 *
945 * 1) lock type -- FL_LEASEs first, then FL_FLOCK, and finally FL_POSIX
946 * 2) lock owner
947 * 3) lock range start
948 * 4) lock range end
949 *
950 * Obviously, the last two criteria only matter for POSIX locks.
951 */
1da177e4
LT
952struct file_lock {
953 struct file_lock *fl_next; /* singly linked list for this inode */
139ca04e 954 struct hlist_node fl_link; /* node in global lists */
1da177e4
LT
955 struct list_head fl_block; /* circular list of blocked processes */
956 fl_owner_t fl_owner;
710b7216 957 unsigned int fl_flags;
afc1246f 958 unsigned char fl_type;
1da177e4 959 unsigned int fl_pid;
7012b02a 960 int fl_link_cpu; /* what cpu's list is this on? */
ab1f1611 961 struct pid *fl_nspid;
1da177e4
LT
962 wait_queue_head_t fl_wait;
963 struct file *fl_file;
1da177e4
LT
964 loff_t fl_start;
965 loff_t fl_end;
966
967 struct fasync_struct * fl_fasync; /* for lease break notifications */
778fc546
BF
968 /* for lease breaks: */
969 unsigned long fl_break_time;
970 unsigned long fl_downgrade_time;
1da177e4 971
6aed6285 972 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
7b021967 973 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1da177e4
LT
974 union {
975 struct nfs_lock_info nfs_fl;
8d0a8a9d 976 struct nfs4_lock_info nfs4_fl;
e8d6c554
DH
977 struct {
978 struct list_head link; /* link in AFS vnode's pending_locks list */
979 int state; /* state of grant or error if -ve */
980 } afs;
1da177e4
LT
981 } fl_u;
982};
983
984/* The following constant reflects the upper bound of the file/locking space */
985#ifndef OFFSET_MAX
986#define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
987#define OFFSET_MAX INT_LIMIT(loff_t)
988#define OFFT_OFFSET_MAX INT_LIMIT(off_t)
989#endif
990
1da177e4
LT
991#include <linux/fcntl.h>
992
bfcd17a6
TP
993extern void send_sigio(struct fown_struct *fown, int fd, int band);
994
bfcd17a6 995#ifdef CONFIG_FILE_LOCKING
1da177e4 996extern int fcntl_getlk(struct file *, struct flock __user *);
c293621b
PS
997extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
998 struct flock __user *);
1da177e4
LT
999
1000#if BITS_PER_LONG == 32
1001extern int fcntl_getlk64(struct file *, struct flock64 __user *);
c293621b
PS
1002extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1003 struct flock64 __user *);
1da177e4
LT
1004#endif
1005
1da177e4
LT
1006extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1007extern int fcntl_getlease(struct file *filp);
1008
1009/* fs/locks.c */
05fa3135 1010void locks_free_lock(struct file_lock *fl);
1da177e4 1011extern void locks_init_lock(struct file_lock *);
c5b1f0d9 1012extern struct file_lock * locks_alloc_lock(void);
1da177e4 1013extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1a747ee0 1014extern void __locks_copy_lock(struct file_lock *, const struct file_lock *);
1da177e4 1015extern void locks_remove_posix(struct file *, fl_owner_t);
78ed8a13 1016extern void locks_remove_file(struct file *);
a9e61e25 1017extern void locks_release_private(struct file_lock *);
6d34ac19 1018extern void posix_test_lock(struct file *, struct file_lock *);
150b3934 1019extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1da177e4 1020extern int posix_lock_file_wait(struct file *, struct file_lock *);
f891a29f 1021extern int posix_unblock_lock(struct file_lock *);
3ee17abd 1022extern int vfs_test_lock(struct file *, struct file_lock *);
150b3934 1023extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
9b9d2ab4 1024extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1da177e4 1025extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
df4e8d2c 1026extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1da177e4 1027extern void lease_get_mtime(struct inode *, struct timespec *time);
0af1a450 1028extern int generic_setlease(struct file *, long, struct file_lock **);
a9933cea 1029extern int vfs_setlease(struct file *, long, struct file_lock **);
1da177e4
LT
1030extern int lease_modify(struct file_lock **, int);
1031extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
1032extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
bfcd17a6 1033#else /* !CONFIG_FILE_LOCKING */
c2aca5e5
SW
1034static inline int fcntl_getlk(struct file *file, struct flock __user *user)
1035{
1036 return -EINVAL;
1037}
1038
1039static inline int fcntl_setlk(unsigned int fd, struct file *file,
1040 unsigned int cmd, struct flock __user *user)
1041{
1042 return -EACCES;
1043}
1044
bfcd17a6 1045#if BITS_PER_LONG == 32
c2aca5e5
SW
1046static inline int fcntl_getlk64(struct file *file, struct flock64 __user *user)
1047{
1048 return -EINVAL;
1049}
1050
1051static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1052 unsigned int cmd, struct flock64 __user *user)
1053{
1054 return -EACCES;
1055}
bfcd17a6 1056#endif
c2aca5e5
SW
1057static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1058{
1059 return 0;
1060}
1061
1062static inline int fcntl_getlease(struct file *filp)
1063{
1064 return 0;
1065}
1066
1067static inline void locks_init_lock(struct file_lock *fl)
1068{
1069 return;
1070}
1071
1072static inline void __locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1073{
1074 return;
1075}
1076
1077static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1078{
1079 return;
1080}
1081
1082static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1083{
1084 return;
1085}
1086
78ed8a13 1087static inline void locks_remove_file(struct file *filp)
c2aca5e5
SW
1088{
1089 return;
1090}
1091
1092static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1093{
1094 return;
1095}
1096
1097static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1098 struct file_lock *conflock)
1099{
1100 return -ENOLCK;
1101}
1102
1103static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1104{
1105 return -ENOLCK;
1106}
1107
f891a29f 1108static inline int posix_unblock_lock(struct file_lock *waiter)
c2aca5e5
SW
1109{
1110 return -ENOENT;
1111}
1112
1113static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1114{
1115 return 0;
1116}
1117
1118static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1119 struct file_lock *fl, struct file_lock *conf)
1120{
1121 return -ENOLCK;
1122}
1123
1124static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1125{
1126 return 0;
1127}
1128
1129static inline int flock_lock_file_wait(struct file *filp,
1130 struct file_lock *request)
1131{
1132 return -ENOLCK;
1133}
1134
df4e8d2c 1135static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
c2aca5e5
SW
1136{
1137 return 0;
1138}
1139
1140static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1141{
1142 return;
1143}
1144
1145static inline int generic_setlease(struct file *filp, long arg,
1146 struct file_lock **flp)
1147{
1148 return -EINVAL;
1149}
1150
1151static inline int vfs_setlease(struct file *filp, long arg,
1152 struct file_lock **lease)
1153{
1154 return -EINVAL;
1155}
1156
1157static inline int lease_modify(struct file_lock **before, int arg)
1158{
1159 return -EINVAL;
1160}
1161
1162static inline int lock_may_read(struct inode *inode, loff_t start,
1163 unsigned long len)
1164{
1165 return 1;
1166}
1167
1168static inline int lock_may_write(struct inode *inode, loff_t start,
1169 unsigned long len)
1170{
1171 return 1;
1172}
bfcd17a6
TP
1173#endif /* !CONFIG_FILE_LOCKING */
1174
1da177e4
LT
1175
1176struct fasync_struct {
989a2979
ED
1177 spinlock_t fa_lock;
1178 int magic;
1179 int fa_fd;
1180 struct fasync_struct *fa_next; /* singly linked list */
1181 struct file *fa_file;
1182 struct rcu_head fa_rcu;
1da177e4
LT
1183};
1184
1185#define FASYNC_MAGIC 0x4601
1186
1187/* SMP safe fasync helpers: */
1188extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
f7347ce4
LT
1189extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1190extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1191extern struct fasync_struct *fasync_alloc(void);
1192extern void fasync_free(struct fasync_struct *);
1193
1da177e4
LT
1194/* can be called from interrupts */
1195extern void kill_fasync(struct fasync_struct **, int, int);
1da177e4 1196
609d7fa9 1197extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1da177e4
LT
1198extern int f_setown(struct file *filp, unsigned long arg, int force);
1199extern void f_delown(struct file *filp);
609d7fa9 1200extern pid_t f_getown(struct file *filp);
1da177e4
LT
1201extern int send_sigurg(struct fown_struct *fown);
1202
5accdf82
JK
1203struct mm_struct;
1204
1da177e4
LT
1205/*
1206 * Umount options
1207 */
1208
1209#define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1210#define MNT_DETACH 0x00000002 /* Just detach from the tree */
1211#define MNT_EXPIRE 0x00000004 /* Mark for expiry */
db1f05bb
MS
1212#define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1213#define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1da177e4
LT
1214
1215extern struct list_head super_blocks;
1216extern spinlock_t sb_lock;
1217
5accdf82
JK
1218/* Possible states of 'frozen' field */
1219enum {
1220 SB_UNFROZEN = 0, /* FS is unfrozen */
1221 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
5accdf82
JK
1222 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1223 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1224 * internal threads if needed) */
1225 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1226};
1227
1228#define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1229
1230struct sb_writers {
1231 /* Counters for counting writers at each level */
1232 struct percpu_counter counter[SB_FREEZE_LEVELS];
1233 wait_queue_head_t wait; /* queue for waiting for
1234 writers / faults to finish */
1235 int frozen; /* Is sb frozen? */
1236 wait_queue_head_t wait_unfrozen; /* queue for waiting for
1237 sb to be thawed */
1238#ifdef CONFIG_DEBUG_LOCK_ALLOC
1239 struct lockdep_map lock_map[SB_FREEZE_LEVELS];
1240#endif
1241};
1242
1da177e4
LT
1243struct super_block {
1244 struct list_head s_list; /* Keep this first */
1245 dev_t s_dev; /* search index; _not_ kdev_t */
270ba5f7
RK
1246 unsigned char s_blocksize_bits;
1247 unsigned long s_blocksize;
42cb56ae 1248 loff_t s_maxbytes; /* Max file size */
1da177e4 1249 struct file_system_type *s_type;
ee9b6d61 1250 const struct super_operations *s_op;
61e225dc 1251 const struct dquot_operations *dq_op;
0d54b217 1252 const struct quotactl_ops *s_qcop;
39655164 1253 const struct export_operations *s_export_op;
1da177e4
LT
1254 unsigned long s_flags;
1255 unsigned long s_magic;
1256 struct dentry *s_root;
1257 struct rw_semaphore s_umount;
1da177e4 1258 int s_count;
1da177e4 1259 atomic_t s_active;
50462062 1260#ifdef CONFIG_SECURITY
1da177e4 1261 void *s_security;
50462062 1262#endif
bb435453 1263 const struct xattr_handler **s_xattr;
1da177e4
LT
1264
1265 struct list_head s_inodes; /* all inodes */
ceb5bdc2 1266 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
39f7c4db 1267 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1da177e4 1268 struct block_device *s_bdev;
32a88aa1 1269 struct backing_dev_info *s_bdi;
acaebfd8 1270 struct mtd_info *s_mtd;
a5166169 1271 struct hlist_node s_instances;
1da177e4
LT
1272 struct quota_info s_dquot; /* Diskquota specific options */
1273
5accdf82 1274 struct sb_writers s_writers;
1da177e4
LT
1275
1276 char s_id[32]; /* Informational name */
0bba0169 1277 u8 s_uuid[16]; /* UUID */
1da177e4
LT
1278
1279 void *s_fs_info; /* Filesystem private info */
8de52778 1280 unsigned int s_max_links;
30c40d2c 1281 fmode_t s_mode;
1da177e4 1282
270ba5f7
RK
1283 /* Granularity of c/m/atime in ns.
1284 Cannot be worse than a second */
1285 u32 s_time_gran;
1286
1da177e4
LT
1287 /*
1288 * The next field is for VFS *only*. No filesystems have any business
1289 * even looking at it. You had been warned.
1290 */
a11f3a05 1291 struct mutex s_vfs_rename_mutex; /* Kludge */
1da177e4 1292
79c0b2df
MS
1293 /*
1294 * Filesystem subtype. If non-empty the filesystem type field
1295 * in /proc/mounts will be "type.subtype"
1296 */
1297 char *s_subtype;
b3b304a2
MS
1298
1299 /*
1300 * Saved mount options for lazy filesystems using
1301 * generic_show_options()
1302 */
4d2deb40 1303 char __rcu *s_options;
c8aebb0c 1304 const struct dentry_operations *s_d_op; /* default d_op for dentries */
9fdfdcf1
DM
1305
1306 /*
1307 * Saved pool identifier for cleancache (-1 means none)
1308 */
1309 int cleancache_poolid;
b0d40c92
DC
1310
1311 struct shrinker s_shrink; /* per-sb shrinker handle */
4ed5e82f 1312
7ada4db8
MS
1313 /* Number of inodes with nlink == 0 but still referenced */
1314 atomic_long_t s_remove_count;
1315
4ed5e82f
MS
1316 /* Being remounted read-only */
1317 int s_readonly_remount;
7b7a8665
CH
1318
1319 /* AIO completions deferred from interrupt context */
1320 struct workqueue_struct *s_dio_done_wq;
f6041567
DC
1321
1322 /*
1323 * Keep the lru lists last in the structure so they always sit on their
1324 * own individual cachelines.
1325 */
1326 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1327 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
e2fec7c3 1328 struct rcu_head rcu;
1da177e4
LT
1329};
1330
1331extern struct timespec current_fs_time(struct super_block *sb);
1332
1333/*
1334 * Snapshotting support.
1335 */
1da177e4 1336
5accdf82
JK
1337void __sb_end_write(struct super_block *sb, int level);
1338int __sb_start_write(struct super_block *sb, int level, bool wait);
1339
1340/**
1341 * sb_end_write - drop write access to a superblock
1342 * @sb: the super we wrote to
1343 *
1344 * Decrement number of writers to the filesystem. Wake up possible waiters
1345 * wanting to freeze the filesystem.
1346 */
1347static inline void sb_end_write(struct super_block *sb)
1348{
1349 __sb_end_write(sb, SB_FREEZE_WRITE);
1350}
1351
1352/**
1353 * sb_end_pagefault - drop write access to a superblock from a page fault
1354 * @sb: the super we wrote to
1355 *
1356 * Decrement number of processes handling write page fault to the filesystem.
1357 * Wake up possible waiters wanting to freeze the filesystem.
1358 */
1359static inline void sb_end_pagefault(struct super_block *sb)
1360{
1361 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1362}
1363
1364/**
1365 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1366 * @sb: the super we wrote to
1367 *
1368 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1369 * waiters wanting to freeze the filesystem.
1370 */
1371static inline void sb_end_intwrite(struct super_block *sb)
1372{
1373 __sb_end_write(sb, SB_FREEZE_FS);
1374}
1375
1376/**
1377 * sb_start_write - get write access to a superblock
1378 * @sb: the super we write to
1379 *
1380 * When a process wants to write data or metadata to a file system (i.e. dirty
1381 * a page or an inode), it should embed the operation in a sb_start_write() -
1382 * sb_end_write() pair to get exclusion against file system freezing. This
1383 * function increments number of writers preventing freezing. If the file
1384 * system is already frozen, the function waits until the file system is
1385 * thawed.
1386 *
1387 * Since freeze protection behaves as a lock, users have to preserve
1388 * ordering of freeze protection and other filesystem locks. Generally,
1389 * freeze protection should be the outermost lock. In particular, we have:
1390 *
1391 * sb_start_write
1392 * -> i_mutex (write path, truncate, directory ops, ...)
1393 * -> s_umount (freeze_super, thaw_super)
1394 */
1395static inline void sb_start_write(struct super_block *sb)
1396{
1397 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1398}
1399
1400static inline int sb_start_write_trylock(struct super_block *sb)
1401{
1402 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1403}
1404
1405/**
1406 * sb_start_pagefault - get write access to a superblock from a page fault
1407 * @sb: the super we write to
1408 *
1409 * When a process starts handling write page fault, it should embed the
1410 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1411 * exclusion against file system freezing. This is needed since the page fault
1412 * is going to dirty a page. This function increments number of running page
1413 * faults preventing freezing. If the file system is already frozen, the
1414 * function waits until the file system is thawed.
1415 *
1416 * Since page fault freeze protection behaves as a lock, users have to preserve
1417 * ordering of freeze protection and other filesystem locks. It is advised to
1418 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1419 * handling code implies lock dependency:
1420 *
1421 * mmap_sem
1422 * -> sb_start_pagefault
1423 */
1424static inline void sb_start_pagefault(struct super_block *sb)
1425{
1426 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1427}
1428
1429/*
1430 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1431 * @sb: the super we write to
1432 *
1433 * This is the third level of protection against filesystem freezing. It is
1434 * free for use by a filesystem. The only requirement is that it must rank
1435 * below sb_start_pagefault.
1436 *
1437 * For example filesystem can call sb_start_intwrite() when starting a
1438 * transaction which somewhat eases handling of freezing for internal sources
1439 * of filesystem changes (internal fs threads, discarding preallocation on file
1440 * close, etc.).
1441 */
1442static inline void sb_start_intwrite(struct super_block *sb)
1443{
1444 __sb_start_write(sb, SB_FREEZE_FS, true);
1445}
1da177e4 1446
1da177e4 1447
2e149670 1448extern bool inode_owner_or_capable(const struct inode *inode);
3bd858ab 1449
1da177e4
LT
1450/*
1451 * VFS helper functions..
1452 */
312b63fb 1453extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
18bb1db3 1454extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1a67aafb 1455extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
db2e747b 1456extern int vfs_symlink(struct inode *, struct dentry *, const char *);
146a8595 1457extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1da177e4 1458extern int vfs_rmdir(struct inode *, struct dentry *);
b21996e3 1459extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
8e6d782c 1460extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **);
1da177e4
LT
1461
1462/*
1463 * VFS dentry helper functions.
1464 */
1465extern void dentry_unhash(struct dentry *dentry);
1466
8c744fb8
CH
1467/*
1468 * VFS file helper functions.
1469 */
a1bd120d 1470extern void inode_init_owner(struct inode *inode, const struct inode *dir,
62bb1091 1471 umode_t mode);
c4b929b8
MF
1472/*
1473 * VFS FS_IOC_FIEMAP helper definitions.
1474 */
1475struct fiemap_extent_info {
1476 unsigned int fi_flags; /* Flags as passed from user */
1477 unsigned int fi_extents_mapped; /* Number of mapped extents */
1478 unsigned int fi_extents_max; /* Size of fiemap_extent array */
ecf5632d
NK
1479 struct fiemap_extent __user *fi_extents_start; /* Start of
1480 fiemap_extent array */
c4b929b8
MF
1481};
1482int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1483 u64 phys, u64 len, u32 flags);
1484int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1485
1da177e4
LT
1486/*
1487 * File types
1488 *
1489 * NOTE! These match bits 12..15 of stat.st_mode
1490 * (ie "(i_mode >> 12) & 15").
1491 */
1492#define DT_UNKNOWN 0
1493#define DT_FIFO 1
1494#define DT_CHR 2
1495#define DT_DIR 4
1496#define DT_BLK 6
1497#define DT_REG 8
1498#define DT_LNK 10
1499#define DT_SOCK 12
1500#define DT_WHT 14
1501
1da177e4
LT
1502/*
1503 * This is the "filldir" function type, used by readdir() to let
1504 * the kernel specify what kind of dirent layout it wants to have.
1505 * This allows the kernel to read directories into kernel space or
1506 * to have different dirent layouts depending on the binary type.
1507 */
afefdbb2 1508typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
5c0ba4e0 1509struct dir_context {
ac6614b7 1510 const filldir_t actor;
bb6f619b 1511 loff_t pos;
5c0ba4e0 1512};
bb6f619b 1513
08f85851 1514struct block_device_operations;
1da177e4 1515
1da177e4
LT
1516/* These macros are for out of kernel modules to test that
1517 * the kernel supports the unlocked_ioctl and compat_ioctl
1518 * fields in struct file_operations. */
1519#define HAVE_COMPAT_IOCTL 1
1520#define HAVE_UNLOCKED_IOCTL 1
1521
1da177e4
LT
1522struct file_operations {
1523 struct module *owner;
1524 loff_t (*llseek) (struct file *, loff_t, int);
1525 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1da177e4 1526 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
027445c3
BP
1527 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1528 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
bb6f619b 1529 int (*iterate) (struct file *, struct dir_context *);
1da177e4 1530 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1da177e4
LT
1531 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1532 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1533 int (*mmap) (struct file *, struct vm_area_struct *);
1534 int (*open) (struct inode *, struct file *);
75e1fcc0 1535 int (*flush) (struct file *, fl_owner_t id);
1da177e4 1536 int (*release) (struct inode *, struct file *);
02c24a82 1537 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1da177e4
LT
1538 int (*aio_fsync) (struct kiocb *, int datasync);
1539 int (*fasync) (int, struct file *, int);
1540 int (*lock) (struct file *, int, struct file_lock *);
1da177e4
LT
1541 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1542 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1543 int (*check_flags)(int);
1da177e4 1544 int (*flock) (struct file *, int, struct file_lock *);
cbb7e577
JA
1545 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1546 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
f9ffed26 1547 int (*setlease)(struct file *, long, struct file_lock **);
2fe17c10
CH
1548 long (*fallocate)(struct file *file, int mode, loff_t offset,
1549 loff_t len);
55985dd7 1550 int (*show_fdinfo)(struct seq_file *m, struct file *f);
1da177e4
LT
1551};
1552
1553struct inode_operations {
00cd8dd3 1554 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
44a7d7a8 1555 void * (*follow_link) (struct dentry *, struct nameidata *);
10556cb2 1556 int (*permission) (struct inode *, int);
4e34e719 1557 struct posix_acl * (*get_acl)(struct inode *, int);
44a7d7a8
NP
1558
1559 int (*readlink) (struct dentry *, char __user *,int);
1560 void (*put_link) (struct dentry *, struct nameidata *, void *);
1561
ebfc3b49 1562 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1da177e4
LT
1563 int (*link) (struct dentry *,struct inode *,struct dentry *);
1564 int (*unlink) (struct inode *,struct dentry *);
1565 int (*symlink) (struct inode *,struct dentry *,const char *);
18bb1db3 1566 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1da177e4 1567 int (*rmdir) (struct inode *,struct dentry *);
1a67aafb 1568 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1da177e4
LT
1569 int (*rename) (struct inode *, struct dentry *,
1570 struct inode *, struct dentry *);
1da177e4
LT
1571 int (*setattr) (struct dentry *, struct iattr *);
1572 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1573 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1574 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1575 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1576 int (*removexattr) (struct dentry *, const char *);
c4b929b8
MF
1577 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1578 u64 len);
c3b2da31 1579 int (*update_time)(struct inode *, struct timespec *, int);
d9585277 1580 int (*atomic_open)(struct inode *, struct dentry *,
30d90494 1581 struct file *, unsigned open_flag,
d9585277 1582 umode_t create_mode, int *opened);
60545d0d 1583 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
893d46e4 1584 int (*set_acl)(struct inode *, struct posix_acl *, int);
44a7d7a8 1585} ____cacheline_aligned;
1da177e4 1586
eed4e51f 1587ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
fcf63409
CY
1588 unsigned long nr_segs, unsigned long fast_segs,
1589 struct iovec *fast_pointer,
ac34ebb3 1590 struct iovec **ret_pointer);
eed4e51f 1591
1da177e4
LT
1592extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1593extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1594extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1595 unsigned long, loff_t *);
1596extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1597 unsigned long, loff_t *);
1598
1da177e4
LT
1599struct super_operations {
1600 struct inode *(*alloc_inode)(struct super_block *sb);
1601 void (*destroy_inode)(struct inode *);
1602
aa385729 1603 void (*dirty_inode) (struct inode *, int flags);
a9185b41 1604 int (*write_inode) (struct inode *, struct writeback_control *wbc);
45321ac5 1605 int (*drop_inode) (struct inode *);
be7ce416 1606 void (*evict_inode) (struct inode *);
1da177e4 1607 void (*put_super) (struct super_block *);
1da177e4 1608 int (*sync_fs)(struct super_block *sb, int wait);
c4be0c1d
TS
1609 int (*freeze_fs) (struct super_block *);
1610 int (*unfreeze_fs) (struct super_block *);
726c3342 1611 int (*statfs) (struct dentry *, struct kstatfs *);
1da177e4 1612 int (*remount_fs) (struct super_block *, int *, char *);
42faad99 1613 void (*umount_begin) (struct super_block *);
1da177e4 1614
34c80b1d 1615 int (*show_options)(struct seq_file *, struct dentry *);
d861c630 1616 int (*show_devname)(struct seq_file *, struct dentry *);
a6322de6 1617 int (*show_path)(struct seq_file *, struct dentry *);
64132379 1618 int (*show_stats)(struct seq_file *, struct dentry *);
0e51a720 1619#ifdef CONFIG_QUOTA
1da177e4
LT
1620 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1621 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
0e51a720 1622#endif
87d8fe1e 1623 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
9b17c623
DC
1624 long (*nr_cached_objects)(struct super_block *, int);
1625 long (*free_cached_objects)(struct super_block *, long, int);
1da177e4
LT
1626};
1627
bbc1096a
DH
1628/*
1629 * Inode flags - they have no relation to superblock flags now
1630 */
1631#define S_SYNC 1 /* Writes are synced at once */
1632#define S_NOATIME 2 /* Do not update access times */
1633#define S_APPEND 4 /* Append-only file */
1634#define S_IMMUTABLE 8 /* Immutable file */
1635#define S_DEAD 16 /* removed, but still open directory */
1636#define S_NOQUOTA 32 /* Inode is not counted to quota */
1637#define S_DIRSYNC 64 /* Directory modifications are synchronous */
1638#define S_NOCMTIME 128 /* Do not update file c/mtime */
1639#define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1640#define S_PRIVATE 512 /* Inode is fs-internal */
1641#define S_IMA 1024 /* Inode has an associated IMA struct */
1642#define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1643#define S_NOSEC 4096 /* no suid or xattr security attributes */
1644
1645/*
1646 * Note that nosuid etc flags are inode-specific: setting some file-system
1647 * flags just means all the inodes inherit those flags by default. It might be
1648 * possible to override it selectively if you really wanted to with some
1649 * ioctl() that is not currently implemented.
1650 *
1651 * Exception: MS_RDONLY is always applied to the entire file system.
1652 *
1653 * Unfortunately, it is possible to change a filesystems flags with it mounted
1654 * with files in use. This means that all of the inodes will not have their
1655 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1656 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1657 */
1658#define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1659
1660#define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1661#define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1662 ((inode)->i_flags & S_SYNC))
1663#define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1664 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1665#define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1666#define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1667#define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1668
1669#define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1670#define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1671#define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1672#define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1673
1674#define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1675#define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1676#define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1677#define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1678#define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1679#define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1680#define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1681
1c0eeaf5 1682/*
250df6ed 1683 * Inode state bits. Protected by inode->i_lock
1c0eeaf5
JE
1684 *
1685 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1686 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1687 *
1688 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1689 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1690 * various stages of removing an inode.
1691 *
eaff8079 1692 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1c0eeaf5 1693 *
e7ca2d41
JE
1694 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1695 * fdatasync(). i_atime is the usual cause.
000cb48e
JK
1696 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1697 * these changes separately from I_DIRTY_SYNC so that we
1698 * don't have to write inode on fdatasync() when only
1699 * mtime has changed in it.
1c0eeaf5 1700 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
eaff8079
CH
1701 * I_NEW Serves as both a mutex and completion notification.
1702 * New inodes set I_NEW. If two processes both create
1703 * the same inode, one of them will release its inode and
1704 * wait for I_NEW to be released before returning.
1705 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1706 * also cause waiting on I_NEW, without I_NEW actually
1707 * being set. find_inode() uses this to prevent returning
1708 * nearly-dead inodes.
1c0eeaf5
JE
1709 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1710 * is zero. I_FREEING must be set when I_WILL_FREE is
1711 * cleared.
1712 * I_FREEING Set when inode is about to be freed but still has dirty
1713 * pages or buffers attached or the inode itself is still
1714 * dirty.
dbd5768f
JK
1715 * I_CLEAR Added by clear_inode(). In this state the inode is
1716 * clean and can be destroyed. Inode keeps I_FREEING.
1c0eeaf5
JE
1717 *
1718 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1719 * prohibited for many purposes. iget() must wait for
1720 * the inode to be completely released, then create it
1721 * anew. Other functions will just ignore such inodes,
eaff8079 1722 * if appropriate. I_NEW is used for waiting.
1c0eeaf5 1723 *
169ebd90
JK
1724 * I_SYNC Writeback of inode is running. The bit is set during
1725 * data writeback, and cleared with a wakeup on the bit
1726 * address once it is done. The bit is also used to pin
1727 * the inode in memory for flusher thread.
1c0eeaf5 1728 *
bd5fe6c5
CH
1729 * I_REFERENCED Marks the inode as recently references on the LRU list.
1730 *
1731 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1732 *
1c0eeaf5 1733 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1c0eeaf5 1734 */
9e38d86f
NP
1735#define I_DIRTY_SYNC (1 << 0)
1736#define I_DIRTY_DATASYNC (1 << 1)
1737#define I_DIRTY_PAGES (1 << 2)
eaff8079
CH
1738#define __I_NEW 3
1739#define I_NEW (1 << __I_NEW)
9e38d86f
NP
1740#define I_WILL_FREE (1 << 4)
1741#define I_FREEING (1 << 5)
1742#define I_CLEAR (1 << 6)
eaff8079 1743#define __I_SYNC 7
1c0eeaf5 1744#define I_SYNC (1 << __I_SYNC)
9e38d86f 1745#define I_REFERENCED (1 << 8)
bd5fe6c5
CH
1746#define __I_DIO_WAKEUP 9
1747#define I_DIO_WAKEUP (1 << I_DIO_WAKEUP)
f4e0c30c 1748#define I_LINKABLE (1 << 10)
1da177e4
LT
1749
1750#define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1751
1752extern void __mark_inode_dirty(struct inode *, int);
1753static inline void mark_inode_dirty(struct inode *inode)
1754{
1755 __mark_inode_dirty(inode, I_DIRTY);
1756}
1757
1758static inline void mark_inode_dirty_sync(struct inode *inode)
1759{
1760 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1761}
1762
7ada4db8
MS
1763extern void inc_nlink(struct inode *inode);
1764extern void drop_nlink(struct inode *inode);
1765extern void clear_nlink(struct inode *inode);
1766extern void set_nlink(struct inode *inode, unsigned int nlink);
d8c76e6f
DH
1767
1768static inline void inode_inc_link_count(struct inode *inode)
1769{
1770 inc_nlink(inode);
66d21730
AD
1771 mark_inode_dirty(inode);
1772}
1773
9a53c3a7
DH
1774static inline void inode_dec_link_count(struct inode *inode)
1775{
1776 drop_nlink(inode);
66d21730
AD
1777 mark_inode_dirty(inode);
1778}
1779
25ec56b5
JNC
1780/**
1781 * inode_inc_iversion - increments i_version
1782 * @inode: inode that need to be updated
1783 *
1784 * Every time the inode is modified, the i_version field will be incremented.
1785 * The filesystem has to be mounted with i_version flag
1786 */
1787
1788static inline void inode_inc_iversion(struct inode *inode)
1789{
1790 spin_lock(&inode->i_lock);
1791 inode->i_version++;
1792 spin_unlock(&inode->i_lock);
1793}
1794
c3b2da31
JB
1795enum file_time_flags {
1796 S_ATIME = 1,
1797 S_MTIME = 2,
1798 S_CTIME = 4,
1799 S_VERSION = 8,
1800};
1801
badcf2b7 1802extern void touch_atime(const struct path *);
1da177e4
LT
1803static inline void file_accessed(struct file *file)
1804{
1805 if (!(file->f_flags & O_NOATIME))
68ac1234 1806 touch_atime(&file->f_path);
1da177e4
LT
1807}
1808
1809int sync_inode(struct inode *inode, struct writeback_control *wbc);
c3765016 1810int sync_inode_metadata(struct inode *inode, int wait);
1da177e4 1811
1da177e4
LT
1812struct file_system_type {
1813 const char *name;
1814 int fs_flags;
bbc1096a
DH
1815#define FS_REQUIRES_DEV 1
1816#define FS_BINARY_MOUNTDATA 2
1817#define FS_HAS_SUBTYPE 4
0c55cfc4
EB
1818#define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
1819#define FS_USERNS_DEV_MOUNT 16 /* A userns mount does not imply MNT_NODEV */
bbc1096a 1820#define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
c96e41e9
AV
1821 struct dentry *(*mount) (struct file_system_type *, int,
1822 const char *, void *);
1da177e4
LT
1823 void (*kill_sb) (struct super_block *);
1824 struct module *owner;
1825 struct file_system_type * next;
a5166169 1826 struct hlist_head fs_supers;
d475fd42 1827
cf516249 1828 struct lock_class_key s_lock_key;
897c6ff9 1829 struct lock_class_key s_umount_key;
51ee049e 1830 struct lock_class_key s_vfs_rename_key;
5accdf82 1831 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
d475fd42
PZ
1832
1833 struct lock_class_key i_lock_key;
1834 struct lock_class_key i_mutex_key;
14358e6d 1835 struct lock_class_key i_mutex_dir_key;
1da177e4
LT
1836};
1837
7f78e035
EB
1838#define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
1839
ceefda69
AV
1840extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1841 void *data, int (*fill_super)(struct super_block *, void *, int));
152a0836
AV
1842extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1843 int flags, const char *dev_name, void *data,
1844 int (*fill_super)(struct super_block *, void *, int));
fc14f2fe
AV
1845extern struct dentry *mount_single(struct file_system_type *fs_type,
1846 int flags, void *data,
1847 int (*fill_super)(struct super_block *, void *, int));
3c26ff6e
AV
1848extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1849 int flags, void *data,
1850 int (*fill_super)(struct super_block *, void *, int));
ea441d11 1851extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
1da177e4
LT
1852void generic_shutdown_super(struct super_block *sb);
1853void kill_block_super(struct super_block *sb);
1854void kill_anon_super(struct super_block *sb);
1855void kill_litter_super(struct super_block *sb);
1856void deactivate_super(struct super_block *sb);
74dbbdd7 1857void deactivate_locked_super(struct super_block *sb);
1da177e4 1858int set_anon_super(struct super_block *s, void *data);
0ee5dc67
AV
1859int get_anon_bdev(dev_t *);
1860void free_anon_bdev(dev_t);
1da177e4
LT
1861struct super_block *sget(struct file_system_type *type,
1862 int (*test)(struct super_block *,void *),
1863 int (*set)(struct super_block *,void *),
9249e17f 1864 int flags, void *data);
51139ada 1865extern struct dentry *mount_pseudo(struct file_system_type *, char *,
c74a1cbb
AV
1866 const struct super_operations *ops,
1867 const struct dentry_operations *dops,
1868 unsigned long);
1da177e4
LT
1869
1870/* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1871#define fops_get(fops) \
1872 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1873#define fops_put(fops) \
1874 do { if (fops) module_put((fops)->owner); } while(0)
e84f9e57
AV
1875/*
1876 * This one is to be used *ONLY* from ->open() instances.
1877 * fops must be non-NULL, pinned down *and* module dependencies
1878 * should be sufficient to pin the caller down as well.
1879 */
1880#define replace_fops(f, fops) \
1881 do { \
1882 struct file *__file = (f); \
1883 fops_put(__file->f_op); \
1884 BUG_ON(!(__file->f_op = (fops))); \
1885 } while(0)
1da177e4
LT
1886
1887extern int register_filesystem(struct file_system_type *);
1888extern int unregister_filesystem(struct file_system_type *);
8bf9725c
PE
1889extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1890#define kern_mount(type) kern_mount_data(type, NULL)
423e0ab0 1891extern void kern_unmount(struct vfsmount *mnt);
1da177e4
LT
1892extern int may_umount_tree(struct vfsmount *);
1893extern int may_umount(struct vfsmount *);
808d4e3c 1894extern long do_mount(const char *, const char *, const char *, unsigned long, void *);
589ff870 1895extern struct vfsmount *collect_mounts(struct path *);
8aec0809 1896extern void drop_collected_mounts(struct vfsmount *);
1f707137
AV
1897extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1898 struct vfsmount *);
ebabe9a9 1899extern int vfs_statfs(struct path *, struct kstatfs *);
c8b91acc
AV
1900extern int user_statfs(const char __user *, struct kstatfs *);
1901extern int fd_statfs(int, struct kstatfs *);
cf31e70d 1902extern int vfs_ustat(dev_t, struct kstatfs *);
18e9e510
JB
1903extern int freeze_super(struct super_block *super);
1904extern int thaw_super(struct super_block *super);
02125a82 1905extern bool our_mnt(struct vfsmount *mnt);
e51db735 1906extern bool fs_fully_visible(struct file_system_type *);
1da177e4 1907
ce3b0f8d
AV
1908extern int current_umask(void);
1909
5a14696c
BF
1910extern void ihold(struct inode * inode);
1911extern void iput(struct inode *);
1912
f87fd4c2 1913/* /sys/fs */
00d26666 1914extern struct kobject *fs_kobj;
f87fd4c2 1915
435f49a5 1916#define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
bfcd17a6 1917
1da177e4
LT
1918#define FLOCK_VERIFY_READ 1
1919#define FLOCK_VERIFY_WRITE 2
1920
bfcd17a6 1921#ifdef CONFIG_FILE_LOCKING
1da177e4
LT
1922extern int locks_mandatory_locked(struct inode *);
1923extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1924
1925/*
1926 * Candidates for mandatory locking have the setgid bit set
1927 * but no group execute bit - an otherwise meaningless combination.
1928 */
a16877ca
PE
1929
1930static inline int __mandatory_lock(struct inode *ino)
1931{
1932 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1933}
1934
1935/*
1936 * ... and these candidates should be on MS_MANDLOCK mounted fs,
1937 * otherwise these will be advisory locks
1938 */
1939
1940static inline int mandatory_lock(struct inode *ino)
1941{
1942 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1943}
1da177e4
LT
1944
1945static inline int locks_verify_locked(struct inode *inode)
1946{
a16877ca 1947 if (mandatory_lock(inode))
1da177e4
LT
1948 return locks_mandatory_locked(inode);
1949 return 0;
1950}
1951
1da177e4
LT
1952static inline int locks_verify_truncate(struct inode *inode,
1953 struct file *filp,
1954 loff_t size)
1955{
a16877ca 1956 if (inode->i_flock && mandatory_lock(inode))
1da177e4
LT
1957 return locks_mandatory_area(
1958 FLOCK_VERIFY_WRITE, inode, filp,
1959 size < inode->i_size ? size : inode->i_size,
1960 (size < inode->i_size ? inode->i_size - size
1961 : size - inode->i_size)
1962 );
1963 return 0;
1964}
1965
1966static inline int break_lease(struct inode *inode, unsigned int mode)
1967{
24cbe784
JL
1968 /*
1969 * Since this check is lockless, we must ensure that any refcounts
1970 * taken are done before checking inode->i_flock. Otherwise, we could
1971 * end up racing with tasks trying to set a new lease on this file.
1972 */
1973 smp_mb();
1da177e4 1974 if (inode->i_flock)
df4e8d2c 1975 return __break_lease(inode, mode, FL_LEASE);
1da177e4
LT
1976 return 0;
1977}
df4e8d2c
BF
1978
1979static inline int break_deleg(struct inode *inode, unsigned int mode)
1980{
1981 if (inode->i_flock)
1982 return __break_lease(inode, mode, FL_DELEG);
1da177e4
LT
1983 return 0;
1984}
df4e8d2c 1985
5a14696c
BF
1986static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
1987{
1988 int ret;
1989
1990 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
1991 if (ret == -EWOULDBLOCK && delegated_inode) {
1992 *delegated_inode = inode;
1993 ihold(inode);
1994 }
1995 return ret;
1996}
1997
1998static inline int break_deleg_wait(struct inode **delegated_inode)
1999{
2000 int ret;
2001
2002 ret = break_deleg(*delegated_inode, O_WRONLY);
2003 iput(*delegated_inode);
2004 *delegated_inode = NULL;
2005 return ret;
2006}
2007
bfcd17a6 2008#else /* !CONFIG_FILE_LOCKING */
af5df566
SW
2009static inline int locks_mandatory_locked(struct inode *inode)
2010{
2011 return 0;
2012}
2013
2014static inline int locks_mandatory_area(int rw, struct inode *inode,
2015 struct file *filp, loff_t offset,
2016 size_t count)
2017{
2018 return 0;
2019}
2020
2021static inline int __mandatory_lock(struct inode *inode)
2022{
2023 return 0;
2024}
2025
2026static inline int mandatory_lock(struct inode *inode)
2027{
2028 return 0;
2029}
2030
2031static inline int locks_verify_locked(struct inode *inode)
2032{
2033 return 0;
2034}
2035
2036static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2037 size_t size)
2038{
2039 return 0;
2040}
2041
2042static inline int break_lease(struct inode *inode, unsigned int mode)
2043{
2044 return 0;
2045}
2046
df4e8d2c
BF
2047static inline int break_deleg(struct inode *inode, unsigned int mode)
2048{
2049 return 0;
2050}
5a14696c
BF
2051
2052static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2053{
2054 return 0;
2055}
2056
2057static inline int break_deleg_wait(struct inode **delegated_inode)
2058{
2059 BUG();
2060 return 0;
2061}
2062
bfcd17a6 2063#endif /* CONFIG_FILE_LOCKING */
1da177e4
LT
2064
2065/* fs/open.c */
adb5c247 2066struct audit_names;
91a27b2a 2067struct filename {
adb5c247
JL
2068 const char *name; /* pointer to actual string */
2069 const __user char *uptr; /* original userland pointer */
2070 struct audit_names *aname;
7950e385 2071 bool separate; /* should "name" be freed? */
91a27b2a 2072};
1da177e4 2073
a02de960 2074extern long vfs_truncate(struct path *, loff_t);
4a30131e
N
2075extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2076 struct file *filp);
3e63cbb1
AJ
2077extern int do_fallocate(struct file *file, int mode, loff_t offset,
2078 loff_t len);
8e8a1407 2079extern long do_sys_open(int dfd, const char __user *filename, int flags,
a218d0fd 2080 umode_t mode);
669abf4e 2081extern struct file *file_open_name(struct filename *, int, umode_t);
a218d0fd 2082extern struct file *filp_open(const char *, int, umode_t);
73d049a4
AV
2083extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2084 const char *, int);
765927b2 2085extern struct file * dentry_open(const struct path *, int, const struct cred *);
1da177e4 2086extern int filp_close(struct file *, fl_owner_t id);
91a27b2a
JL
2087
2088extern struct filename *getname(const char __user *);
2089
47237687
AV
2090enum {
2091 FILE_CREATED = 1,
2092 FILE_OPENED = 2
2093};
30d90494
AV
2094extern int finish_open(struct file *file, struct dentry *dentry,
2095 int (*open)(struct inode *, struct file *),
2096 int *opened);
e45198a6 2097extern int finish_no_open(struct file *file, struct dentry *dentry);
1da177e4 2098
3e63cbb1
AJ
2099/* fs/ioctl.c */
2100
2101extern int ioctl_preallocate(struct file *filp, void __user *argp);
2102
1da177e4
LT
2103/* fs/dcache.c */
2104extern void __init vfs_caches_init_early(void);
2105extern void __init vfs_caches_init(unsigned long);
2106
b86c089b
CL
2107extern struct kmem_cache *names_cachep;
2108
91a27b2a
JL
2109extern void final_putname(struct filename *name);
2110
a608ca21 2111#define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
3446a8aa 2112#define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
1da177e4 2113#ifndef CONFIG_AUDITSYSCALL
91a27b2a 2114#define putname(name) final_putname(name)
1da177e4 2115#else
91a27b2a 2116extern void putname(struct filename *name);
1da177e4
LT
2117#endif
2118
9361401e 2119#ifdef CONFIG_BLOCK
1da177e4 2120extern int register_blkdev(unsigned int, const char *);
f4480240 2121extern void unregister_blkdev(unsigned int, const char *);
1da177e4 2122extern struct block_device *bdget(dev_t);
dddac6a7 2123extern struct block_device *bdgrab(struct block_device *bdev);
1da177e4
LT
2124extern void bd_set_size(struct block_device *, loff_t size);
2125extern void bd_forget(struct inode *inode);
2126extern void bdput(struct block_device *);
585d3bc0 2127extern void invalidate_bdev(struct block_device *);
5c0d6b60 2128extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
585d3bc0 2129extern int sync_blockdev(struct block_device *bdev);
ff01bb48 2130extern void kill_bdev(struct block_device *);
585d3bc0 2131extern struct super_block *freeze_bdev(struct block_device *);
c2d75438 2132extern void emergency_thaw_all(void);
585d3bc0
NP
2133extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2134extern int fsync_bdev(struct block_device *);
a8855990 2135extern int sb_is_blkdev_sb(struct super_block *sb);
9361401e
DH
2136#else
2137static inline void bd_forget(struct inode *inode) {}
47e4491b 2138static inline int sync_blockdev(struct block_device *bdev) { return 0; }
ff01bb48 2139static inline void kill_bdev(struct block_device *bdev) {}
47e4491b
AV
2140static inline void invalidate_bdev(struct block_device *bdev) {}
2141
2142static inline struct super_block *freeze_bdev(struct block_device *sb)
2143{
2144 return NULL;
2145}
2146
2147static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2148{
2149 return 0;
2150}
5c0d6b60
JK
2151
2152static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2153{
2154}
a8855990
JK
2155
2156static inline int sb_is_blkdev_sb(struct super_block *sb)
2157{
2158 return 0;
2159}
9361401e 2160#endif
60b0680f 2161extern int sync_filesystem(struct super_block *);
9361401e 2162extern const struct file_operations def_blk_fops;
4b6f5d20
AV
2163extern const struct file_operations def_chr_fops;
2164extern const struct file_operations bad_sock_fops;
9361401e 2165#ifdef CONFIG_BLOCK
1da177e4 2166extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
56b26add 2167extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
1da177e4 2168extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
e525fd89 2169extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
d4d77629
TH
2170extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2171 void *holder);
2172extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2173 void *holder);
4385bab1 2174extern void blkdev_put(struct block_device *bdev, fmode_t mode);
641dc636 2175#ifdef CONFIG_SYSFS
e09b457b 2176extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
49731baa
TH
2177extern void bd_unlink_disk_holder(struct block_device *bdev,
2178 struct gendisk *disk);
641dc636 2179#else
e09b457b
TH
2180static inline int bd_link_disk_holder(struct block_device *bdev,
2181 struct gendisk *disk)
2182{
2183 return 0;
2184}
49731baa
TH
2185static inline void bd_unlink_disk_holder(struct block_device *bdev,
2186 struct gendisk *disk)
2187{
2188}
641dc636 2189#endif
9361401e 2190#endif
1da177e4
LT
2191
2192/* fs/char_dev.c */
68eef3b4 2193#define CHRDEV_MAJOR_HASH_SIZE 255
1da177e4
LT
2194extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2195extern int register_chrdev_region(dev_t, unsigned, const char *);
1905b1bf
TH
2196extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2197 unsigned int count, const char *name,
2198 const struct file_operations *fops);
2199extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2200 unsigned int count, const char *name);
1da177e4 2201extern void unregister_chrdev_region(dev_t, unsigned);
68eef3b4 2202extern void chrdev_show(struct seq_file *,off_t);
1da177e4 2203
1905b1bf
TH
2204static inline int register_chrdev(unsigned int major, const char *name,
2205 const struct file_operations *fops)
2206{
2207 return __register_chrdev(major, 0, 256, name, fops);
2208}
2209
2210static inline void unregister_chrdev(unsigned int major, const char *name)
2211{
2212 __unregister_chrdev(major, 0, 256, name);
2213}
2214
1da177e4
LT
2215/* fs/block_dev.c */
2216#define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
1f014290 2217#define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
9361401e
DH
2218
2219#ifdef CONFIG_BLOCK
2220#define BLKDEV_MAJOR_HASH_SIZE 255
1da177e4
LT
2221extern const char *__bdevname(dev_t, char *buffer);
2222extern const char *bdevname(struct block_device *bdev, char *buffer);
2223extern struct block_device *lookup_bdev(const char *);
68eef3b4 2224extern void blkdev_show(struct seq_file *,off_t);
30c40d2c 2225
9361401e
DH
2226#else
2227#define BLKDEV_MAJOR_HASH_SIZE 0
2228#endif
1da177e4
LT
2229
2230extern void init_special_inode(struct inode *, umode_t, dev_t);
2231
2232/* Invalid inode operations -- fs/bad_inode.c */
2233extern void make_bad_inode(struct inode *);
2234extern int is_bad_inode(struct inode *);
2235
9361401e 2236#ifdef CONFIG_BLOCK
1da177e4
LT
2237/*
2238 * return READ, READA, or WRITE
2239 */
2240#define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2241
2242/*
2243 * return data direction, READ or WRITE
2244 */
2245#define bio_data_dir(bio) ((bio)->bi_rw & 1)
2246
c3279d14
AP
2247extern void check_disk_size_change(struct gendisk *disk,
2248 struct block_device *bdev);
0c002c2f 2249extern int revalidate_disk(struct gendisk *);
1da177e4 2250extern int check_disk_change(struct block_device *);
93b270f7 2251extern int __invalidate_device(struct block_device *, bool);
1da177e4 2252extern int invalidate_partition(struct gendisk *, int);
9361401e 2253#endif
1da177e4
LT
2254unsigned long invalidate_mapping_pages(struct address_space *mapping,
2255 pgoff_t start, pgoff_t end);
54bc4855 2256
1da177e4
LT
2257static inline void invalidate_remote_inode(struct inode *inode)
2258{
2259 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2260 S_ISLNK(inode->i_mode))
fc0ecff6 2261 invalidate_mapping_pages(inode->i_mapping, 0, -1);
1da177e4
LT
2262}
2263extern int invalidate_inode_pages2(struct address_space *mapping);
2264extern int invalidate_inode_pages2_range(struct address_space *mapping,
2265 pgoff_t start, pgoff_t end);
2266extern int write_inode_now(struct inode *, int);
2267extern int filemap_fdatawrite(struct address_space *);
2268extern int filemap_flush(struct address_space *);
2269extern int filemap_fdatawait(struct address_space *);
d3bccb6f
JK
2270extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2271 loff_t lend);
1da177e4
LT
2272extern int filemap_write_and_wait(struct address_space *mapping);
2273extern int filemap_write_and_wait_range(struct address_space *mapping,
2274 loff_t lstart, loff_t lend);
ebcf28e1
AM
2275extern int __filemap_fdatawrite_range(struct address_space *mapping,
2276 loff_t start, loff_t end, int sync_mode);
f4c0a0fd
JK
2277extern int filemap_fdatawrite_range(struct address_space *mapping,
2278 loff_t start, loff_t end);
ebcf28e1 2279
8018ab05
CH
2280extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2281 int datasync);
2282extern int vfs_fsync(struct file *file, int datasync);
148f948b 2283extern int generic_write_sync(struct file *file, loff_t pos, loff_t count);
1da177e4
LT
2284extern void emergency_sync(void);
2285extern void emergency_remount(void);
9361401e 2286#ifdef CONFIG_BLOCK
1da177e4 2287extern sector_t bmap(struct inode *, sector_t);
9361401e 2288#endif
27ac0ffe 2289extern int notify_change(struct dentry *, struct iattr *, struct inode **);
f419a2e3 2290extern int inode_permission(struct inode *, int);
2830ba7f 2291extern int generic_permission(struct inode *, int);
1da177e4 2292
f696a365
MS
2293static inline bool execute_ok(struct inode *inode)
2294{
2295 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2296}
2297
496ad9aa
AV
2298static inline struct inode *file_inode(struct file *f)
2299{
dd37978c 2300 return f->f_inode;
496ad9aa
AV
2301}
2302
8d71db4f
AV
2303static inline void file_start_write(struct file *file)
2304{
2305 if (!S_ISREG(file_inode(file)->i_mode))
2306 return;
2307 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2308}
2309
5ae98f15
JK
2310static inline bool file_start_write_trylock(struct file *file)
2311{
2312 if (!S_ISREG(file_inode(file)->i_mode))
2313 return true;
2314 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2315}
2316
8d71db4f
AV
2317static inline void file_end_write(struct file *file)
2318{
2319 if (!S_ISREG(file_inode(file)->i_mode))
2320 return;
2321 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2322}
2323
07b8ce1e
AV
2324/*
2325 * get_write_access() gets write permission for a file.
2326 * put_write_access() releases this write permission.
2327 * This is used for regular files.
2328 * We cannot support write (and maybe mmap read-write shared) accesses and
2329 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2330 * can have the following values:
2331 * 0: no writers, no VM_DENYWRITE mappings
2332 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2333 * > 0: (i_writecount) users are writing to the file.
2334 *
2335 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2336 * except for the cases where we don't hold i_writecount yet. Then we need to
2337 * use {get,deny}_write_access() - these functions check the sign and refuse
2338 * to do the change if sign is wrong.
2339 */
2340static inline int get_write_access(struct inode *inode)
2341{
2342 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2343}
2344static inline int deny_write_access(struct file *file)
2345{
496ad9aa 2346 struct inode *inode = file_inode(file);
07b8ce1e
AV
2347 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2348}
1da177e4
LT
2349static inline void put_write_access(struct inode * inode)
2350{
2351 atomic_dec(&inode->i_writecount);
2352}
2353static inline void allow_write_access(struct file *file)
2354{
2355 if (file)
496ad9aa 2356 atomic_inc(&file_inode(file)->i_writecount);
1da177e4 2357}
f1fe29b4
DH
2358static inline bool inode_is_open_for_write(const struct inode *inode)
2359{
2360 return atomic_read(&inode->i_writecount) > 0;
2361}
2362
a5c96ebf
MZ
2363#ifdef CONFIG_IMA
2364static inline void i_readcount_dec(struct inode *inode)
2365{
2366 BUG_ON(!atomic_read(&inode->i_readcount));
2367 atomic_dec(&inode->i_readcount);
2368}
2369static inline void i_readcount_inc(struct inode *inode)
2370{
2371 atomic_inc(&inode->i_readcount);
2372}
2373#else
2374static inline void i_readcount_dec(struct inode *inode)
2375{
2376 return;
2377}
2378static inline void i_readcount_inc(struct inode *inode)
2379{
2380 return;
2381}
2382#endif
ed8cae8b 2383extern int do_pipe_flags(int *, int);
1da177e4 2384
6777d773 2385extern int kernel_read(struct file *, loff_t, char *, unsigned long);
7bb307e8 2386extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
1da177e4
LT
2387extern struct file * open_exec(const char *);
2388
2389/* fs/dcache.c -- generic fs support functions */
2390extern int is_subdir(struct dentry *, struct dentry *);
2096f759 2391extern int path_is_under(struct path *, struct path *);
1da177e4
LT
2392
2393#include <linux/err.h>
2394
2395/* needed for stackable file system support */
965c8e59 2396extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
1da177e4 2397
965c8e59 2398extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
1da177e4 2399
54e34621 2400extern int inode_init_always(struct super_block *, struct inode *);
1da177e4 2401extern void inode_init_once(struct inode *);
2aa15890 2402extern void address_space_init_once(struct address_space *mapping);
1da177e4
LT
2403extern struct inode * igrab(struct inode *);
2404extern ino_t iunique(struct super_block *, ino_t);
2405extern int inode_needs_sync(struct inode *inode);
45321ac5 2406extern int generic_delete_inode(struct inode *inode);
9bcb4b73
AV
2407static inline int generic_drop_inode(struct inode *inode)
2408{
2409 return !inode->i_nlink || inode_unhashed(inode);
2410}
1da177e4 2411
88bd5121
AA
2412extern struct inode *ilookup5_nowait(struct super_block *sb,
2413 unsigned long hashval, int (*test)(struct inode *, void *),
2414 void *data);
1da177e4
LT
2415extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2416 int (*test)(struct inode *, void *), void *data);
2417extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2418
2419extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2420extern struct inode * iget_locked(struct super_block *, unsigned long);
261bca86
AV
2421extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2422extern int insert_inode_locked(struct inode *);
e096d0c7
JB
2423#ifdef CONFIG_DEBUG_LOCK_ALLOC
2424extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2425#else
2426static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2427#endif
1da177e4 2428extern void unlock_new_inode(struct inode *);
85fe4025 2429extern unsigned int get_next_ino(void);
1da177e4 2430
1da177e4 2431extern void __iget(struct inode * inode);
b46980fe 2432extern void iget_failed(struct inode *);
dbd5768f 2433extern void clear_inode(struct inode *);
2e00c97e 2434extern void __destroy_inode(struct inode *);
a209dfc7
ED
2435extern struct inode *new_inode_pseudo(struct super_block *sb);
2436extern struct inode *new_inode(struct super_block *sb);
ff0c7d15 2437extern void free_inode_nonrcu(struct inode *inode);
01de85e0 2438extern int should_remove_suid(struct dentry *);
2f1936b8 2439extern int file_remove_suid(struct file *);
1da177e4
LT
2440
2441extern void __insert_inode_hash(struct inode *, unsigned long hashval);
646ec461
CH
2442static inline void insert_inode_hash(struct inode *inode)
2443{
1da177e4
LT
2444 __insert_inode_hash(inode, inode->i_ino);
2445}
f2ee7abf
ED
2446
2447extern void __remove_inode_hash(struct inode *);
2448static inline void remove_inode_hash(struct inode *inode)
2449{
2450 if (!inode_unhashed(inode))
2451 __remove_inode_hash(inode);
2452}
2453
646ec461 2454extern void inode_sb_list_add(struct inode *inode);
1da177e4 2455
9361401e 2456#ifdef CONFIG_BLOCK
1da177e4
LT
2457extern void submit_bio(int, struct bio *);
2458extern int bdev_read_only(struct block_device *);
9361401e 2459#endif
1da177e4
LT
2460extern int set_blocksize(struct block_device *, int);
2461extern int sb_set_blocksize(struct super_block *, int);
2462extern int sb_min_blocksize(struct super_block *, int);
2463
2464extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2465extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
0b173bc4
KK
2466extern int generic_file_remap_pages(struct vm_area_struct *, unsigned long addr,
2467 unsigned long size, pgoff_t pgoff);
1da177e4 2468extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
1da177e4 2469int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
027445c3 2470extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
e4dd9de3
JK
2471extern ssize_t __generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long,
2472 loff_t *);
027445c3 2473extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
1da177e4
LT
2474extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
2475 unsigned long *, loff_t, loff_t *, size_t, size_t);
2476extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
2477 unsigned long, loff_t, loff_t *, size_t, ssize_t);
2478extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2479extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
0ceb3314
DM
2480extern int generic_segment_checks(const struct iovec *iov,
2481 unsigned long *nr_segs, size_t *count, int access_flags);
88e6faef 2482
eef99380
CH
2483/* fs/block_dev.c */
2484extern ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov,
2485 unsigned long nr_segs, loff_t pos);
02c24a82
JB
2486extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2487 int datasync);
ff01bb48 2488extern void block_sync_page(struct page *page);
eef99380 2489
88e6faef 2490/* fs/splice.c */
cbb7e577 2491extern ssize_t generic_file_splice_read(struct file *, loff_t *,
88e6faef 2492 struct pipe_inode_info *, size_t, unsigned int);
6818173b
MS
2493extern ssize_t default_file_splice_read(struct file *, loff_t *,
2494 struct pipe_inode_info *, size_t, unsigned int);
88e6faef 2495extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
cbb7e577 2496 struct file *, loff_t *, size_t, unsigned int);
88e6faef 2497extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
cbb7e577 2498 struct file *out, loff_t *, size_t len, unsigned int flags);
88e6faef 2499
1da177e4
LT
2500extern void
2501file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
965c8e59
AM
2502extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2503extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
46a1c2c7 2504extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
965c8e59 2505extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
5760495a 2506extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
965c8e59 2507 int whence, loff_t maxsize, loff_t eof);
1bf9d14d
AV
2508extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2509 int whence, loff_t size);
1da177e4
LT
2510extern int generic_file_open(struct inode * inode, struct file * filp);
2511extern int nonseekable_open(struct inode * inode, struct file * filp);
2512
ceffc078 2513#ifdef CONFIG_FS_XIP
eb6fe0c3
CO
2514extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2515 loff_t *ppos);
ceffc078 2516extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
eb6fe0c3
CO
2517extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2518 size_t len, loff_t *ppos);
ceffc078 2519extern int xip_truncate_page(struct address_space *mapping, loff_t from);
6d79125b
CO
2520#else
2521static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2522{
2523 return 0;
2524}
ceffc078
CO
2525#endif
2526
9361401e 2527#ifdef CONFIG_BLOCK
facd07b0
JB
2528typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2529 loff_t file_offset);
1da177e4
LT
2530
2531enum {
1e431f5c
CH
2532 /* need locking between buffered and direct access */
2533 DIO_LOCKING = 0x01,
2534
2535 /* filesystem does not support filling holes */
2536 DIO_SKIP_HOLES = 0x02,
1da177e4
LT
2537};
2538
eafdc7d1
CH
2539void dio_end_io(struct bio *bio, int error);
2540
2541ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2542 struct block_device *bdev, const struct iovec *iov, loff_t offset,
2543 unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
2544 dio_submit_t submit_io, int flags);
7bb46a67 2545
1da177e4 2546static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
aacfc19c
CH
2547 struct inode *inode, const struct iovec *iov, loff_t offset,
2548 unsigned long nr_segs, get_block_t get_block)
1da177e4 2549{
aacfc19c
CH
2550 return __blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov,
2551 offset, nr_segs, get_block, NULL, NULL,
1e431f5c 2552 DIO_LOCKING | DIO_SKIP_HOLES);
1da177e4 2553}
9361401e 2554#endif
1da177e4 2555
1d59d61f
TM
2556void inode_dio_wait(struct inode *inode);
2557void inode_dio_done(struct inode *inode);
2558
4b6f5d20 2559extern const struct file_operations generic_ro_fops;
1da177e4
LT
2560
2561#define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2562
2563extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
1da177e4 2564extern int page_readlink(struct dentry *, char __user *, int);
cc314eef
LT
2565extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2566extern void page_put_link(struct dentry *, struct nameidata *, void *);
0adb25d2 2567extern int __page_symlink(struct inode *inode, const char *symname, int len,
54566b2c 2568 int nofs);
1da177e4 2569extern int page_symlink(struct inode *inode, const char *symname, int len);
c5ef1c42 2570extern const struct inode_operations page_symlink_inode_operations;
87dc800b 2571extern void kfree_put_link(struct dentry *, struct nameidata *, void *);
1da177e4
LT
2572extern int generic_readlink(struct dentry *, char __user *, int);
2573extern void generic_fillattr(struct inode *, struct kstat *);
b7a6ec52 2574int vfs_getattr_nosec(struct path *path, struct kstat *stat);
3dadecce 2575extern int vfs_getattr(struct path *, struct kstat *);
b462707e 2576void __inode_add_bytes(struct inode *inode, loff_t bytes);
1da177e4 2577void inode_add_bytes(struct inode *inode, loff_t bytes);
1c8924eb 2578void __inode_sub_bytes(struct inode *inode, loff_t bytes);
1da177e4
LT
2579void inode_sub_bytes(struct inode *inode, loff_t bytes);
2580loff_t inode_get_bytes(struct inode *inode);
2581void inode_set_bytes(struct inode *inode, loff_t bytes);
2582
2583extern int vfs_readdir(struct file *, filldir_t, void *);
5c0ba4e0 2584extern int iterate_dir(struct file *, struct dir_context *);
1da177e4 2585
c7887325
DH
2586extern int vfs_stat(const char __user *, struct kstat *);
2587extern int vfs_lstat(const char __user *, struct kstat *);
1da177e4 2588extern int vfs_fstat(unsigned int, struct kstat *);
c7887325 2589extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
1da177e4 2590
deb21db7
EZ
2591extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2592 unsigned long arg);
e9079cce 2593extern int __generic_block_fiemap(struct inode *inode,
3a3076f4
JB
2594 struct fiemap_extent_info *fieinfo,
2595 loff_t start, loff_t len,
2596 get_block_t *get_block);
68c9d702
JB
2597extern int generic_block_fiemap(struct inode *inode,
2598 struct fiemap_extent_info *fieinfo, u64 start,
2599 u64 len, get_block_t *get_block);
1da177e4 2600
c18479fe
MS
2601extern void get_filesystem(struct file_system_type *fs);
2602extern void put_filesystem(struct file_system_type *fs);
1da177e4
LT
2603extern struct file_system_type *get_fs_type(const char *name);
2604extern struct super_block *get_super(struct block_device *);
6b6dc836 2605extern struct super_block *get_super_thawed(struct block_device *);
4504230a 2606extern struct super_block *get_active_super(struct block_device *bdev);
1da177e4 2607extern void drop_super(struct super_block *sb);
01a05b33 2608extern void iterate_supers(void (*)(struct super_block *, void *), void *);
43e15cdb
AV
2609extern void iterate_supers_type(struct file_system_type *,
2610 void (*)(struct super_block *, void *), void *);
1da177e4
LT
2611
2612extern int dcache_dir_open(struct inode *, struct file *);
2613extern int dcache_dir_close(struct inode *, struct file *);
2614extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
5f99f4e7 2615extern int dcache_readdir(struct file *, struct dir_context *);
7bb46a67 2616extern int simple_setattr(struct dentry *, struct iattr *);
1da177e4 2617extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
726c3342 2618extern int simple_statfs(struct dentry *, struct kstatfs *);
20955e89 2619extern int simple_open(struct inode *inode, struct file *file);
1da177e4
LT
2620extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2621extern int simple_unlink(struct inode *, struct dentry *);
2622extern int simple_rmdir(struct inode *, struct dentry *);
2623extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
02c24a82 2624extern int noop_fsync(struct file *, loff_t, loff_t, int);
1da177e4
LT
2625extern int simple_empty(struct dentry *);
2626extern int simple_readpage(struct file *file, struct page *page);
afddba49
NP
2627extern int simple_write_begin(struct file *file, struct address_space *mapping,
2628 loff_t pos, unsigned len, unsigned flags,
2629 struct page **pagep, void **fsdata);
2630extern int simple_write_end(struct file *file, struct address_space *mapping,
2631 loff_t pos, unsigned len, unsigned copied,
2632 struct page *page, void *fsdata);
b26d4cd3 2633extern int always_delete_dentry(const struct dentry *);
6987843f 2634extern struct inode *alloc_anon_inode(struct super_block *);
b26d4cd3 2635extern const struct dentry_operations simple_dentry_operations;
1da177e4 2636
00cd8dd3 2637extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
1da177e4 2638extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
4b6f5d20 2639extern const struct file_operations simple_dir_operations;
c5ef1c42 2640extern const struct inode_operations simple_dir_inode_operations;
4b6f5d20 2641struct tree_descr { char *name; const struct file_operations *ops; int mode; };
1da177e4 2642struct dentry *d_alloc_name(struct dentry *, const char *);
7d683a09 2643extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
1f5ce9e9 2644extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
1da177e4
LT
2645extern void simple_release_fs(struct vfsmount **mount, int *count);
2646
93b07113
AM
2647extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2648 loff_t *ppos, const void *from, size_t available);
6a727b43
JS
2649extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2650 const void __user *from, size_t count);
1da177e4 2651
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, \
acaefc25
AB
2738};
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
2776static inline void inode_has_no_xattr(struct inode *inode)
2777{
9e1f1de0 2778 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
69b45732
AK
2779 inode->i_flags |= S_NOSEC;
2780}
2781
5f99f4e7
AV
2782static inline bool dir_emit(struct dir_context *ctx,
2783 const char *name, int namelen,
2784 u64 ino, unsigned type)
2785{
2786 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
2787}
2788static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
2789{
2790 return ctx->actor(ctx, ".", 1, ctx->pos,
2791 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
2792}
2793static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
2794{
2795 return ctx->actor(ctx, "..", 2, ctx->pos,
2796 parent_ino(file->f_path.dentry), DT_DIR) == 0;
2797}
2798static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
2799{
2800 if (ctx->pos == 0) {
2801 if (!dir_emit_dot(file, ctx))
2802 return false;
2803 ctx->pos = 1;
2804 }
2805 if (ctx->pos == 1) {
2806 if (!dir_emit_dotdot(file, ctx))
2807 return false;
2808 ctx->pos = 2;
2809 }
2810 return true;
2811}
5ded75ec
AV
2812static inline bool dir_relax(struct inode *inode)
2813{
2814 mutex_unlock(&inode->i_mutex);
2815 mutex_lock(&inode->i_mutex);
2816 return !IS_DEADDIR(inode);
2817}
5f99f4e7 2818
1da177e4 2819#endif /* _LINUX_FS_H */
This page took 1.310766 seconds and 5 git commands to generate.