]> git.proxmox.com Git - mirror_ubuntu-bionic-kernel.git/blob - include/linux/fs.h
Merge branches 'no-rebases', 'arch-avr32', 'arch-blackfin', 'arch-cris', 'arch-h8300...
[mirror_ubuntu-bionic-kernel.git] / include / linux / fs.h
1 #ifndef _LINUX_FS_H
2 #define _LINUX_FS_H
3
4
5 #include <linux/linkage.h>
6 #include <linux/wait.h>
7 #include <linux/kdev_t.h>
8 #include <linux/dcache.h>
9 #include <linux/path.h>
10 #include <linux/stat.h>
11 #include <linux/cache.h>
12 #include <linux/list.h>
13 #include <linux/radix-tree.h>
14 #include <linux/rbtree.h>
15 #include <linux/init.h>
16 #include <linux/pid.h>
17 #include <linux/bug.h>
18 #include <linux/mutex.h>
19 #include <linux/capability.h>
20 #include <linux/semaphore.h>
21 #include <linux/fiemap.h>
22 #include <linux/rculist_bl.h>
23 #include <linux/atomic.h>
24 #include <linux/shrinker.h>
25 #include <linux/migrate_mode.h>
26 #include <linux/uidgid.h>
27 #include <linux/lockdep.h>
28 #include <linux/percpu-rwsem.h>
29 #include <linux/blk_types.h>
30
31 #include <asm/byteorder.h>
32 #include <uapi/linux/fs.h>
33
34 struct export_operations;
35 struct hd_geometry;
36 struct iovec;
37 struct nameidata;
38 struct kiocb;
39 struct kobject;
40 struct pipe_inode_info;
41 struct poll_table_struct;
42 struct kstatfs;
43 struct vm_area_struct;
44 struct vfsmount;
45 struct cred;
46 struct swap_info_struct;
47
48 extern void __init inode_init(void);
49 extern void __init inode_init_early(void);
50 extern void __init files_init(unsigned long);
51
52 extern struct files_stat_struct files_stat;
53 extern unsigned long get_max_files(void);
54 extern int sysctl_nr_open;
55 extern struct inodes_stat_t inodes_stat;
56 extern int leases_enable, lease_break_time;
57 extern int sysctl_protected_symlinks;
58 extern int sysctl_protected_hardlinks;
59
60 struct buffer_head;
61 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
62 struct buffer_head *bh_result, int create);
63 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
64 ssize_t bytes, void *private, int ret,
65 bool is_async);
66
67 #define MAY_EXEC 0x00000001
68 #define MAY_WRITE 0x00000002
69 #define MAY_READ 0x00000004
70 #define MAY_APPEND 0x00000008
71 #define MAY_ACCESS 0x00000010
72 #define MAY_OPEN 0x00000020
73 #define MAY_CHDIR 0x00000040
74 /* called from RCU mode, don't block */
75 #define MAY_NOT_BLOCK 0x00000080
76
77 /*
78 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
79 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
80 */
81
82 /* file is open for reading */
83 #define FMODE_READ ((__force fmode_t)0x1)
84 /* file is open for writing */
85 #define FMODE_WRITE ((__force fmode_t)0x2)
86 /* file is seekable */
87 #define FMODE_LSEEK ((__force fmode_t)0x4)
88 /* file can be accessed using pread */
89 #define FMODE_PREAD ((__force fmode_t)0x8)
90 /* file can be accessed using pwrite */
91 #define FMODE_PWRITE ((__force fmode_t)0x10)
92 /* File is opened for execution with sys_execve / sys_uselib */
93 #define FMODE_EXEC ((__force fmode_t)0x20)
94 /* File is opened with O_NDELAY (only set for block devices) */
95 #define FMODE_NDELAY ((__force fmode_t)0x40)
96 /* File is opened with O_EXCL (only set for block devices) */
97 #define FMODE_EXCL ((__force fmode_t)0x80)
98 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
99 (specialy hack for floppy.c) */
100 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
101 /* 32bit hashes as llseek() offset (for directories) */
102 #define FMODE_32BITHASH ((__force fmode_t)0x200)
103 /* 64bit hashes as llseek() offset (for directories) */
104 #define FMODE_64BITHASH ((__force fmode_t)0x400)
105
106 /*
107 * Don't update ctime and mtime.
108 *
109 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
110 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
111 */
112 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
113
114 /* Expect random access pattern */
115 #define FMODE_RANDOM ((__force fmode_t)0x1000)
116
117 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
118 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
119
120 /* File is opened with O_PATH; almost nothing can be done with it */
121 #define FMODE_PATH ((__force fmode_t)0x4000)
122
123 /* File was opened by fanotify and shouldn't generate fanotify events */
124 #define FMODE_NONOTIFY ((__force fmode_t)0x1000000)
125
126 /*
127 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
128 * that indicates that they should check the contents of the iovec are
129 * valid, but not check the memory that the iovec elements
130 * points too.
131 */
132 #define CHECK_IOVEC_ONLY -1
133
134 /*
135 * The below are the various read and write types that we support. Some of
136 * them include behavioral modifiers that send information down to the
137 * block layer and IO scheduler. Terminology:
138 *
139 * The block layer uses device plugging to defer IO a little bit, in
140 * the hope that we will see more IO very shortly. This increases
141 * coalescing of adjacent IO and thus reduces the number of IOs we
142 * have to send to the device. It also allows for better queuing,
143 * if the IO isn't mergeable. If the caller is going to be waiting
144 * for the IO, then he must ensure that the device is unplugged so
145 * that the IO is dispatched to the driver.
146 *
147 * All IO is handled async in Linux. This is fine for background
148 * writes, but for reads or writes that someone waits for completion
149 * on, we want to notify the block layer and IO scheduler so that they
150 * know about it. That allows them to make better scheduling
151 * decisions. So when the below references 'sync' and 'async', it
152 * is referencing this priority hint.
153 *
154 * With that in mind, the available types are:
155 *
156 * READ A normal read operation. Device will be plugged.
157 * READ_SYNC A synchronous read. Device is not plugged, caller can
158 * immediately wait on this read without caring about
159 * unplugging.
160 * READA Used for read-ahead operations. Lower priority, and the
161 * block layer could (in theory) choose to ignore this
162 * request if it runs into resource problems.
163 * WRITE A normal async write. Device will be plugged.
164 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
165 * the hint that someone will be waiting on this IO
166 * shortly. The write equivalent of READ_SYNC.
167 * WRITE_ODIRECT Special case write for O_DIRECT only.
168 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
169 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
170 * non-volatile media on completion.
171 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
172 * by a cache flush and data is guaranteed to be on
173 * non-volatile media on completion.
174 *
175 */
176 #define RW_MASK REQ_WRITE
177 #define RWA_MASK REQ_RAHEAD
178
179 #define READ 0
180 #define WRITE RW_MASK
181 #define READA RWA_MASK
182 #define KERNEL_READ (READ|REQ_KERNEL)
183 #define KERNEL_WRITE (WRITE|REQ_KERNEL)
184
185 #define READ_SYNC (READ | REQ_SYNC)
186 #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
187 #define WRITE_ODIRECT (WRITE | REQ_SYNC)
188 #define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
189 #define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
190 #define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
191
192 /*
193 * Attribute flags. These should be or-ed together to figure out what
194 * has been changed!
195 */
196 #define ATTR_MODE (1 << 0)
197 #define ATTR_UID (1 << 1)
198 #define ATTR_GID (1 << 2)
199 #define ATTR_SIZE (1 << 3)
200 #define ATTR_ATIME (1 << 4)
201 #define ATTR_MTIME (1 << 5)
202 #define ATTR_CTIME (1 << 6)
203 #define ATTR_ATIME_SET (1 << 7)
204 #define ATTR_MTIME_SET (1 << 8)
205 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
206 #define ATTR_ATTR_FLAG (1 << 10)
207 #define ATTR_KILL_SUID (1 << 11)
208 #define ATTR_KILL_SGID (1 << 12)
209 #define ATTR_FILE (1 << 13)
210 #define ATTR_KILL_PRIV (1 << 14)
211 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
212 #define ATTR_TIMES_SET (1 << 16)
213
214 /*
215 * This is the Inode Attributes structure, used for notify_change(). It
216 * uses the above definitions as flags, to know which values have changed.
217 * Also, in this manner, a Filesystem can look at only the values it cares
218 * about. Basically, these are the attributes that the VFS layer can
219 * request to change from the FS layer.
220 *
221 * Derek Atkins <warlord@MIT.EDU> 94-10-20
222 */
223 struct iattr {
224 unsigned int ia_valid;
225 umode_t ia_mode;
226 kuid_t ia_uid;
227 kgid_t ia_gid;
228 loff_t ia_size;
229 struct timespec ia_atime;
230 struct timespec ia_mtime;
231 struct timespec ia_ctime;
232
233 /*
234 * Not an attribute, but an auxiliary info for filesystems wanting to
235 * implement an ftruncate() like method. NOTE: filesystem should
236 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
237 */
238 struct file *ia_file;
239 };
240
241 /*
242 * Includes for diskquotas.
243 */
244 #include <linux/quota.h>
245
246 /**
247 * enum positive_aop_returns - aop return codes with specific semantics
248 *
249 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
250 * completed, that the page is still locked, and
251 * should be considered active. The VM uses this hint
252 * to return the page to the active list -- it won't
253 * be a candidate for writeback again in the near
254 * future. Other callers must be careful to unlock
255 * the page if they get this return. Returned by
256 * writepage();
257 *
258 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
259 * unlocked it and the page might have been truncated.
260 * The caller should back up to acquiring a new page and
261 * trying again. The aop will be taking reasonable
262 * precautions not to livelock. If the caller held a page
263 * reference, it should drop it before retrying. Returned
264 * by readpage().
265 *
266 * address_space_operation functions return these large constants to indicate
267 * special semantics to the caller. These are much larger than the bytes in a
268 * page to allow for functions that return the number of bytes operated on in a
269 * given page.
270 */
271
272 enum positive_aop_returns {
273 AOP_WRITEPAGE_ACTIVATE = 0x80000,
274 AOP_TRUNCATED_PAGE = 0x80001,
275 };
276
277 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
278 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
279 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
280 * helper code (eg buffer layer)
281 * to clear GFP_FS from alloc */
282
283 /*
284 * oh the beauties of C type declarations.
285 */
286 struct page;
287 struct address_space;
288 struct writeback_control;
289
290 struct iov_iter {
291 const struct iovec *iov;
292 unsigned long nr_segs;
293 size_t iov_offset;
294 size_t count;
295 };
296
297 size_t iov_iter_copy_from_user_atomic(struct page *page,
298 struct iov_iter *i, unsigned long offset, size_t bytes);
299 size_t iov_iter_copy_from_user(struct page *page,
300 struct iov_iter *i, unsigned long offset, size_t bytes);
301 void iov_iter_advance(struct iov_iter *i, size_t bytes);
302 int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes);
303 size_t iov_iter_single_seg_count(struct iov_iter *i);
304
305 static inline void iov_iter_init(struct iov_iter *i,
306 const struct iovec *iov, unsigned long nr_segs,
307 size_t count, size_t written)
308 {
309 i->iov = iov;
310 i->nr_segs = nr_segs;
311 i->iov_offset = 0;
312 i->count = count + written;
313
314 iov_iter_advance(i, written);
315 }
316
317 static inline size_t iov_iter_count(struct iov_iter *i)
318 {
319 return i->count;
320 }
321
322 /*
323 * "descriptor" for what we're up to with a read.
324 * This allows us to use the same read code yet
325 * have multiple different users of the data that
326 * we read from a file.
327 *
328 * The simplest case just copies the data to user
329 * mode.
330 */
331 typedef struct {
332 size_t written;
333 size_t count;
334 union {
335 char __user *buf;
336 void *data;
337 } arg;
338 int error;
339 } read_descriptor_t;
340
341 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
342 unsigned long, unsigned long);
343
344 struct address_space_operations {
345 int (*writepage)(struct page *page, struct writeback_control *wbc);
346 int (*readpage)(struct file *, struct page *);
347
348 /* Write back some dirty pages from this mapping. */
349 int (*writepages)(struct address_space *, struct writeback_control *);
350
351 /* Set a page dirty. Return true if this dirtied it */
352 int (*set_page_dirty)(struct page *page);
353
354 int (*readpages)(struct file *filp, struct address_space *mapping,
355 struct list_head *pages, unsigned nr_pages);
356
357 int (*write_begin)(struct file *, struct address_space *mapping,
358 loff_t pos, unsigned len, unsigned flags,
359 struct page **pagep, void **fsdata);
360 int (*write_end)(struct file *, struct address_space *mapping,
361 loff_t pos, unsigned len, unsigned copied,
362 struct page *page, void *fsdata);
363
364 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
365 sector_t (*bmap)(struct address_space *, sector_t);
366 void (*invalidatepage) (struct page *, unsigned long);
367 int (*releasepage) (struct page *, gfp_t);
368 void (*freepage)(struct page *);
369 ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
370 loff_t offset, unsigned long nr_segs);
371 int (*get_xip_mem)(struct address_space *, pgoff_t, int,
372 void **, unsigned long *);
373 /*
374 * migrate the contents of a page to the specified target. If sync
375 * is false, it must not block.
376 */
377 int (*migratepage) (struct address_space *,
378 struct page *, struct page *, enum migrate_mode);
379 int (*launder_page) (struct page *);
380 int (*is_partially_uptodate) (struct page *, read_descriptor_t *,
381 unsigned long);
382 int (*error_remove_page)(struct address_space *, struct page *);
383
384 /* swapfile support */
385 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
386 sector_t *span);
387 void (*swap_deactivate)(struct file *file);
388 };
389
390 extern const struct address_space_operations empty_aops;
391
392 /*
393 * pagecache_write_begin/pagecache_write_end must be used by general code
394 * to write into the pagecache.
395 */
396 int pagecache_write_begin(struct file *, struct address_space *mapping,
397 loff_t pos, unsigned len, unsigned flags,
398 struct page **pagep, void **fsdata);
399
400 int pagecache_write_end(struct file *, struct address_space *mapping,
401 loff_t pos, unsigned len, unsigned copied,
402 struct page *page, void *fsdata);
403
404 struct backing_dev_info;
405 struct address_space {
406 struct inode *host; /* owner: inode, block_device */
407 struct radix_tree_root page_tree; /* radix tree of all pages */
408 spinlock_t tree_lock; /* and lock protecting it */
409 unsigned int i_mmap_writable;/* count VM_SHARED mappings */
410 struct rb_root i_mmap; /* tree of private and shared mappings */
411 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
412 struct mutex i_mmap_mutex; /* protect tree, count, list */
413 /* Protected by tree_lock together with the radix tree */
414 unsigned long nrpages; /* number of total pages */
415 pgoff_t writeback_index;/* writeback starts here */
416 const struct address_space_operations *a_ops; /* methods */
417 unsigned long flags; /* error bits/gfp mask */
418 struct backing_dev_info *backing_dev_info; /* device readahead, etc */
419 spinlock_t private_lock; /* for use by the address_space */
420 struct list_head private_list; /* ditto */
421 struct address_space *assoc_mapping; /* ditto */
422 } __attribute__((aligned(sizeof(long))));
423 /*
424 * On most architectures that alignment is already the case; but
425 * must be enforced here for CRIS, to let the least significant bit
426 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
427 */
428 struct request_queue;
429
430 struct block_device {
431 dev_t bd_dev; /* not a kdev_t - it's a search key */
432 int bd_openers;
433 struct inode * bd_inode; /* will die */
434 struct super_block * bd_super;
435 struct mutex bd_mutex; /* open/close mutex */
436 struct list_head bd_inodes;
437 void * bd_claiming;
438 void * bd_holder;
439 int bd_holders;
440 bool bd_write_holder;
441 #ifdef CONFIG_SYSFS
442 struct list_head bd_holder_disks;
443 #endif
444 struct block_device * bd_contains;
445 unsigned bd_block_size;
446 struct hd_struct * bd_part;
447 /* number of times partitions within this device have been opened. */
448 unsigned bd_part_count;
449 int bd_invalidated;
450 struct gendisk * bd_disk;
451 struct request_queue * bd_queue;
452 struct list_head bd_list;
453 /*
454 * Private data. You must have bd_claim'ed the block_device
455 * to use this. NOTE: bd_claim allows an owner to claim
456 * the same device multiple times, the owner must take special
457 * care to not mess up bd_private for that case.
458 */
459 unsigned long bd_private;
460
461 /* The counter of freeze processes */
462 int bd_fsfreeze_count;
463 /* Mutex for freeze */
464 struct mutex bd_fsfreeze_mutex;
465 /* A semaphore that prevents I/O while block size is being changed */
466 struct percpu_rw_semaphore bd_block_size_semaphore;
467 };
468
469 /*
470 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
471 * radix trees
472 */
473 #define PAGECACHE_TAG_DIRTY 0
474 #define PAGECACHE_TAG_WRITEBACK 1
475 #define PAGECACHE_TAG_TOWRITE 2
476
477 int mapping_tagged(struct address_space *mapping, int tag);
478
479 /*
480 * Might pages of this file be mapped into userspace?
481 */
482 static inline int mapping_mapped(struct address_space *mapping)
483 {
484 return !RB_EMPTY_ROOT(&mapping->i_mmap) ||
485 !list_empty(&mapping->i_mmap_nonlinear);
486 }
487
488 /*
489 * Might pages of this file have been modified in userspace?
490 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
491 * marks vma as VM_SHARED if it is shared, and the file was opened for
492 * writing i.e. vma may be mprotected writable even if now readonly.
493 */
494 static inline int mapping_writably_mapped(struct address_space *mapping)
495 {
496 return mapping->i_mmap_writable != 0;
497 }
498
499 /*
500 * Use sequence counter to get consistent i_size on 32-bit processors.
501 */
502 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
503 #include <linux/seqlock.h>
504 #define __NEED_I_SIZE_ORDERED
505 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
506 #else
507 #define i_size_ordered_init(inode) do { } while (0)
508 #endif
509
510 struct posix_acl;
511 #define ACL_NOT_CACHED ((void *)(-1))
512
513 #define IOP_FASTPERM 0x0001
514 #define IOP_LOOKUP 0x0002
515 #define IOP_NOFOLLOW 0x0004
516
517 /*
518 * Keep mostly read-only and often accessed (especially for
519 * the RCU path lookup and 'stat' data) fields at the beginning
520 * of the 'struct inode'
521 */
522 struct inode {
523 umode_t i_mode;
524 unsigned short i_opflags;
525 kuid_t i_uid;
526 kgid_t i_gid;
527 unsigned int i_flags;
528
529 #ifdef CONFIG_FS_POSIX_ACL
530 struct posix_acl *i_acl;
531 struct posix_acl *i_default_acl;
532 #endif
533
534 const struct inode_operations *i_op;
535 struct super_block *i_sb;
536 struct address_space *i_mapping;
537
538 #ifdef CONFIG_SECURITY
539 void *i_security;
540 #endif
541
542 /* Stat data, not accessed from path walking */
543 unsigned long i_ino;
544 /*
545 * Filesystems may only read i_nlink directly. They shall use the
546 * following functions for modification:
547 *
548 * (set|clear|inc|drop)_nlink
549 * inode_(inc|dec)_link_count
550 */
551 union {
552 const unsigned int i_nlink;
553 unsigned int __i_nlink;
554 };
555 dev_t i_rdev;
556 loff_t i_size;
557 struct timespec i_atime;
558 struct timespec i_mtime;
559 struct timespec i_ctime;
560 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
561 unsigned short i_bytes;
562 unsigned int i_blkbits;
563 blkcnt_t i_blocks;
564
565 #ifdef __NEED_I_SIZE_ORDERED
566 seqcount_t i_size_seqcount;
567 #endif
568
569 /* Misc */
570 unsigned long i_state;
571 struct mutex i_mutex;
572
573 unsigned long dirtied_when; /* jiffies of first dirtying */
574
575 struct hlist_node i_hash;
576 struct list_head i_wb_list; /* backing dev IO list */
577 struct list_head i_lru; /* inode LRU list */
578 struct list_head i_sb_list;
579 union {
580 struct hlist_head i_dentry;
581 struct rcu_head i_rcu;
582 };
583 u64 i_version;
584 atomic_t i_count;
585 atomic_t i_dio_count;
586 atomic_t i_writecount;
587 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
588 struct file_lock *i_flock;
589 struct address_space i_data;
590 #ifdef CONFIG_QUOTA
591 struct dquot *i_dquot[MAXQUOTAS];
592 #endif
593 struct list_head i_devices;
594 union {
595 struct pipe_inode_info *i_pipe;
596 struct block_device *i_bdev;
597 struct cdev *i_cdev;
598 };
599
600 __u32 i_generation;
601
602 #ifdef CONFIG_FSNOTIFY
603 __u32 i_fsnotify_mask; /* all events this inode cares about */
604 struct hlist_head i_fsnotify_marks;
605 #endif
606
607 #ifdef CONFIG_IMA
608 atomic_t i_readcount; /* struct files open RO */
609 #endif
610 void *i_private; /* fs or device private pointer */
611 };
612
613 static inline int inode_unhashed(struct inode *inode)
614 {
615 return hlist_unhashed(&inode->i_hash);
616 }
617
618 /*
619 * inode->i_mutex nesting subclasses for the lock validator:
620 *
621 * 0: the object of the current VFS operation
622 * 1: parent
623 * 2: child/target
624 * 3: quota file
625 *
626 * The locking order between these classes is
627 * parent -> child -> normal -> xattr -> quota
628 */
629 enum inode_i_mutex_lock_class
630 {
631 I_MUTEX_NORMAL,
632 I_MUTEX_PARENT,
633 I_MUTEX_CHILD,
634 I_MUTEX_XATTR,
635 I_MUTEX_QUOTA
636 };
637
638 /*
639 * NOTE: in a 32bit arch with a preemptable kernel and
640 * an UP compile the i_size_read/write must be atomic
641 * with respect to the local cpu (unlike with preempt disabled),
642 * but they don't need to be atomic with respect to other cpus like in
643 * true SMP (so they need either to either locally disable irq around
644 * the read or for example on x86 they can be still implemented as a
645 * cmpxchg8b without the need of the lock prefix). For SMP compiles
646 * and 64bit archs it makes no difference if preempt is enabled or not.
647 */
648 static inline loff_t i_size_read(const struct inode *inode)
649 {
650 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
651 loff_t i_size;
652 unsigned int seq;
653
654 do {
655 seq = read_seqcount_begin(&inode->i_size_seqcount);
656 i_size = inode->i_size;
657 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
658 return i_size;
659 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
660 loff_t i_size;
661
662 preempt_disable();
663 i_size = inode->i_size;
664 preempt_enable();
665 return i_size;
666 #else
667 return inode->i_size;
668 #endif
669 }
670
671 /*
672 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
673 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
674 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
675 */
676 static inline void i_size_write(struct inode *inode, loff_t i_size)
677 {
678 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
679 write_seqcount_begin(&inode->i_size_seqcount);
680 inode->i_size = i_size;
681 write_seqcount_end(&inode->i_size_seqcount);
682 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
683 preempt_disable();
684 inode->i_size = i_size;
685 preempt_enable();
686 #else
687 inode->i_size = i_size;
688 #endif
689 }
690
691 /* Helper functions so that in most cases filesystems will
692 * not need to deal directly with kuid_t and kgid_t and can
693 * instead deal with the raw numeric values that are stored
694 * in the filesystem.
695 */
696 static inline uid_t i_uid_read(const struct inode *inode)
697 {
698 return from_kuid(&init_user_ns, inode->i_uid);
699 }
700
701 static inline gid_t i_gid_read(const struct inode *inode)
702 {
703 return from_kgid(&init_user_ns, inode->i_gid);
704 }
705
706 static inline void i_uid_write(struct inode *inode, uid_t uid)
707 {
708 inode->i_uid = make_kuid(&init_user_ns, uid);
709 }
710
711 static inline void i_gid_write(struct inode *inode, gid_t gid)
712 {
713 inode->i_gid = make_kgid(&init_user_ns, gid);
714 }
715
716 static inline unsigned iminor(const struct inode *inode)
717 {
718 return MINOR(inode->i_rdev);
719 }
720
721 static inline unsigned imajor(const struct inode *inode)
722 {
723 return MAJOR(inode->i_rdev);
724 }
725
726 extern struct block_device *I_BDEV(struct inode *inode);
727
728 struct fown_struct {
729 rwlock_t lock; /* protects pid, uid, euid fields */
730 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
731 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
732 kuid_t uid, euid; /* uid/euid of process setting the owner */
733 int signum; /* posix.1b rt signal to be delivered on IO */
734 };
735
736 /*
737 * Track a single file's readahead state
738 */
739 struct file_ra_state {
740 pgoff_t start; /* where readahead started */
741 unsigned int size; /* # of readahead pages */
742 unsigned int async_size; /* do asynchronous readahead when
743 there are only # of pages ahead */
744
745 unsigned int ra_pages; /* Maximum readahead window */
746 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
747 loff_t prev_pos; /* Cache last read() position */
748 };
749
750 /*
751 * Check if @index falls in the readahead windows.
752 */
753 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
754 {
755 return (index >= ra->start &&
756 index < ra->start + ra->size);
757 }
758
759 #define FILE_MNT_WRITE_TAKEN 1
760 #define FILE_MNT_WRITE_RELEASED 2
761
762 struct file {
763 /*
764 * fu_list becomes invalid after file_free is called and queued via
765 * fu_rcuhead for RCU freeing
766 */
767 union {
768 struct list_head fu_list;
769 struct rcu_head fu_rcuhead;
770 } f_u;
771 struct path f_path;
772 #define f_dentry f_path.dentry
773 #define f_vfsmnt f_path.mnt
774 const struct file_operations *f_op;
775
776 /*
777 * Protects f_ep_links, f_flags, f_pos vs i_size in lseek SEEK_CUR.
778 * Must not be taken from IRQ context.
779 */
780 spinlock_t f_lock;
781 #ifdef CONFIG_SMP
782 int f_sb_list_cpu;
783 #endif
784 atomic_long_t f_count;
785 unsigned int f_flags;
786 fmode_t f_mode;
787 loff_t f_pos;
788 struct fown_struct f_owner;
789 const struct cred *f_cred;
790 struct file_ra_state f_ra;
791
792 u64 f_version;
793 #ifdef CONFIG_SECURITY
794 void *f_security;
795 #endif
796 /* needed for tty driver, and maybe others */
797 void *private_data;
798
799 #ifdef CONFIG_EPOLL
800 /* Used by fs/eventpoll.c to link all the hooks to this file */
801 struct list_head f_ep_links;
802 struct list_head f_tfile_llink;
803 #endif /* #ifdef CONFIG_EPOLL */
804 struct address_space *f_mapping;
805 #ifdef CONFIG_DEBUG_WRITECOUNT
806 unsigned long f_mnt_write_state;
807 #endif
808 };
809
810 struct file_handle {
811 __u32 handle_bytes;
812 int handle_type;
813 /* file identifier */
814 unsigned char f_handle[0];
815 };
816
817 static inline struct file *get_file(struct file *f)
818 {
819 atomic_long_inc(&f->f_count);
820 return f;
821 }
822 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
823 #define file_count(x) atomic_long_read(&(x)->f_count)
824
825 #ifdef CONFIG_DEBUG_WRITECOUNT
826 static inline void file_take_write(struct file *f)
827 {
828 WARN_ON(f->f_mnt_write_state != 0);
829 f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN;
830 }
831 static inline void file_release_write(struct file *f)
832 {
833 f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED;
834 }
835 static inline void file_reset_write(struct file *f)
836 {
837 f->f_mnt_write_state = 0;
838 }
839 static inline void file_check_state(struct file *f)
840 {
841 /*
842 * At this point, either both or neither of these bits
843 * should be set.
844 */
845 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN);
846 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED);
847 }
848 static inline int file_check_writeable(struct file *f)
849 {
850 if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN)
851 return 0;
852 printk(KERN_WARNING "writeable file with no "
853 "mnt_want_write()\n");
854 WARN_ON(1);
855 return -EINVAL;
856 }
857 #else /* !CONFIG_DEBUG_WRITECOUNT */
858 static inline void file_take_write(struct file *filp) {}
859 static inline void file_release_write(struct file *filp) {}
860 static inline void file_reset_write(struct file *filp) {}
861 static inline void file_check_state(struct file *filp) {}
862 static inline int file_check_writeable(struct file *filp)
863 {
864 return 0;
865 }
866 #endif /* CONFIG_DEBUG_WRITECOUNT */
867
868 #define MAX_NON_LFS ((1UL<<31) - 1)
869
870 /* Page cache limit. The filesystems should put that into their s_maxbytes
871 limits, otherwise bad things can happen in VM. */
872 #if BITS_PER_LONG==32
873 #define MAX_LFS_FILESIZE (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
874 #elif BITS_PER_LONG==64
875 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
876 #endif
877
878 #define FL_POSIX 1
879 #define FL_FLOCK 2
880 #define FL_ACCESS 8 /* not trying to lock, just looking */
881 #define FL_EXISTS 16 /* when unlocking, test for existence */
882 #define FL_LEASE 32 /* lease held on this file */
883 #define FL_CLOSE 64 /* unlock on close */
884 #define FL_SLEEP 128 /* A blocking lock */
885 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
886 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
887
888 /*
889 * Special return value from posix_lock_file() and vfs_lock_file() for
890 * asynchronous locking.
891 */
892 #define FILE_LOCK_DEFERRED 1
893
894 /*
895 * The POSIX file lock owner is determined by
896 * the "struct files_struct" in the thread group
897 * (or NULL for no owner - BSD locks).
898 *
899 * Lockd stuffs a "host" pointer into this.
900 */
901 typedef struct files_struct *fl_owner_t;
902
903 struct file_lock_operations {
904 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
905 void (*fl_release_private)(struct file_lock *);
906 };
907
908 struct lock_manager_operations {
909 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
910 void (*lm_notify)(struct file_lock *); /* unblock callback */
911 int (*lm_grant)(struct file_lock *, struct file_lock *, int);
912 void (*lm_break)(struct file_lock *);
913 int (*lm_change)(struct file_lock **, int);
914 };
915
916 struct lock_manager {
917 struct list_head list;
918 };
919
920 struct net;
921 void locks_start_grace(struct net *, struct lock_manager *);
922 void locks_end_grace(struct lock_manager *);
923 int locks_in_grace(struct net *);
924
925 /* that will die - we need it for nfs_lock_info */
926 #include <linux/nfs_fs_i.h>
927
928 struct file_lock {
929 struct file_lock *fl_next; /* singly linked list for this inode */
930 struct list_head fl_link; /* doubly linked list of all locks */
931 struct list_head fl_block; /* circular list of blocked processes */
932 fl_owner_t fl_owner;
933 unsigned int fl_flags;
934 unsigned char fl_type;
935 unsigned int fl_pid;
936 struct pid *fl_nspid;
937 wait_queue_head_t fl_wait;
938 struct file *fl_file;
939 loff_t fl_start;
940 loff_t fl_end;
941
942 struct fasync_struct * fl_fasync; /* for lease break notifications */
943 /* for lease breaks: */
944 unsigned long fl_break_time;
945 unsigned long fl_downgrade_time;
946
947 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
948 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
949 union {
950 struct nfs_lock_info nfs_fl;
951 struct nfs4_lock_info nfs4_fl;
952 struct {
953 struct list_head link; /* link in AFS vnode's pending_locks list */
954 int state; /* state of grant or error if -ve */
955 } afs;
956 } fl_u;
957 };
958
959 /* The following constant reflects the upper bound of the file/locking space */
960 #ifndef OFFSET_MAX
961 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
962 #define OFFSET_MAX INT_LIMIT(loff_t)
963 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
964 #endif
965
966 #include <linux/fcntl.h>
967
968 extern void send_sigio(struct fown_struct *fown, int fd, int band);
969
970 #ifdef CONFIG_FILE_LOCKING
971 extern int fcntl_getlk(struct file *, struct flock __user *);
972 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
973 struct flock __user *);
974
975 #if BITS_PER_LONG == 32
976 extern int fcntl_getlk64(struct file *, struct flock64 __user *);
977 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
978 struct flock64 __user *);
979 #endif
980
981 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
982 extern int fcntl_getlease(struct file *filp);
983
984 /* fs/locks.c */
985 void locks_free_lock(struct file_lock *fl);
986 extern void locks_init_lock(struct file_lock *);
987 extern struct file_lock * locks_alloc_lock(void);
988 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
989 extern void __locks_copy_lock(struct file_lock *, const struct file_lock *);
990 extern void locks_remove_posix(struct file *, fl_owner_t);
991 extern void locks_remove_flock(struct file *);
992 extern void locks_release_private(struct file_lock *);
993 extern void posix_test_lock(struct file *, struct file_lock *);
994 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
995 extern int posix_lock_file_wait(struct file *, struct file_lock *);
996 extern int posix_unblock_lock(struct file *, struct file_lock *);
997 extern int vfs_test_lock(struct file *, struct file_lock *);
998 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
999 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1000 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
1001 extern int __break_lease(struct inode *inode, unsigned int flags);
1002 extern void lease_get_mtime(struct inode *, struct timespec *time);
1003 extern int generic_setlease(struct file *, long, struct file_lock **);
1004 extern int vfs_setlease(struct file *, long, struct file_lock **);
1005 extern int lease_modify(struct file_lock **, int);
1006 extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
1007 extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
1008 extern void locks_delete_block(struct file_lock *waiter);
1009 extern void lock_flocks(void);
1010 extern void unlock_flocks(void);
1011 #else /* !CONFIG_FILE_LOCKING */
1012 static inline int fcntl_getlk(struct file *file, struct flock __user *user)
1013 {
1014 return -EINVAL;
1015 }
1016
1017 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1018 unsigned int cmd, struct flock __user *user)
1019 {
1020 return -EACCES;
1021 }
1022
1023 #if BITS_PER_LONG == 32
1024 static inline int fcntl_getlk64(struct file *file, struct flock64 __user *user)
1025 {
1026 return -EINVAL;
1027 }
1028
1029 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1030 unsigned int cmd, struct flock64 __user *user)
1031 {
1032 return -EACCES;
1033 }
1034 #endif
1035 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1036 {
1037 return 0;
1038 }
1039
1040 static inline int fcntl_getlease(struct file *filp)
1041 {
1042 return 0;
1043 }
1044
1045 static inline void locks_init_lock(struct file_lock *fl)
1046 {
1047 return;
1048 }
1049
1050 static inline void __locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1051 {
1052 return;
1053 }
1054
1055 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1056 {
1057 return;
1058 }
1059
1060 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1061 {
1062 return;
1063 }
1064
1065 static inline void locks_remove_flock(struct file *filp)
1066 {
1067 return;
1068 }
1069
1070 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1071 {
1072 return;
1073 }
1074
1075 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1076 struct file_lock *conflock)
1077 {
1078 return -ENOLCK;
1079 }
1080
1081 static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1082 {
1083 return -ENOLCK;
1084 }
1085
1086 static inline int posix_unblock_lock(struct file *filp,
1087 struct file_lock *waiter)
1088 {
1089 return -ENOENT;
1090 }
1091
1092 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1093 {
1094 return 0;
1095 }
1096
1097 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1098 struct file_lock *fl, struct file_lock *conf)
1099 {
1100 return -ENOLCK;
1101 }
1102
1103 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1104 {
1105 return 0;
1106 }
1107
1108 static inline int flock_lock_file_wait(struct file *filp,
1109 struct file_lock *request)
1110 {
1111 return -ENOLCK;
1112 }
1113
1114 static inline int __break_lease(struct inode *inode, unsigned int mode)
1115 {
1116 return 0;
1117 }
1118
1119 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1120 {
1121 return;
1122 }
1123
1124 static inline int generic_setlease(struct file *filp, long arg,
1125 struct file_lock **flp)
1126 {
1127 return -EINVAL;
1128 }
1129
1130 static inline int vfs_setlease(struct file *filp, long arg,
1131 struct file_lock **lease)
1132 {
1133 return -EINVAL;
1134 }
1135
1136 static inline int lease_modify(struct file_lock **before, int arg)
1137 {
1138 return -EINVAL;
1139 }
1140
1141 static inline int lock_may_read(struct inode *inode, loff_t start,
1142 unsigned long len)
1143 {
1144 return 1;
1145 }
1146
1147 static inline int lock_may_write(struct inode *inode, loff_t start,
1148 unsigned long len)
1149 {
1150 return 1;
1151 }
1152
1153 static inline void locks_delete_block(struct file_lock *waiter)
1154 {
1155 }
1156
1157 static inline void lock_flocks(void)
1158 {
1159 }
1160
1161 static inline void unlock_flocks(void)
1162 {
1163 }
1164
1165 #endif /* !CONFIG_FILE_LOCKING */
1166
1167
1168 struct fasync_struct {
1169 spinlock_t fa_lock;
1170 int magic;
1171 int fa_fd;
1172 struct fasync_struct *fa_next; /* singly linked list */
1173 struct file *fa_file;
1174 struct rcu_head fa_rcu;
1175 };
1176
1177 #define FASYNC_MAGIC 0x4601
1178
1179 /* SMP safe fasync helpers: */
1180 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1181 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1182 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1183 extern struct fasync_struct *fasync_alloc(void);
1184 extern void fasync_free(struct fasync_struct *);
1185
1186 /* can be called from interrupts */
1187 extern void kill_fasync(struct fasync_struct **, int, int);
1188
1189 extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1190 extern int f_setown(struct file *filp, unsigned long arg, int force);
1191 extern void f_delown(struct file *filp);
1192 extern pid_t f_getown(struct file *filp);
1193 extern int send_sigurg(struct fown_struct *fown);
1194
1195 struct mm_struct;
1196
1197 /*
1198 * Umount options
1199 */
1200
1201 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1202 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1203 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1204 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1205 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1206
1207 extern struct list_head super_blocks;
1208 extern spinlock_t sb_lock;
1209
1210 /* Possible states of 'frozen' field */
1211 enum {
1212 SB_UNFROZEN = 0, /* FS is unfrozen */
1213 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1214 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1215 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1216 * internal threads if needed) */
1217 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1218 };
1219
1220 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1221
1222 struct sb_writers {
1223 /* Counters for counting writers at each level */
1224 struct percpu_counter counter[SB_FREEZE_LEVELS];
1225 wait_queue_head_t wait; /* queue for waiting for
1226 writers / faults to finish */
1227 int frozen; /* Is sb frozen? */
1228 wait_queue_head_t wait_unfrozen; /* queue for waiting for
1229 sb to be thawed */
1230 #ifdef CONFIG_DEBUG_LOCK_ALLOC
1231 struct lockdep_map lock_map[SB_FREEZE_LEVELS];
1232 #endif
1233 };
1234
1235 struct super_block {
1236 struct list_head s_list; /* Keep this first */
1237 dev_t s_dev; /* search index; _not_ kdev_t */
1238 unsigned char s_blocksize_bits;
1239 unsigned long s_blocksize;
1240 loff_t s_maxbytes; /* Max file size */
1241 struct file_system_type *s_type;
1242 const struct super_operations *s_op;
1243 const struct dquot_operations *dq_op;
1244 const struct quotactl_ops *s_qcop;
1245 const struct export_operations *s_export_op;
1246 unsigned long s_flags;
1247 unsigned long s_magic;
1248 struct dentry *s_root;
1249 struct rw_semaphore s_umount;
1250 int s_count;
1251 atomic_t s_active;
1252 #ifdef CONFIG_SECURITY
1253 void *s_security;
1254 #endif
1255 const struct xattr_handler **s_xattr;
1256
1257 struct list_head s_inodes; /* all inodes */
1258 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1259 #ifdef CONFIG_SMP
1260 struct list_head __percpu *s_files;
1261 #else
1262 struct list_head s_files;
1263 #endif
1264 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1265 /* s_dentry_lru, s_nr_dentry_unused protected by dcache.c lru locks */
1266 struct list_head s_dentry_lru; /* unused dentry lru */
1267 int s_nr_dentry_unused; /* # of dentry on lru */
1268
1269 /* s_inode_lru_lock protects s_inode_lru and s_nr_inodes_unused */
1270 spinlock_t s_inode_lru_lock ____cacheline_aligned_in_smp;
1271 struct list_head s_inode_lru; /* unused inode lru */
1272 int s_nr_inodes_unused; /* # of inodes on lru */
1273
1274 struct block_device *s_bdev;
1275 struct backing_dev_info *s_bdi;
1276 struct mtd_info *s_mtd;
1277 struct hlist_node s_instances;
1278 struct quota_info s_dquot; /* Diskquota specific options */
1279
1280 struct sb_writers s_writers;
1281
1282 char s_id[32]; /* Informational name */
1283 u8 s_uuid[16]; /* UUID */
1284
1285 void *s_fs_info; /* Filesystem private info */
1286 unsigned int s_max_links;
1287 fmode_t s_mode;
1288
1289 /* Granularity of c/m/atime in ns.
1290 Cannot be worse than a second */
1291 u32 s_time_gran;
1292
1293 /*
1294 * The next field is for VFS *only*. No filesystems have any business
1295 * even looking at it. You had been warned.
1296 */
1297 struct mutex s_vfs_rename_mutex; /* Kludge */
1298
1299 /*
1300 * Filesystem subtype. If non-empty the filesystem type field
1301 * in /proc/mounts will be "type.subtype"
1302 */
1303 char *s_subtype;
1304
1305 /*
1306 * Saved mount options for lazy filesystems using
1307 * generic_show_options()
1308 */
1309 char __rcu *s_options;
1310 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1311
1312 /*
1313 * Saved pool identifier for cleancache (-1 means none)
1314 */
1315 int cleancache_poolid;
1316
1317 struct shrinker s_shrink; /* per-sb shrinker handle */
1318
1319 /* Number of inodes with nlink == 0 but still referenced */
1320 atomic_long_t s_remove_count;
1321
1322 /* Being remounted read-only */
1323 int s_readonly_remount;
1324 };
1325
1326 /* superblock cache pruning functions */
1327 extern void prune_icache_sb(struct super_block *sb, int nr_to_scan);
1328 extern void prune_dcache_sb(struct super_block *sb, int nr_to_scan);
1329
1330 extern struct timespec current_fs_time(struct super_block *sb);
1331
1332 /*
1333 * Snapshotting support.
1334 */
1335
1336 void __sb_end_write(struct super_block *sb, int level);
1337 int __sb_start_write(struct super_block *sb, int level, bool wait);
1338
1339 /**
1340 * sb_end_write - drop write access to a superblock
1341 * @sb: the super we wrote to
1342 *
1343 * Decrement number of writers to the filesystem. Wake up possible waiters
1344 * wanting to freeze the filesystem.
1345 */
1346 static inline void sb_end_write(struct super_block *sb)
1347 {
1348 __sb_end_write(sb, SB_FREEZE_WRITE);
1349 }
1350
1351 /**
1352 * sb_end_pagefault - drop write access to a superblock from a page fault
1353 * @sb: the super we wrote to
1354 *
1355 * Decrement number of processes handling write page fault to the filesystem.
1356 * Wake up possible waiters wanting to freeze the filesystem.
1357 */
1358 static inline void sb_end_pagefault(struct super_block *sb)
1359 {
1360 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1361 }
1362
1363 /**
1364 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1365 * @sb: the super we wrote to
1366 *
1367 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1368 * waiters wanting to freeze the filesystem.
1369 */
1370 static inline void sb_end_intwrite(struct super_block *sb)
1371 {
1372 __sb_end_write(sb, SB_FREEZE_FS);
1373 }
1374
1375 /**
1376 * sb_start_write - get write access to a superblock
1377 * @sb: the super we write to
1378 *
1379 * When a process wants to write data or metadata to a file system (i.e. dirty
1380 * a page or an inode), it should embed the operation in a sb_start_write() -
1381 * sb_end_write() pair to get exclusion against file system freezing. This
1382 * function increments number of writers preventing freezing. If the file
1383 * system is already frozen, the function waits until the file system is
1384 * thawed.
1385 *
1386 * Since freeze protection behaves as a lock, users have to preserve
1387 * ordering of freeze protection and other filesystem locks. Generally,
1388 * freeze protection should be the outermost lock. In particular, we have:
1389 *
1390 * sb_start_write
1391 * -> i_mutex (write path, truncate, directory ops, ...)
1392 * -> s_umount (freeze_super, thaw_super)
1393 */
1394 static inline void sb_start_write(struct super_block *sb)
1395 {
1396 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1397 }
1398
1399 static inline int sb_start_write_trylock(struct super_block *sb)
1400 {
1401 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1402 }
1403
1404 /**
1405 * sb_start_pagefault - get write access to a superblock from a page fault
1406 * @sb: the super we write to
1407 *
1408 * When a process starts handling write page fault, it should embed the
1409 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1410 * exclusion against file system freezing. This is needed since the page fault
1411 * is going to dirty a page. This function increments number of running page
1412 * faults preventing freezing. If the file system is already frozen, the
1413 * function waits until the file system is thawed.
1414 *
1415 * Since page fault freeze protection behaves as a lock, users have to preserve
1416 * ordering of freeze protection and other filesystem locks. It is advised to
1417 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1418 * handling code implies lock dependency:
1419 *
1420 * mmap_sem
1421 * -> sb_start_pagefault
1422 */
1423 static inline void sb_start_pagefault(struct super_block *sb)
1424 {
1425 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1426 }
1427
1428 /*
1429 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1430 * @sb: the super we write to
1431 *
1432 * This is the third level of protection against filesystem freezing. It is
1433 * free for use by a filesystem. The only requirement is that it must rank
1434 * below sb_start_pagefault.
1435 *
1436 * For example filesystem can call sb_start_intwrite() when starting a
1437 * transaction which somewhat eases handling of freezing for internal sources
1438 * of filesystem changes (internal fs threads, discarding preallocation on file
1439 * close, etc.).
1440 */
1441 static inline void sb_start_intwrite(struct super_block *sb)
1442 {
1443 __sb_start_write(sb, SB_FREEZE_FS, true);
1444 }
1445
1446
1447 extern bool inode_owner_or_capable(const struct inode *inode);
1448
1449 /* not quite ready to be deprecated, but... */
1450 extern void lock_super(struct super_block *);
1451 extern void unlock_super(struct super_block *);
1452
1453 /*
1454 * VFS helper functions..
1455 */
1456 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1457 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1458 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1459 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1460 extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
1461 extern int vfs_rmdir(struct inode *, struct dentry *);
1462 extern int vfs_unlink(struct inode *, struct dentry *);
1463 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1464
1465 /*
1466 * VFS dentry helper functions.
1467 */
1468 extern void dentry_unhash(struct dentry *dentry);
1469
1470 /*
1471 * VFS file helper functions.
1472 */
1473 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1474 umode_t mode);
1475 /*
1476 * VFS FS_IOC_FIEMAP helper definitions.
1477 */
1478 struct fiemap_extent_info {
1479 unsigned int fi_flags; /* Flags as passed from user */
1480 unsigned int fi_extents_mapped; /* Number of mapped extents */
1481 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1482 struct fiemap_extent __user *fi_extents_start; /* Start of
1483 fiemap_extent array */
1484 };
1485 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1486 u64 phys, u64 len, u32 flags);
1487 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1488
1489 /*
1490 * File types
1491 *
1492 * NOTE! These match bits 12..15 of stat.st_mode
1493 * (ie "(i_mode >> 12) & 15").
1494 */
1495 #define DT_UNKNOWN 0
1496 #define DT_FIFO 1
1497 #define DT_CHR 2
1498 #define DT_DIR 4
1499 #define DT_BLK 6
1500 #define DT_REG 8
1501 #define DT_LNK 10
1502 #define DT_SOCK 12
1503 #define DT_WHT 14
1504
1505 /*
1506 * This is the "filldir" function type, used by readdir() to let
1507 * the kernel specify what kind of dirent layout it wants to have.
1508 * This allows the kernel to read directories into kernel space or
1509 * to have different dirent layouts depending on the binary type.
1510 */
1511 typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
1512 struct block_device_operations;
1513
1514 /* These macros are for out of kernel modules to test that
1515 * the kernel supports the unlocked_ioctl and compat_ioctl
1516 * fields in struct file_operations. */
1517 #define HAVE_COMPAT_IOCTL 1
1518 #define HAVE_UNLOCKED_IOCTL 1
1519
1520 struct file_operations {
1521 struct module *owner;
1522 loff_t (*llseek) (struct file *, loff_t, int);
1523 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1524 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1525 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1526 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1527 int (*readdir) (struct file *, void *, filldir_t);
1528 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1529 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1530 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1531 int (*mmap) (struct file *, struct vm_area_struct *);
1532 int (*open) (struct inode *, struct file *);
1533 int (*flush) (struct file *, fl_owner_t id);
1534 int (*release) (struct inode *, struct file *);
1535 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1536 int (*aio_fsync) (struct kiocb *, int datasync);
1537 int (*fasync) (int, struct file *, int);
1538 int (*lock) (struct file *, int, struct file_lock *);
1539 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1540 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1541 int (*check_flags)(int);
1542 int (*flock) (struct file *, int, struct file_lock *);
1543 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1544 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1545 int (*setlease)(struct file *, long, struct file_lock **);
1546 long (*fallocate)(struct file *file, int mode, loff_t offset,
1547 loff_t len);
1548 };
1549
1550 struct inode_operations {
1551 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1552 void * (*follow_link) (struct dentry *, struct nameidata *);
1553 int (*permission) (struct inode *, int);
1554 struct posix_acl * (*get_acl)(struct inode *, int);
1555
1556 int (*readlink) (struct dentry *, char __user *,int);
1557 void (*put_link) (struct dentry *, struct nameidata *, void *);
1558
1559 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1560 int (*link) (struct dentry *,struct inode *,struct dentry *);
1561 int (*unlink) (struct inode *,struct dentry *);
1562 int (*symlink) (struct inode *,struct dentry *,const char *);
1563 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1564 int (*rmdir) (struct inode *,struct dentry *);
1565 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1566 int (*rename) (struct inode *, struct dentry *,
1567 struct inode *, struct dentry *);
1568 void (*truncate) (struct inode *);
1569 int (*setattr) (struct dentry *, struct iattr *);
1570 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1571 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1572 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1573 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1574 int (*removexattr) (struct dentry *, const char *);
1575 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1576 u64 len);
1577 int (*update_time)(struct inode *, struct timespec *, int);
1578 int (*atomic_open)(struct inode *, struct dentry *,
1579 struct file *, unsigned open_flag,
1580 umode_t create_mode, int *opened);
1581 } ____cacheline_aligned;
1582
1583 struct seq_file;
1584
1585 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1586 unsigned long nr_segs, unsigned long fast_segs,
1587 struct iovec *fast_pointer,
1588 struct iovec **ret_pointer);
1589
1590 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1591 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1592 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1593 unsigned long, loff_t *);
1594 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1595 unsigned long, loff_t *);
1596
1597 struct super_operations {
1598 struct inode *(*alloc_inode)(struct super_block *sb);
1599 void (*destroy_inode)(struct inode *);
1600
1601 void (*dirty_inode) (struct inode *, int flags);
1602 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1603 int (*drop_inode) (struct inode *);
1604 void (*evict_inode) (struct inode *);
1605 void (*put_super) (struct super_block *);
1606 int (*sync_fs)(struct super_block *sb, int wait);
1607 int (*freeze_fs) (struct super_block *);
1608 int (*unfreeze_fs) (struct super_block *);
1609 int (*statfs) (struct dentry *, struct kstatfs *);
1610 int (*remount_fs) (struct super_block *, int *, char *);
1611 void (*umount_begin) (struct super_block *);
1612
1613 int (*show_options)(struct seq_file *, struct dentry *);
1614 int (*show_devname)(struct seq_file *, struct dentry *);
1615 int (*show_path)(struct seq_file *, struct dentry *);
1616 int (*show_stats)(struct seq_file *, struct dentry *);
1617 #ifdef CONFIG_QUOTA
1618 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1619 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1620 #endif
1621 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1622 int (*nr_cached_objects)(struct super_block *);
1623 void (*free_cached_objects)(struct super_block *, int);
1624 };
1625
1626 /*
1627 * Inode flags - they have no relation to superblock flags now
1628 */
1629 #define S_SYNC 1 /* Writes are synced at once */
1630 #define S_NOATIME 2 /* Do not update access times */
1631 #define S_APPEND 4 /* Append-only file */
1632 #define S_IMMUTABLE 8 /* Immutable file */
1633 #define S_DEAD 16 /* removed, but still open directory */
1634 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1635 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1636 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1637 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1638 #define S_PRIVATE 512 /* Inode is fs-internal */
1639 #define S_IMA 1024 /* Inode has an associated IMA struct */
1640 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1641 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1642
1643 /*
1644 * Note that nosuid etc flags are inode-specific: setting some file-system
1645 * flags just means all the inodes inherit those flags by default. It might be
1646 * possible to override it selectively if you really wanted to with some
1647 * ioctl() that is not currently implemented.
1648 *
1649 * Exception: MS_RDONLY is always applied to the entire file system.
1650 *
1651 * Unfortunately, it is possible to change a filesystems flags with it mounted
1652 * with files in use. This means that all of the inodes will not have their
1653 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1654 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1655 */
1656 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1657
1658 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1659 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1660 ((inode)->i_flags & S_SYNC))
1661 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1662 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1663 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1664 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1665 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1666
1667 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1668 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1669 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1670 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1671
1672 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1673 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1674 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1675 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1676 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1677 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1678 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1679
1680 /*
1681 * Inode state bits. Protected by inode->i_lock
1682 *
1683 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1684 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1685 *
1686 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1687 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1688 * various stages of removing an inode.
1689 *
1690 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1691 *
1692 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1693 * fdatasync(). i_atime is the usual cause.
1694 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1695 * these changes separately from I_DIRTY_SYNC so that we
1696 * don't have to write inode on fdatasync() when only
1697 * mtime has changed in it.
1698 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1699 * I_NEW Serves as both a mutex and completion notification.
1700 * New inodes set I_NEW. If two processes both create
1701 * the same inode, one of them will release its inode and
1702 * wait for I_NEW to be released before returning.
1703 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1704 * also cause waiting on I_NEW, without I_NEW actually
1705 * being set. find_inode() uses this to prevent returning
1706 * nearly-dead inodes.
1707 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1708 * is zero. I_FREEING must be set when I_WILL_FREE is
1709 * cleared.
1710 * I_FREEING Set when inode is about to be freed but still has dirty
1711 * pages or buffers attached or the inode itself is still
1712 * dirty.
1713 * I_CLEAR Added by clear_inode(). In this state the inode is
1714 * clean and can be destroyed. Inode keeps I_FREEING.
1715 *
1716 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1717 * prohibited for many purposes. iget() must wait for
1718 * the inode to be completely released, then create it
1719 * anew. Other functions will just ignore such inodes,
1720 * if appropriate. I_NEW is used for waiting.
1721 *
1722 * I_SYNC Writeback of inode is running. The bit is set during
1723 * data writeback, and cleared with a wakeup on the bit
1724 * address once it is done. The bit is also used to pin
1725 * the inode in memory for flusher thread.
1726 *
1727 * I_REFERENCED Marks the inode as recently references on the LRU list.
1728 *
1729 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1730 *
1731 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1732 */
1733 #define I_DIRTY_SYNC (1 << 0)
1734 #define I_DIRTY_DATASYNC (1 << 1)
1735 #define I_DIRTY_PAGES (1 << 2)
1736 #define __I_NEW 3
1737 #define I_NEW (1 << __I_NEW)
1738 #define I_WILL_FREE (1 << 4)
1739 #define I_FREEING (1 << 5)
1740 #define I_CLEAR (1 << 6)
1741 #define __I_SYNC 7
1742 #define I_SYNC (1 << __I_SYNC)
1743 #define I_REFERENCED (1 << 8)
1744 #define __I_DIO_WAKEUP 9
1745 #define I_DIO_WAKEUP (1 << I_DIO_WAKEUP)
1746
1747 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1748
1749 extern void __mark_inode_dirty(struct inode *, int);
1750 static inline void mark_inode_dirty(struct inode *inode)
1751 {
1752 __mark_inode_dirty(inode, I_DIRTY);
1753 }
1754
1755 static inline void mark_inode_dirty_sync(struct inode *inode)
1756 {
1757 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1758 }
1759
1760 extern void inc_nlink(struct inode *inode);
1761 extern void drop_nlink(struct inode *inode);
1762 extern void clear_nlink(struct inode *inode);
1763 extern void set_nlink(struct inode *inode, unsigned int nlink);
1764
1765 static inline void inode_inc_link_count(struct inode *inode)
1766 {
1767 inc_nlink(inode);
1768 mark_inode_dirty(inode);
1769 }
1770
1771 static inline void inode_dec_link_count(struct inode *inode)
1772 {
1773 drop_nlink(inode);
1774 mark_inode_dirty(inode);
1775 }
1776
1777 /**
1778 * inode_inc_iversion - increments i_version
1779 * @inode: inode that need to be updated
1780 *
1781 * Every time the inode is modified, the i_version field will be incremented.
1782 * The filesystem has to be mounted with i_version flag
1783 */
1784
1785 static inline void inode_inc_iversion(struct inode *inode)
1786 {
1787 spin_lock(&inode->i_lock);
1788 inode->i_version++;
1789 spin_unlock(&inode->i_lock);
1790 }
1791
1792 enum file_time_flags {
1793 S_ATIME = 1,
1794 S_MTIME = 2,
1795 S_CTIME = 4,
1796 S_VERSION = 8,
1797 };
1798
1799 extern void touch_atime(struct path *);
1800 static inline void file_accessed(struct file *file)
1801 {
1802 if (!(file->f_flags & O_NOATIME))
1803 touch_atime(&file->f_path);
1804 }
1805
1806 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1807 int sync_inode_metadata(struct inode *inode, int wait);
1808
1809 struct file_system_type {
1810 const char *name;
1811 int fs_flags;
1812 #define FS_REQUIRES_DEV 1
1813 #define FS_BINARY_MOUNTDATA 2
1814 #define FS_HAS_SUBTYPE 4
1815 #define FS_REVAL_DOT 16384 /* Check the paths ".", ".." for staleness */
1816 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
1817 struct dentry *(*mount) (struct file_system_type *, int,
1818 const char *, void *);
1819 void (*kill_sb) (struct super_block *);
1820 struct module *owner;
1821 struct file_system_type * next;
1822 struct hlist_head fs_supers;
1823
1824 struct lock_class_key s_lock_key;
1825 struct lock_class_key s_umount_key;
1826 struct lock_class_key s_vfs_rename_key;
1827 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
1828
1829 struct lock_class_key i_lock_key;
1830 struct lock_class_key i_mutex_key;
1831 struct lock_class_key i_mutex_dir_key;
1832 };
1833
1834 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1835 void *data, int (*fill_super)(struct super_block *, void *, int));
1836 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1837 int flags, const char *dev_name, void *data,
1838 int (*fill_super)(struct super_block *, void *, int));
1839 extern struct dentry *mount_single(struct file_system_type *fs_type,
1840 int flags, void *data,
1841 int (*fill_super)(struct super_block *, void *, int));
1842 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1843 int flags, void *data,
1844 int (*fill_super)(struct super_block *, void *, int));
1845 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
1846 void generic_shutdown_super(struct super_block *sb);
1847 void kill_block_super(struct super_block *sb);
1848 void kill_anon_super(struct super_block *sb);
1849 void kill_litter_super(struct super_block *sb);
1850 void deactivate_super(struct super_block *sb);
1851 void deactivate_locked_super(struct super_block *sb);
1852 int set_anon_super(struct super_block *s, void *data);
1853 int get_anon_bdev(dev_t *);
1854 void free_anon_bdev(dev_t);
1855 struct super_block *sget(struct file_system_type *type,
1856 int (*test)(struct super_block *,void *),
1857 int (*set)(struct super_block *,void *),
1858 int flags, void *data);
1859 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
1860 const struct super_operations *ops,
1861 const struct dentry_operations *dops,
1862 unsigned long);
1863
1864 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1865 #define fops_get(fops) \
1866 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1867 #define fops_put(fops) \
1868 do { if (fops) module_put((fops)->owner); } while(0)
1869
1870 extern int register_filesystem(struct file_system_type *);
1871 extern int unregister_filesystem(struct file_system_type *);
1872 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1873 #define kern_mount(type) kern_mount_data(type, NULL)
1874 extern void kern_unmount(struct vfsmount *mnt);
1875 extern int may_umount_tree(struct vfsmount *);
1876 extern int may_umount(struct vfsmount *);
1877 extern long do_mount(const char *, const char *, const char *, unsigned long, void *);
1878 extern struct vfsmount *collect_mounts(struct path *);
1879 extern void drop_collected_mounts(struct vfsmount *);
1880 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1881 struct vfsmount *);
1882 extern int vfs_statfs(struct path *, struct kstatfs *);
1883 extern int user_statfs(const char __user *, struct kstatfs *);
1884 extern int fd_statfs(int, struct kstatfs *);
1885 extern int vfs_ustat(dev_t, struct kstatfs *);
1886 extern int freeze_super(struct super_block *super);
1887 extern int thaw_super(struct super_block *super);
1888 extern bool our_mnt(struct vfsmount *mnt);
1889
1890 extern int current_umask(void);
1891
1892 /* /sys/fs */
1893 extern struct kobject *fs_kobj;
1894
1895 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
1896 extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1897
1898 #define FLOCK_VERIFY_READ 1
1899 #define FLOCK_VERIFY_WRITE 2
1900
1901 #ifdef CONFIG_FILE_LOCKING
1902 extern int locks_mandatory_locked(struct inode *);
1903 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1904
1905 /*
1906 * Candidates for mandatory locking have the setgid bit set
1907 * but no group execute bit - an otherwise meaningless combination.
1908 */
1909
1910 static inline int __mandatory_lock(struct inode *ino)
1911 {
1912 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1913 }
1914
1915 /*
1916 * ... and these candidates should be on MS_MANDLOCK mounted fs,
1917 * otherwise these will be advisory locks
1918 */
1919
1920 static inline int mandatory_lock(struct inode *ino)
1921 {
1922 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1923 }
1924
1925 static inline int locks_verify_locked(struct inode *inode)
1926 {
1927 if (mandatory_lock(inode))
1928 return locks_mandatory_locked(inode);
1929 return 0;
1930 }
1931
1932 static inline int locks_verify_truncate(struct inode *inode,
1933 struct file *filp,
1934 loff_t size)
1935 {
1936 if (inode->i_flock && mandatory_lock(inode))
1937 return locks_mandatory_area(
1938 FLOCK_VERIFY_WRITE, inode, filp,
1939 size < inode->i_size ? size : inode->i_size,
1940 (size < inode->i_size ? inode->i_size - size
1941 : size - inode->i_size)
1942 );
1943 return 0;
1944 }
1945
1946 static inline int break_lease(struct inode *inode, unsigned int mode)
1947 {
1948 if (inode->i_flock)
1949 return __break_lease(inode, mode);
1950 return 0;
1951 }
1952 #else /* !CONFIG_FILE_LOCKING */
1953 static inline int locks_mandatory_locked(struct inode *inode)
1954 {
1955 return 0;
1956 }
1957
1958 static inline int locks_mandatory_area(int rw, struct inode *inode,
1959 struct file *filp, loff_t offset,
1960 size_t count)
1961 {
1962 return 0;
1963 }
1964
1965 static inline int __mandatory_lock(struct inode *inode)
1966 {
1967 return 0;
1968 }
1969
1970 static inline int mandatory_lock(struct inode *inode)
1971 {
1972 return 0;
1973 }
1974
1975 static inline int locks_verify_locked(struct inode *inode)
1976 {
1977 return 0;
1978 }
1979
1980 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
1981 size_t size)
1982 {
1983 return 0;
1984 }
1985
1986 static inline int break_lease(struct inode *inode, unsigned int mode)
1987 {
1988 return 0;
1989 }
1990
1991 #endif /* CONFIG_FILE_LOCKING */
1992
1993 /* fs/open.c */
1994 struct audit_names;
1995 struct filename {
1996 const char *name; /* pointer to actual string */
1997 const __user char *uptr; /* original userland pointer */
1998 struct audit_names *aname;
1999 bool separate; /* should "name" be freed? */
2000 };
2001
2002 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2003 struct file *filp);
2004 extern int do_fallocate(struct file *file, int mode, loff_t offset,
2005 loff_t len);
2006 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2007 umode_t mode);
2008 extern struct file *file_open_name(struct filename *, int, umode_t);
2009 extern struct file *filp_open(const char *, int, umode_t);
2010 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2011 const char *, int);
2012 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2013 extern int filp_close(struct file *, fl_owner_t id);
2014
2015 extern struct filename *getname(const char __user *);
2016
2017 enum {
2018 FILE_CREATED = 1,
2019 FILE_OPENED = 2
2020 };
2021 extern int finish_open(struct file *file, struct dentry *dentry,
2022 int (*open)(struct inode *, struct file *),
2023 int *opened);
2024 extern int finish_no_open(struct file *file, struct dentry *dentry);
2025
2026 /* fs/ioctl.c */
2027
2028 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2029
2030 /* fs/dcache.c */
2031 extern void __init vfs_caches_init_early(void);
2032 extern void __init vfs_caches_init(unsigned long);
2033
2034 extern struct kmem_cache *names_cachep;
2035
2036 extern void final_putname(struct filename *name);
2037
2038 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2039 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2040 #ifndef CONFIG_AUDITSYSCALL
2041 #define putname(name) final_putname(name)
2042 #else
2043 extern void putname(struct filename *name);
2044 #endif
2045
2046 #ifdef CONFIG_BLOCK
2047 extern int register_blkdev(unsigned int, const char *);
2048 extern void unregister_blkdev(unsigned int, const char *);
2049 extern struct block_device *bdget(dev_t);
2050 extern struct block_device *bdgrab(struct block_device *bdev);
2051 extern void bd_set_size(struct block_device *, loff_t size);
2052 extern sector_t blkdev_max_block(struct block_device *bdev);
2053 extern void bd_forget(struct inode *inode);
2054 extern void bdput(struct block_device *);
2055 extern void invalidate_bdev(struct block_device *);
2056 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2057 extern int sync_blockdev(struct block_device *bdev);
2058 extern void kill_bdev(struct block_device *);
2059 extern struct super_block *freeze_bdev(struct block_device *);
2060 extern void emergency_thaw_all(void);
2061 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2062 extern int fsync_bdev(struct block_device *);
2063 #else
2064 static inline void bd_forget(struct inode *inode) {}
2065 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2066 static inline void kill_bdev(struct block_device *bdev) {}
2067 static inline void invalidate_bdev(struct block_device *bdev) {}
2068
2069 static inline struct super_block *freeze_bdev(struct block_device *sb)
2070 {
2071 return NULL;
2072 }
2073
2074 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2075 {
2076 return 0;
2077 }
2078
2079 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2080 {
2081 }
2082 #endif
2083 extern int sync_filesystem(struct super_block *);
2084 extern const struct file_operations def_blk_fops;
2085 extern const struct file_operations def_chr_fops;
2086 extern const struct file_operations bad_sock_fops;
2087 extern const struct file_operations def_fifo_fops;
2088 #ifdef CONFIG_BLOCK
2089 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2090 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2091 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2092 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2093 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2094 void *holder);
2095 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2096 void *holder);
2097 extern int blkdev_put(struct block_device *bdev, fmode_t mode);
2098 #ifdef CONFIG_SYSFS
2099 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2100 extern void bd_unlink_disk_holder(struct block_device *bdev,
2101 struct gendisk *disk);
2102 #else
2103 static inline int bd_link_disk_holder(struct block_device *bdev,
2104 struct gendisk *disk)
2105 {
2106 return 0;
2107 }
2108 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2109 struct gendisk *disk)
2110 {
2111 }
2112 #endif
2113 #endif
2114
2115 /* fs/char_dev.c */
2116 #define CHRDEV_MAJOR_HASH_SIZE 255
2117 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2118 extern int register_chrdev_region(dev_t, unsigned, const char *);
2119 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2120 unsigned int count, const char *name,
2121 const struct file_operations *fops);
2122 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2123 unsigned int count, const char *name);
2124 extern void unregister_chrdev_region(dev_t, unsigned);
2125 extern void chrdev_show(struct seq_file *,off_t);
2126
2127 static inline int register_chrdev(unsigned int major, const char *name,
2128 const struct file_operations *fops)
2129 {
2130 return __register_chrdev(major, 0, 256, name, fops);
2131 }
2132
2133 static inline void unregister_chrdev(unsigned int major, const char *name)
2134 {
2135 __unregister_chrdev(major, 0, 256, name);
2136 }
2137
2138 /* fs/block_dev.c */
2139 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2140 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2141
2142 #ifdef CONFIG_BLOCK
2143 #define BLKDEV_MAJOR_HASH_SIZE 255
2144 extern const char *__bdevname(dev_t, char *buffer);
2145 extern const char *bdevname(struct block_device *bdev, char *buffer);
2146 extern struct block_device *lookup_bdev(const char *);
2147 extern void blkdev_show(struct seq_file *,off_t);
2148
2149 #else
2150 #define BLKDEV_MAJOR_HASH_SIZE 0
2151 #endif
2152
2153 extern void init_special_inode(struct inode *, umode_t, dev_t);
2154
2155 /* Invalid inode operations -- fs/bad_inode.c */
2156 extern void make_bad_inode(struct inode *);
2157 extern int is_bad_inode(struct inode *);
2158
2159 extern const struct file_operations read_pipefifo_fops;
2160 extern const struct file_operations write_pipefifo_fops;
2161 extern const struct file_operations rdwr_pipefifo_fops;
2162
2163 #ifdef CONFIG_BLOCK
2164 /*
2165 * return READ, READA, or WRITE
2166 */
2167 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2168
2169 /*
2170 * return data direction, READ or WRITE
2171 */
2172 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2173
2174 extern void check_disk_size_change(struct gendisk *disk,
2175 struct block_device *bdev);
2176 extern int revalidate_disk(struct gendisk *);
2177 extern int check_disk_change(struct block_device *);
2178 extern int __invalidate_device(struct block_device *, bool);
2179 extern int invalidate_partition(struct gendisk *, int);
2180 #endif
2181 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2182 pgoff_t start, pgoff_t end);
2183
2184 static inline void invalidate_remote_inode(struct inode *inode)
2185 {
2186 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2187 S_ISLNK(inode->i_mode))
2188 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2189 }
2190 extern int invalidate_inode_pages2(struct address_space *mapping);
2191 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2192 pgoff_t start, pgoff_t end);
2193 extern int write_inode_now(struct inode *, int);
2194 extern int filemap_fdatawrite(struct address_space *);
2195 extern int filemap_flush(struct address_space *);
2196 extern int filemap_fdatawait(struct address_space *);
2197 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2198 loff_t lend);
2199 extern int filemap_write_and_wait(struct address_space *mapping);
2200 extern int filemap_write_and_wait_range(struct address_space *mapping,
2201 loff_t lstart, loff_t lend);
2202 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2203 loff_t start, loff_t end, int sync_mode);
2204 extern int filemap_fdatawrite_range(struct address_space *mapping,
2205 loff_t start, loff_t end);
2206
2207 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2208 int datasync);
2209 extern int vfs_fsync(struct file *file, int datasync);
2210 extern int generic_write_sync(struct file *file, loff_t pos, loff_t count);
2211 extern void emergency_sync(void);
2212 extern void emergency_remount(void);
2213 #ifdef CONFIG_BLOCK
2214 extern sector_t bmap(struct inode *, sector_t);
2215 #endif
2216 extern int notify_change(struct dentry *, struct iattr *);
2217 extern int inode_permission(struct inode *, int);
2218 extern int generic_permission(struct inode *, int);
2219
2220 static inline bool execute_ok(struct inode *inode)
2221 {
2222 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2223 }
2224
2225 /*
2226 * get_write_access() gets write permission for a file.
2227 * put_write_access() releases this write permission.
2228 * This is used for regular files.
2229 * We cannot support write (and maybe mmap read-write shared) accesses and
2230 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2231 * can have the following values:
2232 * 0: no writers, no VM_DENYWRITE mappings
2233 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2234 * > 0: (i_writecount) users are writing to the file.
2235 *
2236 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2237 * except for the cases where we don't hold i_writecount yet. Then we need to
2238 * use {get,deny}_write_access() - these functions check the sign and refuse
2239 * to do the change if sign is wrong.
2240 */
2241 static inline int get_write_access(struct inode *inode)
2242 {
2243 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2244 }
2245 static inline int deny_write_access(struct file *file)
2246 {
2247 struct inode *inode = file->f_path.dentry->d_inode;
2248 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2249 }
2250 static inline void put_write_access(struct inode * inode)
2251 {
2252 atomic_dec(&inode->i_writecount);
2253 }
2254 static inline void allow_write_access(struct file *file)
2255 {
2256 if (file)
2257 atomic_inc(&file->f_path.dentry->d_inode->i_writecount);
2258 }
2259 #ifdef CONFIG_IMA
2260 static inline void i_readcount_dec(struct inode *inode)
2261 {
2262 BUG_ON(!atomic_read(&inode->i_readcount));
2263 atomic_dec(&inode->i_readcount);
2264 }
2265 static inline void i_readcount_inc(struct inode *inode)
2266 {
2267 atomic_inc(&inode->i_readcount);
2268 }
2269 #else
2270 static inline void i_readcount_dec(struct inode *inode)
2271 {
2272 return;
2273 }
2274 static inline void i_readcount_inc(struct inode *inode)
2275 {
2276 return;
2277 }
2278 #endif
2279 extern int do_pipe_flags(int *, int);
2280
2281 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2282 extern struct file * open_exec(const char *);
2283
2284 /* fs/dcache.c -- generic fs support functions */
2285 extern int is_subdir(struct dentry *, struct dentry *);
2286 extern int path_is_under(struct path *, struct path *);
2287 extern ino_t find_inode_number(struct dentry *, struct qstr *);
2288
2289 #include <linux/err.h>
2290
2291 /* needed for stackable file system support */
2292 extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
2293
2294 extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
2295
2296 extern int inode_init_always(struct super_block *, struct inode *);
2297 extern void inode_init_once(struct inode *);
2298 extern void address_space_init_once(struct address_space *mapping);
2299 extern void ihold(struct inode * inode);
2300 extern void iput(struct inode *);
2301 extern struct inode * igrab(struct inode *);
2302 extern ino_t iunique(struct super_block *, ino_t);
2303 extern int inode_needs_sync(struct inode *inode);
2304 extern int generic_delete_inode(struct inode *inode);
2305 static inline int generic_drop_inode(struct inode *inode)
2306 {
2307 return !inode->i_nlink || inode_unhashed(inode);
2308 }
2309
2310 extern struct inode *ilookup5_nowait(struct super_block *sb,
2311 unsigned long hashval, int (*test)(struct inode *, void *),
2312 void *data);
2313 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2314 int (*test)(struct inode *, void *), void *data);
2315 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2316
2317 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2318 extern struct inode * iget_locked(struct super_block *, unsigned long);
2319 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2320 extern int insert_inode_locked(struct inode *);
2321 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2322 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2323 #else
2324 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2325 #endif
2326 extern void unlock_new_inode(struct inode *);
2327 extern unsigned int get_next_ino(void);
2328
2329 extern void __iget(struct inode * inode);
2330 extern void iget_failed(struct inode *);
2331 extern void clear_inode(struct inode *);
2332 extern void __destroy_inode(struct inode *);
2333 extern struct inode *new_inode_pseudo(struct super_block *sb);
2334 extern struct inode *new_inode(struct super_block *sb);
2335 extern void free_inode_nonrcu(struct inode *inode);
2336 extern int should_remove_suid(struct dentry *);
2337 extern int file_remove_suid(struct file *);
2338
2339 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2340 static inline void insert_inode_hash(struct inode *inode)
2341 {
2342 __insert_inode_hash(inode, inode->i_ino);
2343 }
2344
2345 extern void __remove_inode_hash(struct inode *);
2346 static inline void remove_inode_hash(struct inode *inode)
2347 {
2348 if (!inode_unhashed(inode))
2349 __remove_inode_hash(inode);
2350 }
2351
2352 extern void inode_sb_list_add(struct inode *inode);
2353
2354 #ifdef CONFIG_BLOCK
2355 extern void submit_bio(int, struct bio *);
2356 extern int bdev_read_only(struct block_device *);
2357 #endif
2358 extern int set_blocksize(struct block_device *, int);
2359 extern int sb_set_blocksize(struct super_block *, int);
2360 extern int sb_min_blocksize(struct super_block *, int);
2361
2362 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2363 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2364 extern int generic_file_remap_pages(struct vm_area_struct *, unsigned long addr,
2365 unsigned long size, pgoff_t pgoff);
2366 extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
2367 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
2368 extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2369 extern ssize_t __generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long,
2370 loff_t *);
2371 extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2372 extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
2373 unsigned long *, loff_t, loff_t *, size_t, size_t);
2374 extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
2375 unsigned long, loff_t, loff_t *, size_t, ssize_t);
2376 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2377 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2378 extern int generic_segment_checks(const struct iovec *iov,
2379 unsigned long *nr_segs, size_t *count, int access_flags);
2380
2381 /* fs/block_dev.c */
2382 extern ssize_t blkdev_aio_read(struct kiocb *iocb, const struct iovec *iov,
2383 unsigned long nr_segs, loff_t pos);
2384 extern ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov,
2385 unsigned long nr_segs, loff_t pos);
2386 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2387 int datasync);
2388 extern void block_sync_page(struct page *page);
2389
2390 /* fs/splice.c */
2391 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2392 struct pipe_inode_info *, size_t, unsigned int);
2393 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2394 struct pipe_inode_info *, size_t, unsigned int);
2395 extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
2396 struct file *, loff_t *, size_t, unsigned int);
2397 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2398 struct file *out, loff_t *, size_t len, unsigned int flags);
2399 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2400 size_t len, unsigned int flags);
2401
2402 extern void
2403 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2404 extern loff_t noop_llseek(struct file *file, loff_t offset, int origin);
2405 extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
2406 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
2407 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2408 int origin, loff_t maxsize, loff_t eof);
2409 extern int generic_file_open(struct inode * inode, struct file * filp);
2410 extern int nonseekable_open(struct inode * inode, struct file * filp);
2411
2412 #ifdef CONFIG_FS_XIP
2413 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2414 loff_t *ppos);
2415 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
2416 extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2417 size_t len, loff_t *ppos);
2418 extern int xip_truncate_page(struct address_space *mapping, loff_t from);
2419 #else
2420 static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2421 {
2422 return 0;
2423 }
2424 #endif
2425
2426 #ifdef CONFIG_BLOCK
2427 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2428 loff_t file_offset);
2429
2430 enum {
2431 /* need locking between buffered and direct access */
2432 DIO_LOCKING = 0x01,
2433
2434 /* filesystem does not support filling holes */
2435 DIO_SKIP_HOLES = 0x02,
2436 };
2437
2438 void dio_end_io(struct bio *bio, int error);
2439
2440 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2441 struct block_device *bdev, const struct iovec *iov, loff_t offset,
2442 unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
2443 dio_submit_t submit_io, int flags);
2444
2445 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
2446 struct inode *inode, const struct iovec *iov, loff_t offset,
2447 unsigned long nr_segs, get_block_t get_block)
2448 {
2449 return __blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov,
2450 offset, nr_segs, get_block, NULL, NULL,
2451 DIO_LOCKING | DIO_SKIP_HOLES);
2452 }
2453 #endif
2454
2455 void inode_dio_wait(struct inode *inode);
2456 void inode_dio_done(struct inode *inode);
2457
2458 extern const struct file_operations generic_ro_fops;
2459
2460 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2461
2462 extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
2463 extern int vfs_follow_link(struct nameidata *, const char *);
2464 extern int page_readlink(struct dentry *, char __user *, int);
2465 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2466 extern void page_put_link(struct dentry *, struct nameidata *, void *);
2467 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2468 int nofs);
2469 extern int page_symlink(struct inode *inode, const char *symname, int len);
2470 extern const struct inode_operations page_symlink_inode_operations;
2471 extern int generic_readlink(struct dentry *, char __user *, int);
2472 extern void generic_fillattr(struct inode *, struct kstat *);
2473 extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2474 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2475 void inode_add_bytes(struct inode *inode, loff_t bytes);
2476 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2477 loff_t inode_get_bytes(struct inode *inode);
2478 void inode_set_bytes(struct inode *inode, loff_t bytes);
2479
2480 extern int vfs_readdir(struct file *, filldir_t, void *);
2481
2482 extern int vfs_stat(const char __user *, struct kstat *);
2483 extern int vfs_lstat(const char __user *, struct kstat *);
2484 extern int vfs_fstat(unsigned int, struct kstat *);
2485 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2486
2487 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2488 unsigned long arg);
2489 extern int __generic_block_fiemap(struct inode *inode,
2490 struct fiemap_extent_info *fieinfo,
2491 loff_t start, loff_t len,
2492 get_block_t *get_block);
2493 extern int generic_block_fiemap(struct inode *inode,
2494 struct fiemap_extent_info *fieinfo, u64 start,
2495 u64 len, get_block_t *get_block);
2496
2497 extern void get_filesystem(struct file_system_type *fs);
2498 extern void put_filesystem(struct file_system_type *fs);
2499 extern struct file_system_type *get_fs_type(const char *name);
2500 extern struct super_block *get_super(struct block_device *);
2501 extern struct super_block *get_super_thawed(struct block_device *);
2502 extern struct super_block *get_active_super(struct block_device *bdev);
2503 extern void drop_super(struct super_block *sb);
2504 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2505 extern void iterate_supers_type(struct file_system_type *,
2506 void (*)(struct super_block *, void *), void *);
2507
2508 extern int dcache_dir_open(struct inode *, struct file *);
2509 extern int dcache_dir_close(struct inode *, struct file *);
2510 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2511 extern int dcache_readdir(struct file *, void *, filldir_t);
2512 extern int simple_setattr(struct dentry *, struct iattr *);
2513 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2514 extern int simple_statfs(struct dentry *, struct kstatfs *);
2515 extern int simple_open(struct inode *inode, struct file *file);
2516 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2517 extern int simple_unlink(struct inode *, struct dentry *);
2518 extern int simple_rmdir(struct inode *, struct dentry *);
2519 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2520 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2521 extern int simple_empty(struct dentry *);
2522 extern int simple_readpage(struct file *file, struct page *page);
2523 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2524 loff_t pos, unsigned len, unsigned flags,
2525 struct page **pagep, void **fsdata);
2526 extern int simple_write_end(struct file *file, struct address_space *mapping,
2527 loff_t pos, unsigned len, unsigned copied,
2528 struct page *page, void *fsdata);
2529
2530 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
2531 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2532 extern const struct file_operations simple_dir_operations;
2533 extern const struct inode_operations simple_dir_inode_operations;
2534 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2535 struct dentry *d_alloc_name(struct dentry *, const char *);
2536 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2537 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2538 extern void simple_release_fs(struct vfsmount **mount, int *count);
2539
2540 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2541 loff_t *ppos, const void *from, size_t available);
2542 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2543 const void __user *from, size_t count);
2544
2545 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2546
2547 extern int generic_check_addressable(unsigned, u64);
2548
2549 #ifdef CONFIG_MIGRATION
2550 extern int buffer_migrate_page(struct address_space *,
2551 struct page *, struct page *,
2552 enum migrate_mode);
2553 #else
2554 #define buffer_migrate_page NULL
2555 #endif
2556
2557 extern int inode_change_ok(const struct inode *, struct iattr *);
2558 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2559 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2560
2561 extern int file_update_time(struct file *file);
2562
2563 extern int generic_show_options(struct seq_file *m, struct dentry *root);
2564 extern void save_mount_options(struct super_block *sb, char *options);
2565 extern void replace_mount_options(struct super_block *sb, char *options);
2566
2567 static inline ino_t parent_ino(struct dentry *dentry)
2568 {
2569 ino_t res;
2570
2571 /*
2572 * Don't strictly need d_lock here? If the parent ino could change
2573 * then surely we'd have a deeper race in the caller?
2574 */
2575 spin_lock(&dentry->d_lock);
2576 res = dentry->d_parent->d_inode->i_ino;
2577 spin_unlock(&dentry->d_lock);
2578 return res;
2579 }
2580
2581 /* Transaction based IO helpers */
2582
2583 /*
2584 * An argresp is stored in an allocated page and holds the
2585 * size of the argument or response, along with its content
2586 */
2587 struct simple_transaction_argresp {
2588 ssize_t size;
2589 char data[0];
2590 };
2591
2592 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2593
2594 char *simple_transaction_get(struct file *file, const char __user *buf,
2595 size_t size);
2596 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2597 size_t size, loff_t *pos);
2598 int simple_transaction_release(struct inode *inode, struct file *file);
2599
2600 void simple_transaction_set(struct file *file, size_t n);
2601
2602 /*
2603 * simple attribute files
2604 *
2605 * These attributes behave similar to those in sysfs:
2606 *
2607 * Writing to an attribute immediately sets a value, an open file can be
2608 * written to multiple times.
2609 *
2610 * Reading from an attribute creates a buffer from the value that might get
2611 * read with multiple read calls. When the attribute has been read
2612 * completely, no further read calls are possible until the file is opened
2613 * again.
2614 *
2615 * All attributes contain a text representation of a numeric value
2616 * that are accessed with the get() and set() functions.
2617 */
2618 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2619 static int __fops ## _open(struct inode *inode, struct file *file) \
2620 { \
2621 __simple_attr_check_format(__fmt, 0ull); \
2622 return simple_attr_open(inode, file, __get, __set, __fmt); \
2623 } \
2624 static const struct file_operations __fops = { \
2625 .owner = THIS_MODULE, \
2626 .open = __fops ## _open, \
2627 .release = simple_attr_release, \
2628 .read = simple_attr_read, \
2629 .write = simple_attr_write, \
2630 .llseek = generic_file_llseek, \
2631 };
2632
2633 static inline __printf(1, 2)
2634 void __simple_attr_check_format(const char *fmt, ...)
2635 {
2636 /* don't do anything, just let the compiler check the arguments; */
2637 }
2638
2639 int simple_attr_open(struct inode *inode, struct file *file,
2640 int (*get)(void *, u64 *), int (*set)(void *, u64),
2641 const char *fmt);
2642 int simple_attr_release(struct inode *inode, struct file *file);
2643 ssize_t simple_attr_read(struct file *file, char __user *buf,
2644 size_t len, loff_t *ppos);
2645 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2646 size_t len, loff_t *ppos);
2647
2648 struct ctl_table;
2649 int proc_nr_files(struct ctl_table *table, int write,
2650 void __user *buffer, size_t *lenp, loff_t *ppos);
2651 int proc_nr_dentry(struct ctl_table *table, int write,
2652 void __user *buffer, size_t *lenp, loff_t *ppos);
2653 int proc_nr_inodes(struct ctl_table *table, int write,
2654 void __user *buffer, size_t *lenp, loff_t *ppos);
2655 int __init get_filesystem_list(char *buf);
2656
2657 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
2658 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
2659
2660 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2661 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2662 (flag & __FMODE_NONOTIFY)))
2663
2664 static inline int is_sxid(umode_t mode)
2665 {
2666 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
2667 }
2668
2669 static inline void inode_has_no_xattr(struct inode *inode)
2670 {
2671 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
2672 inode->i_flags |= S_NOSEC;
2673 }
2674
2675 #endif /* _LINUX_FS_H */