]> git.proxmox.com Git - mirror_ubuntu-artful-kernel.git/blob - include/linux/fs.h
Merge branches 'acpi-fan', 'acpi-video' and 'acpi-ec'
[mirror_ubuntu-artful-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/list_lru.h>
14 #include <linux/llist.h>
15 #include <linux/radix-tree.h>
16 #include <linux/rbtree.h>
17 #include <linux/init.h>
18 #include <linux/pid.h>
19 #include <linux/bug.h>
20 #include <linux/mutex.h>
21 #include <linux/capability.h>
22 #include <linux/semaphore.h>
23 #include <linux/fiemap.h>
24 #include <linux/rculist_bl.h>
25 #include <linux/atomic.h>
26 #include <linux/shrinker.h>
27 #include <linux/migrate_mode.h>
28 #include <linux/uidgid.h>
29 #include <linux/lockdep.h>
30 #include <linux/percpu-rwsem.h>
31 #include <linux/blk_types.h>
32
33 #include <asm/byteorder.h>
34 #include <uapi/linux/fs.h>
35
36 struct export_operations;
37 struct hd_geometry;
38 struct iovec;
39 struct nameidata;
40 struct kiocb;
41 struct kobject;
42 struct pipe_inode_info;
43 struct poll_table_struct;
44 struct kstatfs;
45 struct vm_area_struct;
46 struct vfsmount;
47 struct cred;
48 struct swap_info_struct;
49 struct seq_file;
50 struct workqueue_struct;
51 struct iov_iter;
52
53 extern void __init inode_init(void);
54 extern void __init inode_init_early(void);
55 extern void __init files_init(unsigned long);
56
57 extern struct files_stat_struct files_stat;
58 extern unsigned long get_max_files(void);
59 extern int sysctl_nr_open;
60 extern struct inodes_stat_t inodes_stat;
61 extern int leases_enable, lease_break_time;
62 extern int sysctl_protected_symlinks;
63 extern int sysctl_protected_hardlinks;
64
65 struct buffer_head;
66 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
67 struct buffer_head *bh_result, int create);
68 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
69 ssize_t bytes, void *private);
70
71 #define MAY_EXEC 0x00000001
72 #define MAY_WRITE 0x00000002
73 #define MAY_READ 0x00000004
74 #define MAY_APPEND 0x00000008
75 #define MAY_ACCESS 0x00000010
76 #define MAY_OPEN 0x00000020
77 #define MAY_CHDIR 0x00000040
78 /* called from RCU mode, don't block */
79 #define MAY_NOT_BLOCK 0x00000080
80
81 /*
82 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
83 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
84 */
85
86 /* file is open for reading */
87 #define FMODE_READ ((__force fmode_t)0x1)
88 /* file is open for writing */
89 #define FMODE_WRITE ((__force fmode_t)0x2)
90 /* file is seekable */
91 #define FMODE_LSEEK ((__force fmode_t)0x4)
92 /* file can be accessed using pread */
93 #define FMODE_PREAD ((__force fmode_t)0x8)
94 /* file can be accessed using pwrite */
95 #define FMODE_PWRITE ((__force fmode_t)0x10)
96 /* File is opened for execution with sys_execve / sys_uselib */
97 #define FMODE_EXEC ((__force fmode_t)0x20)
98 /* File is opened with O_NDELAY (only set for block devices) */
99 #define FMODE_NDELAY ((__force fmode_t)0x40)
100 /* File is opened with O_EXCL (only set for block devices) */
101 #define FMODE_EXCL ((__force fmode_t)0x80)
102 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
103 (specialy hack for floppy.c) */
104 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
105 /* 32bit hashes as llseek() offset (for directories) */
106 #define FMODE_32BITHASH ((__force fmode_t)0x200)
107 /* 64bit hashes as llseek() offset (for directories) */
108 #define FMODE_64BITHASH ((__force fmode_t)0x400)
109
110 /*
111 * Don't update ctime and mtime.
112 *
113 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
114 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
115 */
116 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
117
118 /* Expect random access pattern */
119 #define FMODE_RANDOM ((__force fmode_t)0x1000)
120
121 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
122 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
123
124 /* File is opened with O_PATH; almost nothing can be done with it */
125 #define FMODE_PATH ((__force fmode_t)0x4000)
126
127 /* File needs atomic accesses to f_pos */
128 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
129 /* Write access to underlying fs */
130 #define FMODE_WRITER ((__force fmode_t)0x10000)
131 /* Has read method(s) */
132 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
133 /* Has write method(s) */
134 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
135
136 /* File was opened by fanotify and shouldn't generate fanotify events */
137 #define FMODE_NONOTIFY ((__force fmode_t)0x1000000)
138
139 /*
140 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
141 * that indicates that they should check the contents of the iovec are
142 * valid, but not check the memory that the iovec elements
143 * points too.
144 */
145 #define CHECK_IOVEC_ONLY -1
146
147 /*
148 * The below are the various read and write types that we support. Some of
149 * them include behavioral modifiers that send information down to the
150 * block layer and IO scheduler. Terminology:
151 *
152 * The block layer uses device plugging to defer IO a little bit, in
153 * the hope that we will see more IO very shortly. This increases
154 * coalescing of adjacent IO and thus reduces the number of IOs we
155 * have to send to the device. It also allows for better queuing,
156 * if the IO isn't mergeable. If the caller is going to be waiting
157 * for the IO, then he must ensure that the device is unplugged so
158 * that the IO is dispatched to the driver.
159 *
160 * All IO is handled async in Linux. This is fine for background
161 * writes, but for reads or writes that someone waits for completion
162 * on, we want to notify the block layer and IO scheduler so that they
163 * know about it. That allows them to make better scheduling
164 * decisions. So when the below references 'sync' and 'async', it
165 * is referencing this priority hint.
166 *
167 * With that in mind, the available types are:
168 *
169 * READ A normal read operation. Device will be plugged.
170 * READ_SYNC A synchronous read. Device is not plugged, caller can
171 * immediately wait on this read without caring about
172 * unplugging.
173 * READA Used for read-ahead operations. Lower priority, and the
174 * block layer could (in theory) choose to ignore this
175 * request if it runs into resource problems.
176 * WRITE A normal async write. Device will be plugged.
177 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
178 * the hint that someone will be waiting on this IO
179 * shortly. The write equivalent of READ_SYNC.
180 * WRITE_ODIRECT Special case write for O_DIRECT only.
181 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
182 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
183 * non-volatile media on completion.
184 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
185 * by a cache flush and data is guaranteed to be on
186 * non-volatile media on completion.
187 *
188 */
189 #define RW_MASK REQ_WRITE
190 #define RWA_MASK REQ_RAHEAD
191
192 #define READ 0
193 #define WRITE RW_MASK
194 #define READA RWA_MASK
195
196 #define READ_SYNC (READ | REQ_SYNC)
197 #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
198 #define WRITE_ODIRECT (WRITE | REQ_SYNC)
199 #define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
200 #define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
201 #define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
202
203 /*
204 * Attribute flags. These should be or-ed together to figure out what
205 * has been changed!
206 */
207 #define ATTR_MODE (1 << 0)
208 #define ATTR_UID (1 << 1)
209 #define ATTR_GID (1 << 2)
210 #define ATTR_SIZE (1 << 3)
211 #define ATTR_ATIME (1 << 4)
212 #define ATTR_MTIME (1 << 5)
213 #define ATTR_CTIME (1 << 6)
214 #define ATTR_ATIME_SET (1 << 7)
215 #define ATTR_MTIME_SET (1 << 8)
216 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
217 #define ATTR_ATTR_FLAG (1 << 10)
218 #define ATTR_KILL_SUID (1 << 11)
219 #define ATTR_KILL_SGID (1 << 12)
220 #define ATTR_FILE (1 << 13)
221 #define ATTR_KILL_PRIV (1 << 14)
222 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
223 #define ATTR_TIMES_SET (1 << 16)
224
225 /*
226 * Whiteout is represented by a char device. The following constants define the
227 * mode and device number to use.
228 */
229 #define WHITEOUT_MODE 0
230 #define WHITEOUT_DEV 0
231
232 /*
233 * This is the Inode Attributes structure, used for notify_change(). It
234 * uses the above definitions as flags, to know which values have changed.
235 * Also, in this manner, a Filesystem can look at only the values it cares
236 * about. Basically, these are the attributes that the VFS layer can
237 * request to change from the FS layer.
238 *
239 * Derek Atkins <warlord@MIT.EDU> 94-10-20
240 */
241 struct iattr {
242 unsigned int ia_valid;
243 umode_t ia_mode;
244 kuid_t ia_uid;
245 kgid_t ia_gid;
246 loff_t ia_size;
247 struct timespec ia_atime;
248 struct timespec ia_mtime;
249 struct timespec ia_ctime;
250
251 /*
252 * Not an attribute, but an auxiliary info for filesystems wanting to
253 * implement an ftruncate() like method. NOTE: filesystem should
254 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
255 */
256 struct file *ia_file;
257 };
258
259 /*
260 * Includes for diskquotas.
261 */
262 #include <linux/quota.h>
263
264 /*
265 * Maximum number of layers of fs stack. Needs to be limited to
266 * prevent kernel stack overflow
267 */
268 #define FILESYSTEM_MAX_STACK_DEPTH 2
269
270 /**
271 * enum positive_aop_returns - aop return codes with specific semantics
272 *
273 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
274 * completed, that the page is still locked, and
275 * should be considered active. The VM uses this hint
276 * to return the page to the active list -- it won't
277 * be a candidate for writeback again in the near
278 * future. Other callers must be careful to unlock
279 * the page if they get this return. Returned by
280 * writepage();
281 *
282 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
283 * unlocked it and the page might have been truncated.
284 * The caller should back up to acquiring a new page and
285 * trying again. The aop will be taking reasonable
286 * precautions not to livelock. If the caller held a page
287 * reference, it should drop it before retrying. Returned
288 * by readpage().
289 *
290 * address_space_operation functions return these large constants to indicate
291 * special semantics to the caller. These are much larger than the bytes in a
292 * page to allow for functions that return the number of bytes operated on in a
293 * given page.
294 */
295
296 enum positive_aop_returns {
297 AOP_WRITEPAGE_ACTIVATE = 0x80000,
298 AOP_TRUNCATED_PAGE = 0x80001,
299 };
300
301 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
302 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
303 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
304 * helper code (eg buffer layer)
305 * to clear GFP_FS from alloc */
306
307 /*
308 * oh the beauties of C type declarations.
309 */
310 struct page;
311 struct address_space;
312 struct writeback_control;
313
314 /*
315 * "descriptor" for what we're up to with a read.
316 * This allows us to use the same read code yet
317 * have multiple different users of the data that
318 * we read from a file.
319 *
320 * The simplest case just copies the data to user
321 * mode.
322 */
323 typedef struct {
324 size_t written;
325 size_t count;
326 union {
327 char __user *buf;
328 void *data;
329 } arg;
330 int error;
331 } read_descriptor_t;
332
333 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
334 unsigned long, unsigned long);
335
336 struct address_space_operations {
337 int (*writepage)(struct page *page, struct writeback_control *wbc);
338 int (*readpage)(struct file *, struct page *);
339
340 /* Write back some dirty pages from this mapping. */
341 int (*writepages)(struct address_space *, struct writeback_control *);
342
343 /* Set a page dirty. Return true if this dirtied it */
344 int (*set_page_dirty)(struct page *page);
345
346 int (*readpages)(struct file *filp, struct address_space *mapping,
347 struct list_head *pages, unsigned nr_pages);
348
349 int (*write_begin)(struct file *, struct address_space *mapping,
350 loff_t pos, unsigned len, unsigned flags,
351 struct page **pagep, void **fsdata);
352 int (*write_end)(struct file *, struct address_space *mapping,
353 loff_t pos, unsigned len, unsigned copied,
354 struct page *page, void *fsdata);
355
356 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
357 sector_t (*bmap)(struct address_space *, sector_t);
358 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
359 int (*releasepage) (struct page *, gfp_t);
360 void (*freepage)(struct page *);
361 ssize_t (*direct_IO)(int, struct kiocb *, struct iov_iter *iter, loff_t offset);
362 int (*get_xip_mem)(struct address_space *, pgoff_t, int,
363 void **, unsigned long *);
364 /*
365 * migrate the contents of a page to the specified target. If
366 * migrate_mode is MIGRATE_ASYNC, it must not block.
367 */
368 int (*migratepage) (struct address_space *,
369 struct page *, struct page *, enum migrate_mode);
370 int (*launder_page) (struct page *);
371 int (*is_partially_uptodate) (struct page *, unsigned long,
372 unsigned long);
373 void (*is_dirty_writeback) (struct page *, bool *, bool *);
374 int (*error_remove_page)(struct address_space *, struct page *);
375
376 /* swapfile support */
377 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
378 sector_t *span);
379 void (*swap_deactivate)(struct file *file);
380 };
381
382 extern const struct address_space_operations empty_aops;
383
384 /*
385 * pagecache_write_begin/pagecache_write_end must be used by general code
386 * to write into the pagecache.
387 */
388 int pagecache_write_begin(struct file *, struct address_space *mapping,
389 loff_t pos, unsigned len, unsigned flags,
390 struct page **pagep, void **fsdata);
391
392 int pagecache_write_end(struct file *, struct address_space *mapping,
393 loff_t pos, unsigned len, unsigned copied,
394 struct page *page, void *fsdata);
395
396 struct backing_dev_info;
397 struct address_space {
398 struct inode *host; /* owner: inode, block_device */
399 struct radix_tree_root page_tree; /* radix tree of all pages */
400 spinlock_t tree_lock; /* and lock protecting it */
401 atomic_t i_mmap_writable;/* count VM_SHARED mappings */
402 struct rb_root i_mmap; /* tree of private and shared mappings */
403 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
404 struct mutex i_mmap_mutex; /* protect tree, count, list */
405 /* Protected by tree_lock together with the radix tree */
406 unsigned long nrpages; /* number of total pages */
407 unsigned long nrshadows; /* number of shadow entries */
408 pgoff_t writeback_index;/* writeback starts here */
409 const struct address_space_operations *a_ops; /* methods */
410 unsigned long flags; /* error bits/gfp mask */
411 struct backing_dev_info *backing_dev_info; /* device readahead, etc */
412 spinlock_t private_lock; /* for use by the address_space */
413 struct list_head private_list; /* ditto */
414 void *private_data; /* ditto */
415 } __attribute__((aligned(sizeof(long))));
416 /*
417 * On most architectures that alignment is already the case; but
418 * must be enforced here for CRIS, to let the least significant bit
419 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
420 */
421 struct request_queue;
422
423 struct block_device {
424 dev_t bd_dev; /* not a kdev_t - it's a search key */
425 int bd_openers;
426 struct inode * bd_inode; /* will die */
427 struct super_block * bd_super;
428 struct mutex bd_mutex; /* open/close mutex */
429 struct list_head bd_inodes;
430 void * bd_claiming;
431 void * bd_holder;
432 int bd_holders;
433 bool bd_write_holder;
434 #ifdef CONFIG_SYSFS
435 struct list_head bd_holder_disks;
436 #endif
437 struct block_device * bd_contains;
438 unsigned bd_block_size;
439 struct hd_struct * bd_part;
440 /* number of times partitions within this device have been opened. */
441 unsigned bd_part_count;
442 int bd_invalidated;
443 struct gendisk * bd_disk;
444 struct request_queue * bd_queue;
445 struct list_head bd_list;
446 /*
447 * Private data. You must have bd_claim'ed the block_device
448 * to use this. NOTE: bd_claim allows an owner to claim
449 * the same device multiple times, the owner must take special
450 * care to not mess up bd_private for that case.
451 */
452 unsigned long bd_private;
453
454 /* The counter of freeze processes */
455 int bd_fsfreeze_count;
456 /* Mutex for freeze */
457 struct mutex bd_fsfreeze_mutex;
458 };
459
460 /*
461 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
462 * radix trees
463 */
464 #define PAGECACHE_TAG_DIRTY 0
465 #define PAGECACHE_TAG_WRITEBACK 1
466 #define PAGECACHE_TAG_TOWRITE 2
467
468 int mapping_tagged(struct address_space *mapping, int tag);
469
470 /*
471 * Might pages of this file be mapped into userspace?
472 */
473 static inline int mapping_mapped(struct address_space *mapping)
474 {
475 return !RB_EMPTY_ROOT(&mapping->i_mmap) ||
476 !list_empty(&mapping->i_mmap_nonlinear);
477 }
478
479 /*
480 * Might pages of this file have been modified in userspace?
481 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
482 * marks vma as VM_SHARED if it is shared, and the file was opened for
483 * writing i.e. vma may be mprotected writable even if now readonly.
484 *
485 * If i_mmap_writable is negative, no new writable mappings are allowed. You
486 * can only deny writable mappings, if none exists right now.
487 */
488 static inline int mapping_writably_mapped(struct address_space *mapping)
489 {
490 return atomic_read(&mapping->i_mmap_writable) > 0;
491 }
492
493 static inline int mapping_map_writable(struct address_space *mapping)
494 {
495 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
496 0 : -EPERM;
497 }
498
499 static inline void mapping_unmap_writable(struct address_space *mapping)
500 {
501 atomic_dec(&mapping->i_mmap_writable);
502 }
503
504 static inline int mapping_deny_writable(struct address_space *mapping)
505 {
506 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
507 0 : -EBUSY;
508 }
509
510 static inline void mapping_allow_writable(struct address_space *mapping)
511 {
512 atomic_inc(&mapping->i_mmap_writable);
513 }
514
515 /*
516 * Use sequence counter to get consistent i_size on 32-bit processors.
517 */
518 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
519 #include <linux/seqlock.h>
520 #define __NEED_I_SIZE_ORDERED
521 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
522 #else
523 #define i_size_ordered_init(inode) do { } while (0)
524 #endif
525
526 struct posix_acl;
527 #define ACL_NOT_CACHED ((void *)(-1))
528
529 #define IOP_FASTPERM 0x0001
530 #define IOP_LOOKUP 0x0002
531 #define IOP_NOFOLLOW 0x0004
532
533 /*
534 * Keep mostly read-only and often accessed (especially for
535 * the RCU path lookup and 'stat' data) fields at the beginning
536 * of the 'struct inode'
537 */
538 struct inode {
539 umode_t i_mode;
540 unsigned short i_opflags;
541 kuid_t i_uid;
542 kgid_t i_gid;
543 unsigned int i_flags;
544
545 #ifdef CONFIG_FS_POSIX_ACL
546 struct posix_acl *i_acl;
547 struct posix_acl *i_default_acl;
548 #endif
549
550 const struct inode_operations *i_op;
551 struct super_block *i_sb;
552 struct address_space *i_mapping;
553
554 #ifdef CONFIG_SECURITY
555 void *i_security;
556 #endif
557
558 /* Stat data, not accessed from path walking */
559 unsigned long i_ino;
560 /*
561 * Filesystems may only read i_nlink directly. They shall use the
562 * following functions for modification:
563 *
564 * (set|clear|inc|drop)_nlink
565 * inode_(inc|dec)_link_count
566 */
567 union {
568 const unsigned int i_nlink;
569 unsigned int __i_nlink;
570 };
571 dev_t i_rdev;
572 loff_t i_size;
573 struct timespec i_atime;
574 struct timespec i_mtime;
575 struct timespec i_ctime;
576 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
577 unsigned short i_bytes;
578 unsigned int i_blkbits;
579 blkcnt_t i_blocks;
580
581 #ifdef __NEED_I_SIZE_ORDERED
582 seqcount_t i_size_seqcount;
583 #endif
584
585 /* Misc */
586 unsigned long i_state;
587 struct mutex i_mutex;
588
589 unsigned long dirtied_when; /* jiffies of first dirtying */
590
591 struct hlist_node i_hash;
592 struct list_head i_wb_list; /* backing dev IO list */
593 struct list_head i_lru; /* inode LRU list */
594 struct list_head i_sb_list;
595 union {
596 struct hlist_head i_dentry;
597 struct rcu_head i_rcu;
598 };
599 u64 i_version;
600 atomic_t i_count;
601 atomic_t i_dio_count;
602 atomic_t i_writecount;
603 #ifdef CONFIG_IMA
604 atomic_t i_readcount; /* struct files open RO */
605 #endif
606 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
607 struct file_lock *i_flock;
608 struct address_space i_data;
609 struct list_head i_devices;
610 union {
611 struct pipe_inode_info *i_pipe;
612 struct block_device *i_bdev;
613 struct cdev *i_cdev;
614 };
615
616 __u32 i_generation;
617
618 #ifdef CONFIG_FSNOTIFY
619 __u32 i_fsnotify_mask; /* all events this inode cares about */
620 struct hlist_head i_fsnotify_marks;
621 #endif
622
623 void *i_private; /* fs or device private pointer */
624 };
625
626 static inline int inode_unhashed(struct inode *inode)
627 {
628 return hlist_unhashed(&inode->i_hash);
629 }
630
631 /*
632 * inode->i_mutex nesting subclasses for the lock validator:
633 *
634 * 0: the object of the current VFS operation
635 * 1: parent
636 * 2: child/target
637 * 3: xattr
638 * 4: second non-directory
639 * 5: second parent (when locking independent directories in rename)
640 *
641 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
642 * non-directories at once.
643 *
644 * The locking order between these classes is
645 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
646 */
647 enum inode_i_mutex_lock_class
648 {
649 I_MUTEX_NORMAL,
650 I_MUTEX_PARENT,
651 I_MUTEX_CHILD,
652 I_MUTEX_XATTR,
653 I_MUTEX_NONDIR2,
654 I_MUTEX_PARENT2,
655 };
656
657 void lock_two_nondirectories(struct inode *, struct inode*);
658 void unlock_two_nondirectories(struct inode *, struct inode*);
659
660 /*
661 * NOTE: in a 32bit arch with a preemptable kernel and
662 * an UP compile the i_size_read/write must be atomic
663 * with respect to the local cpu (unlike with preempt disabled),
664 * but they don't need to be atomic with respect to other cpus like in
665 * true SMP (so they need either to either locally disable irq around
666 * the read or for example on x86 they can be still implemented as a
667 * cmpxchg8b without the need of the lock prefix). For SMP compiles
668 * and 64bit archs it makes no difference if preempt is enabled or not.
669 */
670 static inline loff_t i_size_read(const struct inode *inode)
671 {
672 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
673 loff_t i_size;
674 unsigned int seq;
675
676 do {
677 seq = read_seqcount_begin(&inode->i_size_seqcount);
678 i_size = inode->i_size;
679 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
680 return i_size;
681 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
682 loff_t i_size;
683
684 preempt_disable();
685 i_size = inode->i_size;
686 preempt_enable();
687 return i_size;
688 #else
689 return inode->i_size;
690 #endif
691 }
692
693 /*
694 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
695 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
696 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
697 */
698 static inline void i_size_write(struct inode *inode, loff_t i_size)
699 {
700 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
701 preempt_disable();
702 write_seqcount_begin(&inode->i_size_seqcount);
703 inode->i_size = i_size;
704 write_seqcount_end(&inode->i_size_seqcount);
705 preempt_enable();
706 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
707 preempt_disable();
708 inode->i_size = i_size;
709 preempt_enable();
710 #else
711 inode->i_size = i_size;
712 #endif
713 }
714
715 /* Helper functions so that in most cases filesystems will
716 * not need to deal directly with kuid_t and kgid_t and can
717 * instead deal with the raw numeric values that are stored
718 * in the filesystem.
719 */
720 static inline uid_t i_uid_read(const struct inode *inode)
721 {
722 return from_kuid(&init_user_ns, inode->i_uid);
723 }
724
725 static inline gid_t i_gid_read(const struct inode *inode)
726 {
727 return from_kgid(&init_user_ns, inode->i_gid);
728 }
729
730 static inline void i_uid_write(struct inode *inode, uid_t uid)
731 {
732 inode->i_uid = make_kuid(&init_user_ns, uid);
733 }
734
735 static inline void i_gid_write(struct inode *inode, gid_t gid)
736 {
737 inode->i_gid = make_kgid(&init_user_ns, gid);
738 }
739
740 static inline unsigned iminor(const struct inode *inode)
741 {
742 return MINOR(inode->i_rdev);
743 }
744
745 static inline unsigned imajor(const struct inode *inode)
746 {
747 return MAJOR(inode->i_rdev);
748 }
749
750 extern struct block_device *I_BDEV(struct inode *inode);
751
752 struct fown_struct {
753 rwlock_t lock; /* protects pid, uid, euid fields */
754 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
755 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
756 kuid_t uid, euid; /* uid/euid of process setting the owner */
757 int signum; /* posix.1b rt signal to be delivered on IO */
758 };
759
760 /*
761 * Track a single file's readahead state
762 */
763 struct file_ra_state {
764 pgoff_t start; /* where readahead started */
765 unsigned int size; /* # of readahead pages */
766 unsigned int async_size; /* do asynchronous readahead when
767 there are only # of pages ahead */
768
769 unsigned int ra_pages; /* Maximum readahead window */
770 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
771 loff_t prev_pos; /* Cache last read() position */
772 };
773
774 /*
775 * Check if @index falls in the readahead windows.
776 */
777 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
778 {
779 return (index >= ra->start &&
780 index < ra->start + ra->size);
781 }
782
783 struct file {
784 union {
785 struct llist_node fu_llist;
786 struct rcu_head fu_rcuhead;
787 } f_u;
788 struct path f_path;
789 struct inode *f_inode; /* cached value */
790 const struct file_operations *f_op;
791
792 /*
793 * Protects f_ep_links, f_flags.
794 * Must not be taken from IRQ context.
795 */
796 spinlock_t f_lock;
797 atomic_long_t f_count;
798 unsigned int f_flags;
799 fmode_t f_mode;
800 struct mutex f_pos_lock;
801 loff_t f_pos;
802 struct fown_struct f_owner;
803 const struct cred *f_cred;
804 struct file_ra_state f_ra;
805
806 u64 f_version;
807 #ifdef CONFIG_SECURITY
808 void *f_security;
809 #endif
810 /* needed for tty driver, and maybe others */
811 void *private_data;
812
813 #ifdef CONFIG_EPOLL
814 /* Used by fs/eventpoll.c to link all the hooks to this file */
815 struct list_head f_ep_links;
816 struct list_head f_tfile_llink;
817 #endif /* #ifdef CONFIG_EPOLL */
818 struct address_space *f_mapping;
819 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
820
821 struct file_handle {
822 __u32 handle_bytes;
823 int handle_type;
824 /* file identifier */
825 unsigned char f_handle[0];
826 };
827
828 static inline struct file *get_file(struct file *f)
829 {
830 atomic_long_inc(&f->f_count);
831 return f;
832 }
833 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
834 #define file_count(x) atomic_long_read(&(x)->f_count)
835
836 #define MAX_NON_LFS ((1UL<<31) - 1)
837
838 /* Page cache limit. The filesystems should put that into their s_maxbytes
839 limits, otherwise bad things can happen in VM. */
840 #if BITS_PER_LONG==32
841 #define MAX_LFS_FILESIZE (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
842 #elif BITS_PER_LONG==64
843 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
844 #endif
845
846 #define FL_POSIX 1
847 #define FL_FLOCK 2
848 #define FL_DELEG 4 /* NFSv4 delegation */
849 #define FL_ACCESS 8 /* not trying to lock, just looking */
850 #define FL_EXISTS 16 /* when unlocking, test for existence */
851 #define FL_LEASE 32 /* lease held on this file */
852 #define FL_CLOSE 64 /* unlock on close */
853 #define FL_SLEEP 128 /* A blocking lock */
854 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
855 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
856 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
857
858 /*
859 * Special return value from posix_lock_file() and vfs_lock_file() for
860 * asynchronous locking.
861 */
862 #define FILE_LOCK_DEFERRED 1
863
864 /* legacy typedef, should eventually be removed */
865 typedef void *fl_owner_t;
866
867 struct file_lock_operations {
868 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
869 void (*fl_release_private)(struct file_lock *);
870 };
871
872 struct lock_manager_operations {
873 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
874 unsigned long (*lm_owner_key)(struct file_lock *);
875 void (*lm_get_owner)(struct file_lock *, struct file_lock *);
876 void (*lm_put_owner)(struct file_lock *);
877 void (*lm_notify)(struct file_lock *); /* unblock callback */
878 int (*lm_grant)(struct file_lock *, int);
879 bool (*lm_break)(struct file_lock *);
880 int (*lm_change)(struct file_lock **, int, struct list_head *);
881 void (*lm_setup)(struct file_lock *, void **);
882 };
883
884 struct lock_manager {
885 struct list_head list;
886 };
887
888 struct net;
889 void locks_start_grace(struct net *, struct lock_manager *);
890 void locks_end_grace(struct lock_manager *);
891 int locks_in_grace(struct net *);
892
893 /* that will die - we need it for nfs_lock_info */
894 #include <linux/nfs_fs_i.h>
895
896 /*
897 * struct file_lock represents a generic "file lock". It's used to represent
898 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
899 * note that the same struct is used to represent both a request for a lock and
900 * the lock itself, but the same object is never used for both.
901 *
902 * FIXME: should we create a separate "struct lock_request" to help distinguish
903 * these two uses?
904 *
905 * The i_flock list is ordered by:
906 *
907 * 1) lock type -- FL_LEASEs first, then FL_FLOCK, and finally FL_POSIX
908 * 2) lock owner
909 * 3) lock range start
910 * 4) lock range end
911 *
912 * Obviously, the last two criteria only matter for POSIX locks.
913 */
914 struct file_lock {
915 struct file_lock *fl_next; /* singly linked list for this inode */
916 struct hlist_node fl_link; /* node in global lists */
917 struct list_head fl_block; /* circular list of blocked processes */
918 fl_owner_t fl_owner;
919 unsigned int fl_flags;
920 unsigned char fl_type;
921 unsigned int fl_pid;
922 int fl_link_cpu; /* what cpu's list is this on? */
923 struct pid *fl_nspid;
924 wait_queue_head_t fl_wait;
925 struct file *fl_file;
926 loff_t fl_start;
927 loff_t fl_end;
928
929 struct fasync_struct * fl_fasync; /* for lease break notifications */
930 /* for lease breaks: */
931 unsigned long fl_break_time;
932 unsigned long fl_downgrade_time;
933
934 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
935 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
936 union {
937 struct nfs_lock_info nfs_fl;
938 struct nfs4_lock_info nfs4_fl;
939 struct {
940 struct list_head link; /* link in AFS vnode's pending_locks list */
941 int state; /* state of grant or error if -ve */
942 } afs;
943 } fl_u;
944 };
945
946 /* The following constant reflects the upper bound of the file/locking space */
947 #ifndef OFFSET_MAX
948 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
949 #define OFFSET_MAX INT_LIMIT(loff_t)
950 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
951 #endif
952
953 #include <linux/fcntl.h>
954
955 extern void send_sigio(struct fown_struct *fown, int fd, int band);
956
957 #ifdef CONFIG_FILE_LOCKING
958 extern int fcntl_getlk(struct file *, unsigned int, struct flock __user *);
959 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
960 struct flock __user *);
961
962 #if BITS_PER_LONG == 32
963 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 __user *);
964 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
965 struct flock64 __user *);
966 #endif
967
968 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
969 extern int fcntl_getlease(struct file *filp);
970
971 /* fs/locks.c */
972 void locks_free_lock(struct file_lock *fl);
973 extern void locks_init_lock(struct file_lock *);
974 extern struct file_lock * locks_alloc_lock(void);
975 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
976 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
977 extern void locks_remove_posix(struct file *, fl_owner_t);
978 extern void locks_remove_file(struct file *);
979 extern void locks_release_private(struct file_lock *);
980 extern void posix_test_lock(struct file *, struct file_lock *);
981 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
982 extern int posix_lock_file_wait(struct file *, struct file_lock *);
983 extern int posix_unblock_lock(struct file_lock *);
984 extern int vfs_test_lock(struct file *, struct file_lock *);
985 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
986 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
987 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
988 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
989 extern void lease_get_mtime(struct inode *, struct timespec *time);
990 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
991 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
992 extern int lease_modify(struct file_lock **, int, struct list_head *);
993 #else /* !CONFIG_FILE_LOCKING */
994 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
995 struct flock __user *user)
996 {
997 return -EINVAL;
998 }
999
1000 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1001 unsigned int cmd, struct flock __user *user)
1002 {
1003 return -EACCES;
1004 }
1005
1006 #if BITS_PER_LONG == 32
1007 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1008 struct flock64 __user *user)
1009 {
1010 return -EINVAL;
1011 }
1012
1013 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1014 unsigned int cmd, struct flock64 __user *user)
1015 {
1016 return -EACCES;
1017 }
1018 #endif
1019 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1020 {
1021 return -EINVAL;
1022 }
1023
1024 static inline int fcntl_getlease(struct file *filp)
1025 {
1026 return F_UNLCK;
1027 }
1028
1029 static inline void locks_init_lock(struct file_lock *fl)
1030 {
1031 return;
1032 }
1033
1034 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1035 {
1036 return;
1037 }
1038
1039 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1040 {
1041 return;
1042 }
1043
1044 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1045 {
1046 return;
1047 }
1048
1049 static inline void locks_remove_file(struct file *filp)
1050 {
1051 return;
1052 }
1053
1054 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1055 {
1056 return;
1057 }
1058
1059 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1060 struct file_lock *conflock)
1061 {
1062 return -ENOLCK;
1063 }
1064
1065 static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1066 {
1067 return -ENOLCK;
1068 }
1069
1070 static inline int posix_unblock_lock(struct file_lock *waiter)
1071 {
1072 return -ENOENT;
1073 }
1074
1075 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1076 {
1077 return 0;
1078 }
1079
1080 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1081 struct file_lock *fl, struct file_lock *conf)
1082 {
1083 return -ENOLCK;
1084 }
1085
1086 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1087 {
1088 return 0;
1089 }
1090
1091 static inline int flock_lock_file_wait(struct file *filp,
1092 struct file_lock *request)
1093 {
1094 return -ENOLCK;
1095 }
1096
1097 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1098 {
1099 return 0;
1100 }
1101
1102 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1103 {
1104 return;
1105 }
1106
1107 static inline int generic_setlease(struct file *filp, long arg,
1108 struct file_lock **flp, void **priv)
1109 {
1110 return -EINVAL;
1111 }
1112
1113 static inline int vfs_setlease(struct file *filp, long arg,
1114 struct file_lock **lease, void **priv)
1115 {
1116 return -EINVAL;
1117 }
1118
1119 static inline int lease_modify(struct file_lock **before, int arg,
1120 struct list_head *dispose)
1121 {
1122 return -EINVAL;
1123 }
1124 #endif /* !CONFIG_FILE_LOCKING */
1125
1126
1127 struct fasync_struct {
1128 spinlock_t fa_lock;
1129 int magic;
1130 int fa_fd;
1131 struct fasync_struct *fa_next; /* singly linked list */
1132 struct file *fa_file;
1133 struct rcu_head fa_rcu;
1134 };
1135
1136 #define FASYNC_MAGIC 0x4601
1137
1138 /* SMP safe fasync helpers: */
1139 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1140 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1141 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1142 extern struct fasync_struct *fasync_alloc(void);
1143 extern void fasync_free(struct fasync_struct *);
1144
1145 /* can be called from interrupts */
1146 extern void kill_fasync(struct fasync_struct **, int, int);
1147
1148 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1149 extern void f_setown(struct file *filp, unsigned long arg, int force);
1150 extern void f_delown(struct file *filp);
1151 extern pid_t f_getown(struct file *filp);
1152 extern int send_sigurg(struct fown_struct *fown);
1153
1154 struct mm_struct;
1155
1156 /*
1157 * Umount options
1158 */
1159
1160 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1161 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1162 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1163 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1164 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1165
1166 extern struct list_head super_blocks;
1167 extern spinlock_t sb_lock;
1168
1169 /* Possible states of 'frozen' field */
1170 enum {
1171 SB_UNFROZEN = 0, /* FS is unfrozen */
1172 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1173 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1174 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1175 * internal threads if needed) */
1176 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1177 };
1178
1179 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1180
1181 struct sb_writers {
1182 /* Counters for counting writers at each level */
1183 struct percpu_counter counter[SB_FREEZE_LEVELS];
1184 wait_queue_head_t wait; /* queue for waiting for
1185 writers / faults to finish */
1186 int frozen; /* Is sb frozen? */
1187 wait_queue_head_t wait_unfrozen; /* queue for waiting for
1188 sb to be thawed */
1189 #ifdef CONFIG_DEBUG_LOCK_ALLOC
1190 struct lockdep_map lock_map[SB_FREEZE_LEVELS];
1191 #endif
1192 };
1193
1194 struct super_block {
1195 struct list_head s_list; /* Keep this first */
1196 dev_t s_dev; /* search index; _not_ kdev_t */
1197 unsigned char s_blocksize_bits;
1198 unsigned long s_blocksize;
1199 loff_t s_maxbytes; /* Max file size */
1200 struct file_system_type *s_type;
1201 const struct super_operations *s_op;
1202 const struct dquot_operations *dq_op;
1203 const struct quotactl_ops *s_qcop;
1204 const struct export_operations *s_export_op;
1205 unsigned long s_flags;
1206 unsigned long s_magic;
1207 struct dentry *s_root;
1208 struct rw_semaphore s_umount;
1209 int s_count;
1210 atomic_t s_active;
1211 #ifdef CONFIG_SECURITY
1212 void *s_security;
1213 #endif
1214 const struct xattr_handler **s_xattr;
1215
1216 struct list_head s_inodes; /* all inodes */
1217 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1218 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1219 struct block_device *s_bdev;
1220 struct backing_dev_info *s_bdi;
1221 struct mtd_info *s_mtd;
1222 struct hlist_node s_instances;
1223 unsigned int s_quota_types; /* Bitmask of supported quota types */
1224 struct quota_info s_dquot; /* Diskquota specific options */
1225
1226 struct sb_writers s_writers;
1227
1228 char s_id[32]; /* Informational name */
1229 u8 s_uuid[16]; /* UUID */
1230
1231 void *s_fs_info; /* Filesystem private info */
1232 unsigned int s_max_links;
1233 fmode_t s_mode;
1234
1235 /* Granularity of c/m/atime in ns.
1236 Cannot be worse than a second */
1237 u32 s_time_gran;
1238
1239 /*
1240 * The next field is for VFS *only*. No filesystems have any business
1241 * even looking at it. You had been warned.
1242 */
1243 struct mutex s_vfs_rename_mutex; /* Kludge */
1244
1245 /*
1246 * Filesystem subtype. If non-empty the filesystem type field
1247 * in /proc/mounts will be "type.subtype"
1248 */
1249 char *s_subtype;
1250
1251 /*
1252 * Saved mount options for lazy filesystems using
1253 * generic_show_options()
1254 */
1255 char __rcu *s_options;
1256 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1257
1258 /*
1259 * Saved pool identifier for cleancache (-1 means none)
1260 */
1261 int cleancache_poolid;
1262
1263 struct shrinker s_shrink; /* per-sb shrinker handle */
1264
1265 /* Number of inodes with nlink == 0 but still referenced */
1266 atomic_long_t s_remove_count;
1267
1268 /* Being remounted read-only */
1269 int s_readonly_remount;
1270
1271 /* AIO completions deferred from interrupt context */
1272 struct workqueue_struct *s_dio_done_wq;
1273 struct hlist_head s_pins;
1274
1275 /*
1276 * Keep the lru lists last in the structure so they always sit on their
1277 * own individual cachelines.
1278 */
1279 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1280 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
1281 struct rcu_head rcu;
1282
1283 /*
1284 * Indicates how deep in a filesystem stack this SB is
1285 */
1286 int s_stack_depth;
1287 };
1288
1289 extern struct timespec current_fs_time(struct super_block *sb);
1290
1291 /*
1292 * Snapshotting support.
1293 */
1294
1295 void __sb_end_write(struct super_block *sb, int level);
1296 int __sb_start_write(struct super_block *sb, int level, bool wait);
1297
1298 /**
1299 * sb_end_write - drop write access to a superblock
1300 * @sb: the super we wrote to
1301 *
1302 * Decrement number of writers to the filesystem. Wake up possible waiters
1303 * wanting to freeze the filesystem.
1304 */
1305 static inline void sb_end_write(struct super_block *sb)
1306 {
1307 __sb_end_write(sb, SB_FREEZE_WRITE);
1308 }
1309
1310 /**
1311 * sb_end_pagefault - drop write access to a superblock from a page fault
1312 * @sb: the super we wrote to
1313 *
1314 * Decrement number of processes handling write page fault to the filesystem.
1315 * Wake up possible waiters wanting to freeze the filesystem.
1316 */
1317 static inline void sb_end_pagefault(struct super_block *sb)
1318 {
1319 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1320 }
1321
1322 /**
1323 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1324 * @sb: the super we wrote to
1325 *
1326 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1327 * waiters wanting to freeze the filesystem.
1328 */
1329 static inline void sb_end_intwrite(struct super_block *sb)
1330 {
1331 __sb_end_write(sb, SB_FREEZE_FS);
1332 }
1333
1334 /**
1335 * sb_start_write - get write access to a superblock
1336 * @sb: the super we write to
1337 *
1338 * When a process wants to write data or metadata to a file system (i.e. dirty
1339 * a page or an inode), it should embed the operation in a sb_start_write() -
1340 * sb_end_write() pair to get exclusion against file system freezing. This
1341 * function increments number of writers preventing freezing. If the file
1342 * system is already frozen, the function waits until the file system is
1343 * thawed.
1344 *
1345 * Since freeze protection behaves as a lock, users have to preserve
1346 * ordering of freeze protection and other filesystem locks. Generally,
1347 * freeze protection should be the outermost lock. In particular, we have:
1348 *
1349 * sb_start_write
1350 * -> i_mutex (write path, truncate, directory ops, ...)
1351 * -> s_umount (freeze_super, thaw_super)
1352 */
1353 static inline void sb_start_write(struct super_block *sb)
1354 {
1355 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1356 }
1357
1358 static inline int sb_start_write_trylock(struct super_block *sb)
1359 {
1360 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1361 }
1362
1363 /**
1364 * sb_start_pagefault - get write access to a superblock from a page fault
1365 * @sb: the super we write to
1366 *
1367 * When a process starts handling write page fault, it should embed the
1368 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1369 * exclusion against file system freezing. This is needed since the page fault
1370 * is going to dirty a page. This function increments number of running page
1371 * faults preventing freezing. If the file system is already frozen, the
1372 * function waits until the file system is thawed.
1373 *
1374 * Since page fault freeze protection behaves as a lock, users have to preserve
1375 * ordering of freeze protection and other filesystem locks. It is advised to
1376 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1377 * handling code implies lock dependency:
1378 *
1379 * mmap_sem
1380 * -> sb_start_pagefault
1381 */
1382 static inline void sb_start_pagefault(struct super_block *sb)
1383 {
1384 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1385 }
1386
1387 /*
1388 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1389 * @sb: the super we write to
1390 *
1391 * This is the third level of protection against filesystem freezing. It is
1392 * free for use by a filesystem. The only requirement is that it must rank
1393 * below sb_start_pagefault.
1394 *
1395 * For example filesystem can call sb_start_intwrite() when starting a
1396 * transaction which somewhat eases handling of freezing for internal sources
1397 * of filesystem changes (internal fs threads, discarding preallocation on file
1398 * close, etc.).
1399 */
1400 static inline void sb_start_intwrite(struct super_block *sb)
1401 {
1402 __sb_start_write(sb, SB_FREEZE_FS, true);
1403 }
1404
1405
1406 extern bool inode_owner_or_capable(const struct inode *inode);
1407
1408 /*
1409 * VFS helper functions..
1410 */
1411 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1412 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1413 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1414 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1415 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1416 extern int vfs_rmdir(struct inode *, struct dentry *);
1417 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1418 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1419 extern int vfs_whiteout(struct inode *, struct dentry *);
1420
1421 /*
1422 * VFS dentry helper functions.
1423 */
1424 extern void dentry_unhash(struct dentry *dentry);
1425
1426 /*
1427 * VFS file helper functions.
1428 */
1429 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1430 umode_t mode);
1431 /*
1432 * VFS FS_IOC_FIEMAP helper definitions.
1433 */
1434 struct fiemap_extent_info {
1435 unsigned int fi_flags; /* Flags as passed from user */
1436 unsigned int fi_extents_mapped; /* Number of mapped extents */
1437 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1438 struct fiemap_extent __user *fi_extents_start; /* Start of
1439 fiemap_extent array */
1440 };
1441 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1442 u64 phys, u64 len, u32 flags);
1443 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1444
1445 /*
1446 * File types
1447 *
1448 * NOTE! These match bits 12..15 of stat.st_mode
1449 * (ie "(i_mode >> 12) & 15").
1450 */
1451 #define DT_UNKNOWN 0
1452 #define DT_FIFO 1
1453 #define DT_CHR 2
1454 #define DT_DIR 4
1455 #define DT_BLK 6
1456 #define DT_REG 8
1457 #define DT_LNK 10
1458 #define DT_SOCK 12
1459 #define DT_WHT 14
1460
1461 /*
1462 * This is the "filldir" function type, used by readdir() to let
1463 * the kernel specify what kind of dirent layout it wants to have.
1464 * This allows the kernel to read directories into kernel space or
1465 * to have different dirent layouts depending on the binary type.
1466 */
1467 struct dir_context;
1468 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1469 unsigned);
1470
1471 struct dir_context {
1472 const filldir_t actor;
1473 loff_t pos;
1474 };
1475
1476 struct block_device_operations;
1477
1478 /* These macros are for out of kernel modules to test that
1479 * the kernel supports the unlocked_ioctl and compat_ioctl
1480 * fields in struct file_operations. */
1481 #define HAVE_COMPAT_IOCTL 1
1482 #define HAVE_UNLOCKED_IOCTL 1
1483
1484 struct iov_iter;
1485
1486 struct file_operations {
1487 struct module *owner;
1488 loff_t (*llseek) (struct file *, loff_t, int);
1489 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1490 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1491 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1492 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1493 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1494 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1495 int (*iterate) (struct file *, struct dir_context *);
1496 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1497 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1498 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1499 int (*mmap) (struct file *, struct vm_area_struct *);
1500 int (*open) (struct inode *, struct file *);
1501 int (*flush) (struct file *, fl_owner_t id);
1502 int (*release) (struct inode *, struct file *);
1503 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1504 int (*aio_fsync) (struct kiocb *, int datasync);
1505 int (*fasync) (int, struct file *, int);
1506 int (*lock) (struct file *, int, struct file_lock *);
1507 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1508 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1509 int (*check_flags)(int);
1510 int (*flock) (struct file *, int, struct file_lock *);
1511 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1512 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1513 int (*setlease)(struct file *, long, struct file_lock **, void **);
1514 long (*fallocate)(struct file *file, int mode, loff_t offset,
1515 loff_t len);
1516 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1517 };
1518
1519 struct inode_operations {
1520 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1521 void * (*follow_link) (struct dentry *, struct nameidata *);
1522 int (*permission) (struct inode *, int);
1523 struct posix_acl * (*get_acl)(struct inode *, int);
1524
1525 int (*readlink) (struct dentry *, char __user *,int);
1526 void (*put_link) (struct dentry *, struct nameidata *, void *);
1527
1528 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1529 int (*link) (struct dentry *,struct inode *,struct dentry *);
1530 int (*unlink) (struct inode *,struct dentry *);
1531 int (*symlink) (struct inode *,struct dentry *,const char *);
1532 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1533 int (*rmdir) (struct inode *,struct dentry *);
1534 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1535 int (*rename) (struct inode *, struct dentry *,
1536 struct inode *, struct dentry *);
1537 int (*rename2) (struct inode *, struct dentry *,
1538 struct inode *, struct dentry *, unsigned int);
1539 int (*setattr) (struct dentry *, struct iattr *);
1540 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1541 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1542 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1543 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1544 int (*removexattr) (struct dentry *, const char *);
1545 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1546 u64 len);
1547 int (*update_time)(struct inode *, struct timespec *, int);
1548 int (*atomic_open)(struct inode *, struct dentry *,
1549 struct file *, unsigned open_flag,
1550 umode_t create_mode, int *opened);
1551 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1552 int (*set_acl)(struct inode *, struct posix_acl *, int);
1553
1554 /* WARNING: probably going away soon, do not use! */
1555 int (*dentry_open)(struct dentry *, struct file *, const struct cred *);
1556 } ____cacheline_aligned;
1557
1558 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1559 unsigned long nr_segs, unsigned long fast_segs,
1560 struct iovec *fast_pointer,
1561 struct iovec **ret_pointer);
1562
1563 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1564 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1565 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1566 unsigned long, loff_t *);
1567 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1568 unsigned long, loff_t *);
1569
1570 struct super_operations {
1571 struct inode *(*alloc_inode)(struct super_block *sb);
1572 void (*destroy_inode)(struct inode *);
1573
1574 void (*dirty_inode) (struct inode *, int flags);
1575 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1576 int (*drop_inode) (struct inode *);
1577 void (*evict_inode) (struct inode *);
1578 void (*put_super) (struct super_block *);
1579 int (*sync_fs)(struct super_block *sb, int wait);
1580 int (*freeze_super) (struct super_block *);
1581 int (*freeze_fs) (struct super_block *);
1582 int (*thaw_super) (struct super_block *);
1583 int (*unfreeze_fs) (struct super_block *);
1584 int (*statfs) (struct dentry *, struct kstatfs *);
1585 int (*remount_fs) (struct super_block *, int *, char *);
1586 void (*umount_begin) (struct super_block *);
1587
1588 int (*show_options)(struct seq_file *, struct dentry *);
1589 int (*show_devname)(struct seq_file *, struct dentry *);
1590 int (*show_path)(struct seq_file *, struct dentry *);
1591 int (*show_stats)(struct seq_file *, struct dentry *);
1592 #ifdef CONFIG_QUOTA
1593 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1594 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1595 struct dquot **(*get_dquots)(struct inode *);
1596 #endif
1597 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1598 long (*nr_cached_objects)(struct super_block *, int);
1599 long (*free_cached_objects)(struct super_block *, long, int);
1600 };
1601
1602 /*
1603 * Inode flags - they have no relation to superblock flags now
1604 */
1605 #define S_SYNC 1 /* Writes are synced at once */
1606 #define S_NOATIME 2 /* Do not update access times */
1607 #define S_APPEND 4 /* Append-only file */
1608 #define S_IMMUTABLE 8 /* Immutable file */
1609 #define S_DEAD 16 /* removed, but still open directory */
1610 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1611 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1612 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1613 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1614 #define S_PRIVATE 512 /* Inode is fs-internal */
1615 #define S_IMA 1024 /* Inode has an associated IMA struct */
1616 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1617 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1618
1619 /*
1620 * Note that nosuid etc flags are inode-specific: setting some file-system
1621 * flags just means all the inodes inherit those flags by default. It might be
1622 * possible to override it selectively if you really wanted to with some
1623 * ioctl() that is not currently implemented.
1624 *
1625 * Exception: MS_RDONLY is always applied to the entire file system.
1626 *
1627 * Unfortunately, it is possible to change a filesystems flags with it mounted
1628 * with files in use. This means that all of the inodes will not have their
1629 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1630 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1631 */
1632 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1633
1634 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1635 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1636 ((inode)->i_flags & S_SYNC))
1637 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1638 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1639 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1640 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1641 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1642
1643 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1644 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1645 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1646 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1647
1648 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1649 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1650 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1651 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1652 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1653 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1654 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1655
1656 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1657 (inode)->i_rdev == WHITEOUT_DEV)
1658
1659 /*
1660 * Inode state bits. Protected by inode->i_lock
1661 *
1662 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1663 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1664 *
1665 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1666 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1667 * various stages of removing an inode.
1668 *
1669 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1670 *
1671 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1672 * fdatasync(). i_atime is the usual cause.
1673 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1674 * these changes separately from I_DIRTY_SYNC so that we
1675 * don't have to write inode on fdatasync() when only
1676 * mtime has changed in it.
1677 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1678 * I_NEW Serves as both a mutex and completion notification.
1679 * New inodes set I_NEW. If two processes both create
1680 * the same inode, one of them will release its inode and
1681 * wait for I_NEW to be released before returning.
1682 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1683 * also cause waiting on I_NEW, without I_NEW actually
1684 * being set. find_inode() uses this to prevent returning
1685 * nearly-dead inodes.
1686 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1687 * is zero. I_FREEING must be set when I_WILL_FREE is
1688 * cleared.
1689 * I_FREEING Set when inode is about to be freed but still has dirty
1690 * pages or buffers attached or the inode itself is still
1691 * dirty.
1692 * I_CLEAR Added by clear_inode(). In this state the inode is
1693 * clean and can be destroyed. Inode keeps I_FREEING.
1694 *
1695 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1696 * prohibited for many purposes. iget() must wait for
1697 * the inode to be completely released, then create it
1698 * anew. Other functions will just ignore such inodes,
1699 * if appropriate. I_NEW is used for waiting.
1700 *
1701 * I_SYNC Writeback of inode is running. The bit is set during
1702 * data writeback, and cleared with a wakeup on the bit
1703 * address once it is done. The bit is also used to pin
1704 * the inode in memory for flusher thread.
1705 *
1706 * I_REFERENCED Marks the inode as recently references on the LRU list.
1707 *
1708 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1709 *
1710 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1711 */
1712 #define I_DIRTY_SYNC (1 << 0)
1713 #define I_DIRTY_DATASYNC (1 << 1)
1714 #define I_DIRTY_PAGES (1 << 2)
1715 #define __I_NEW 3
1716 #define I_NEW (1 << __I_NEW)
1717 #define I_WILL_FREE (1 << 4)
1718 #define I_FREEING (1 << 5)
1719 #define I_CLEAR (1 << 6)
1720 #define __I_SYNC 7
1721 #define I_SYNC (1 << __I_SYNC)
1722 #define I_REFERENCED (1 << 8)
1723 #define __I_DIO_WAKEUP 9
1724 #define I_DIO_WAKEUP (1 << I_DIO_WAKEUP)
1725 #define I_LINKABLE (1 << 10)
1726
1727 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1728
1729 extern void __mark_inode_dirty(struct inode *, int);
1730 static inline void mark_inode_dirty(struct inode *inode)
1731 {
1732 __mark_inode_dirty(inode, I_DIRTY);
1733 }
1734
1735 static inline void mark_inode_dirty_sync(struct inode *inode)
1736 {
1737 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1738 }
1739
1740 extern void inc_nlink(struct inode *inode);
1741 extern void drop_nlink(struct inode *inode);
1742 extern void clear_nlink(struct inode *inode);
1743 extern void set_nlink(struct inode *inode, unsigned int nlink);
1744
1745 static inline void inode_inc_link_count(struct inode *inode)
1746 {
1747 inc_nlink(inode);
1748 mark_inode_dirty(inode);
1749 }
1750
1751 static inline void inode_dec_link_count(struct inode *inode)
1752 {
1753 drop_nlink(inode);
1754 mark_inode_dirty(inode);
1755 }
1756
1757 /**
1758 * inode_inc_iversion - increments i_version
1759 * @inode: inode that need to be updated
1760 *
1761 * Every time the inode is modified, the i_version field will be incremented.
1762 * The filesystem has to be mounted with i_version flag
1763 */
1764
1765 static inline void inode_inc_iversion(struct inode *inode)
1766 {
1767 spin_lock(&inode->i_lock);
1768 inode->i_version++;
1769 spin_unlock(&inode->i_lock);
1770 }
1771
1772 enum file_time_flags {
1773 S_ATIME = 1,
1774 S_MTIME = 2,
1775 S_CTIME = 4,
1776 S_VERSION = 8,
1777 };
1778
1779 extern void touch_atime(const struct path *);
1780 static inline void file_accessed(struct file *file)
1781 {
1782 if (!(file->f_flags & O_NOATIME))
1783 touch_atime(&file->f_path);
1784 }
1785
1786 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1787 int sync_inode_metadata(struct inode *inode, int wait);
1788
1789 struct file_system_type {
1790 const char *name;
1791 int fs_flags;
1792 #define FS_REQUIRES_DEV 1
1793 #define FS_BINARY_MOUNTDATA 2
1794 #define FS_HAS_SUBTYPE 4
1795 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
1796 #define FS_USERNS_DEV_MOUNT 16 /* A userns mount does not imply MNT_NODEV */
1797 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
1798 struct dentry *(*mount) (struct file_system_type *, int,
1799 const char *, void *);
1800 void (*kill_sb) (struct super_block *);
1801 struct module *owner;
1802 struct file_system_type * next;
1803 struct hlist_head fs_supers;
1804
1805 struct lock_class_key s_lock_key;
1806 struct lock_class_key s_umount_key;
1807 struct lock_class_key s_vfs_rename_key;
1808 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
1809
1810 struct lock_class_key i_lock_key;
1811 struct lock_class_key i_mutex_key;
1812 struct lock_class_key i_mutex_dir_key;
1813 };
1814
1815 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
1816
1817 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1818 void *data, int (*fill_super)(struct super_block *, void *, int));
1819 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1820 int flags, const char *dev_name, void *data,
1821 int (*fill_super)(struct super_block *, void *, int));
1822 extern struct dentry *mount_single(struct file_system_type *fs_type,
1823 int flags, void *data,
1824 int (*fill_super)(struct super_block *, void *, int));
1825 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1826 int flags, void *data,
1827 int (*fill_super)(struct super_block *, void *, int));
1828 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
1829 void generic_shutdown_super(struct super_block *sb);
1830 void kill_block_super(struct super_block *sb);
1831 void kill_anon_super(struct super_block *sb);
1832 void kill_litter_super(struct super_block *sb);
1833 void deactivate_super(struct super_block *sb);
1834 void deactivate_locked_super(struct super_block *sb);
1835 int set_anon_super(struct super_block *s, void *data);
1836 int get_anon_bdev(dev_t *);
1837 void free_anon_bdev(dev_t);
1838 struct super_block *sget(struct file_system_type *type,
1839 int (*test)(struct super_block *,void *),
1840 int (*set)(struct super_block *,void *),
1841 int flags, void *data);
1842 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
1843 const struct super_operations *ops,
1844 const struct dentry_operations *dops,
1845 unsigned long);
1846
1847 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1848 #define fops_get(fops) \
1849 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1850 #define fops_put(fops) \
1851 do { if (fops) module_put((fops)->owner); } while(0)
1852 /*
1853 * This one is to be used *ONLY* from ->open() instances.
1854 * fops must be non-NULL, pinned down *and* module dependencies
1855 * should be sufficient to pin the caller down as well.
1856 */
1857 #define replace_fops(f, fops) \
1858 do { \
1859 struct file *__file = (f); \
1860 fops_put(__file->f_op); \
1861 BUG_ON(!(__file->f_op = (fops))); \
1862 } while(0)
1863
1864 extern int register_filesystem(struct file_system_type *);
1865 extern int unregister_filesystem(struct file_system_type *);
1866 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1867 #define kern_mount(type) kern_mount_data(type, NULL)
1868 extern void kern_unmount(struct vfsmount *mnt);
1869 extern int may_umount_tree(struct vfsmount *);
1870 extern int may_umount(struct vfsmount *);
1871 extern long do_mount(const char *, const char __user *,
1872 const char *, unsigned long, void *);
1873 extern struct vfsmount *collect_mounts(struct path *);
1874 extern void drop_collected_mounts(struct vfsmount *);
1875 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1876 struct vfsmount *);
1877 extern int vfs_statfs(struct path *, struct kstatfs *);
1878 extern int user_statfs(const char __user *, struct kstatfs *);
1879 extern int fd_statfs(int, struct kstatfs *);
1880 extern int vfs_ustat(dev_t, struct kstatfs *);
1881 extern int freeze_super(struct super_block *super);
1882 extern int thaw_super(struct super_block *super);
1883 extern bool our_mnt(struct vfsmount *mnt);
1884 extern bool fs_fully_visible(struct file_system_type *);
1885
1886 extern int current_umask(void);
1887
1888 extern void ihold(struct inode * inode);
1889 extern void iput(struct inode *);
1890
1891 static inline struct inode *file_inode(const struct file *f)
1892 {
1893 return f->f_inode;
1894 }
1895
1896 /* /sys/fs */
1897 extern struct kobject *fs_kobj;
1898
1899 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
1900
1901 #define FLOCK_VERIFY_READ 1
1902 #define FLOCK_VERIFY_WRITE 2
1903
1904 #ifdef CONFIG_FILE_LOCKING
1905 extern int locks_mandatory_locked(struct file *);
1906 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1907
1908 /*
1909 * Candidates for mandatory locking have the setgid bit set
1910 * but no group execute bit - an otherwise meaningless combination.
1911 */
1912
1913 static inline int __mandatory_lock(struct inode *ino)
1914 {
1915 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1916 }
1917
1918 /*
1919 * ... and these candidates should be on MS_MANDLOCK mounted fs,
1920 * otherwise these will be advisory locks
1921 */
1922
1923 static inline int mandatory_lock(struct inode *ino)
1924 {
1925 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1926 }
1927
1928 static inline int locks_verify_locked(struct file *file)
1929 {
1930 if (mandatory_lock(file_inode(file)))
1931 return locks_mandatory_locked(file);
1932 return 0;
1933 }
1934
1935 static inline int locks_verify_truncate(struct inode *inode,
1936 struct file *filp,
1937 loff_t size)
1938 {
1939 if (inode->i_flock && mandatory_lock(inode))
1940 return locks_mandatory_area(
1941 FLOCK_VERIFY_WRITE, inode, filp,
1942 size < inode->i_size ? size : inode->i_size,
1943 (size < inode->i_size ? inode->i_size - size
1944 : size - inode->i_size)
1945 );
1946 return 0;
1947 }
1948
1949 static inline int break_lease(struct inode *inode, unsigned int mode)
1950 {
1951 /*
1952 * Since this check is lockless, we must ensure that any refcounts
1953 * taken are done before checking inode->i_flock. Otherwise, we could
1954 * end up racing with tasks trying to set a new lease on this file.
1955 */
1956 smp_mb();
1957 if (inode->i_flock)
1958 return __break_lease(inode, mode, FL_LEASE);
1959 return 0;
1960 }
1961
1962 static inline int break_deleg(struct inode *inode, unsigned int mode)
1963 {
1964 /*
1965 * Since this check is lockless, we must ensure that any refcounts
1966 * taken are done before checking inode->i_flock. Otherwise, we could
1967 * end up racing with tasks trying to set a new lease on this file.
1968 */
1969 smp_mb();
1970 if (inode->i_flock)
1971 return __break_lease(inode, mode, FL_DELEG);
1972 return 0;
1973 }
1974
1975 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
1976 {
1977 int ret;
1978
1979 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
1980 if (ret == -EWOULDBLOCK && delegated_inode) {
1981 *delegated_inode = inode;
1982 ihold(inode);
1983 }
1984 return ret;
1985 }
1986
1987 static inline int break_deleg_wait(struct inode **delegated_inode)
1988 {
1989 int ret;
1990
1991 ret = break_deleg(*delegated_inode, O_WRONLY);
1992 iput(*delegated_inode);
1993 *delegated_inode = NULL;
1994 return ret;
1995 }
1996
1997 #else /* !CONFIG_FILE_LOCKING */
1998 static inline int locks_mandatory_locked(struct file *file)
1999 {
2000 return 0;
2001 }
2002
2003 static inline int locks_mandatory_area(int rw, struct inode *inode,
2004 struct file *filp, loff_t offset,
2005 size_t count)
2006 {
2007 return 0;
2008 }
2009
2010 static inline int __mandatory_lock(struct inode *inode)
2011 {
2012 return 0;
2013 }
2014
2015 static inline int mandatory_lock(struct inode *inode)
2016 {
2017 return 0;
2018 }
2019
2020 static inline int locks_verify_locked(struct file *file)
2021 {
2022 return 0;
2023 }
2024
2025 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2026 size_t size)
2027 {
2028 return 0;
2029 }
2030
2031 static inline int break_lease(struct inode *inode, unsigned int mode)
2032 {
2033 return 0;
2034 }
2035
2036 static inline int break_deleg(struct inode *inode, unsigned int mode)
2037 {
2038 return 0;
2039 }
2040
2041 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2042 {
2043 return 0;
2044 }
2045
2046 static inline int break_deleg_wait(struct inode **delegated_inode)
2047 {
2048 BUG();
2049 return 0;
2050 }
2051
2052 #endif /* CONFIG_FILE_LOCKING */
2053
2054 /* fs/open.c */
2055 struct audit_names;
2056 struct filename {
2057 const char *name; /* pointer to actual string */
2058 const __user char *uptr; /* original userland pointer */
2059 struct audit_names *aname;
2060 bool separate; /* should "name" be freed? */
2061 };
2062
2063 extern long vfs_truncate(struct path *, loff_t);
2064 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2065 struct file *filp);
2066 extern int do_fallocate(struct file *file, int mode, loff_t offset,
2067 loff_t len);
2068 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2069 umode_t mode);
2070 extern struct file *file_open_name(struct filename *, int, umode_t);
2071 extern struct file *filp_open(const char *, int, umode_t);
2072 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2073 const char *, int);
2074 extern int vfs_open(const struct path *, struct file *, const struct cred *);
2075 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2076 extern int filp_close(struct file *, fl_owner_t id);
2077
2078 extern struct filename *getname(const char __user *);
2079 extern struct filename *getname_kernel(const char *);
2080
2081 enum {
2082 FILE_CREATED = 1,
2083 FILE_OPENED = 2
2084 };
2085 extern int finish_open(struct file *file, struct dentry *dentry,
2086 int (*open)(struct inode *, struct file *),
2087 int *opened);
2088 extern int finish_no_open(struct file *file, struct dentry *dentry);
2089
2090 /* fs/ioctl.c */
2091
2092 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2093
2094 /* fs/dcache.c */
2095 extern void __init vfs_caches_init_early(void);
2096 extern void __init vfs_caches_init(unsigned long);
2097
2098 extern struct kmem_cache *names_cachep;
2099
2100 extern void final_putname(struct filename *name);
2101
2102 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2103 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2104 #ifndef CONFIG_AUDITSYSCALL
2105 #define putname(name) final_putname(name)
2106 #else
2107 extern void putname(struct filename *name);
2108 #endif
2109
2110 #ifdef CONFIG_BLOCK
2111 extern int register_blkdev(unsigned int, const char *);
2112 extern void unregister_blkdev(unsigned int, const char *);
2113 extern struct block_device *bdget(dev_t);
2114 extern struct block_device *bdgrab(struct block_device *bdev);
2115 extern void bd_set_size(struct block_device *, loff_t size);
2116 extern void bd_forget(struct inode *inode);
2117 extern void bdput(struct block_device *);
2118 extern void invalidate_bdev(struct block_device *);
2119 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2120 extern int sync_blockdev(struct block_device *bdev);
2121 extern void kill_bdev(struct block_device *);
2122 extern struct super_block *freeze_bdev(struct block_device *);
2123 extern void emergency_thaw_all(void);
2124 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2125 extern int fsync_bdev(struct block_device *);
2126 extern int sb_is_blkdev_sb(struct super_block *sb);
2127 #else
2128 static inline void bd_forget(struct inode *inode) {}
2129 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2130 static inline void kill_bdev(struct block_device *bdev) {}
2131 static inline void invalidate_bdev(struct block_device *bdev) {}
2132
2133 static inline struct super_block *freeze_bdev(struct block_device *sb)
2134 {
2135 return NULL;
2136 }
2137
2138 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2139 {
2140 return 0;
2141 }
2142
2143 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2144 {
2145 }
2146
2147 static inline int sb_is_blkdev_sb(struct super_block *sb)
2148 {
2149 return 0;
2150 }
2151 #endif
2152 extern int sync_filesystem(struct super_block *);
2153 extern const struct file_operations def_blk_fops;
2154 extern const struct file_operations def_chr_fops;
2155 extern const struct file_operations bad_sock_fops;
2156 #ifdef CONFIG_BLOCK
2157 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2158 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2159 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2160 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2161 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2162 void *holder);
2163 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2164 void *holder);
2165 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2166 #ifdef CONFIG_SYSFS
2167 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2168 extern void bd_unlink_disk_holder(struct block_device *bdev,
2169 struct gendisk *disk);
2170 #else
2171 static inline int bd_link_disk_holder(struct block_device *bdev,
2172 struct gendisk *disk)
2173 {
2174 return 0;
2175 }
2176 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2177 struct gendisk *disk)
2178 {
2179 }
2180 #endif
2181 #endif
2182
2183 /* fs/char_dev.c */
2184 #define CHRDEV_MAJOR_HASH_SIZE 255
2185 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2186 extern int register_chrdev_region(dev_t, unsigned, const char *);
2187 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2188 unsigned int count, const char *name,
2189 const struct file_operations *fops);
2190 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2191 unsigned int count, const char *name);
2192 extern void unregister_chrdev_region(dev_t, unsigned);
2193 extern void chrdev_show(struct seq_file *,off_t);
2194
2195 static inline int register_chrdev(unsigned int major, const char *name,
2196 const struct file_operations *fops)
2197 {
2198 return __register_chrdev(major, 0, 256, name, fops);
2199 }
2200
2201 static inline void unregister_chrdev(unsigned int major, const char *name)
2202 {
2203 __unregister_chrdev(major, 0, 256, name);
2204 }
2205
2206 /* fs/block_dev.c */
2207 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2208 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2209
2210 #ifdef CONFIG_BLOCK
2211 #define BLKDEV_MAJOR_HASH_SIZE 255
2212 extern const char *__bdevname(dev_t, char *buffer);
2213 extern const char *bdevname(struct block_device *bdev, char *buffer);
2214 extern struct block_device *lookup_bdev(const char *);
2215 extern void blkdev_show(struct seq_file *,off_t);
2216
2217 #else
2218 #define BLKDEV_MAJOR_HASH_SIZE 0
2219 #endif
2220
2221 extern void init_special_inode(struct inode *, umode_t, dev_t);
2222
2223 /* Invalid inode operations -- fs/bad_inode.c */
2224 extern void make_bad_inode(struct inode *);
2225 extern int is_bad_inode(struct inode *);
2226
2227 #ifdef CONFIG_BLOCK
2228 /*
2229 * return READ, READA, or WRITE
2230 */
2231 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2232
2233 /*
2234 * return data direction, READ or WRITE
2235 */
2236 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2237
2238 extern void check_disk_size_change(struct gendisk *disk,
2239 struct block_device *bdev);
2240 extern int revalidate_disk(struct gendisk *);
2241 extern int check_disk_change(struct block_device *);
2242 extern int __invalidate_device(struct block_device *, bool);
2243 extern int invalidate_partition(struct gendisk *, int);
2244 #endif
2245 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2246 pgoff_t start, pgoff_t end);
2247
2248 static inline void invalidate_remote_inode(struct inode *inode)
2249 {
2250 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2251 S_ISLNK(inode->i_mode))
2252 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2253 }
2254 extern int invalidate_inode_pages2(struct address_space *mapping);
2255 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2256 pgoff_t start, pgoff_t end);
2257 extern int write_inode_now(struct inode *, int);
2258 extern int filemap_fdatawrite(struct address_space *);
2259 extern int filemap_flush(struct address_space *);
2260 extern int filemap_fdatawait(struct address_space *);
2261 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2262 loff_t lend);
2263 extern int filemap_write_and_wait(struct address_space *mapping);
2264 extern int filemap_write_and_wait_range(struct address_space *mapping,
2265 loff_t lstart, loff_t lend);
2266 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2267 loff_t start, loff_t end, int sync_mode);
2268 extern int filemap_fdatawrite_range(struct address_space *mapping,
2269 loff_t start, loff_t end);
2270
2271 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2272 int datasync);
2273 extern int vfs_fsync(struct file *file, int datasync);
2274 static inline int generic_write_sync(struct file *file, loff_t pos, loff_t count)
2275 {
2276 if (!(file->f_flags & O_DSYNC) && !IS_SYNC(file->f_mapping->host))
2277 return 0;
2278 return vfs_fsync_range(file, pos, pos + count - 1,
2279 (file->f_flags & __O_SYNC) ? 0 : 1);
2280 }
2281 extern void emergency_sync(void);
2282 extern void emergency_remount(void);
2283 #ifdef CONFIG_BLOCK
2284 extern sector_t bmap(struct inode *, sector_t);
2285 #endif
2286 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2287 extern int inode_permission(struct inode *, int);
2288 extern int __inode_permission(struct inode *, int);
2289 extern int generic_permission(struct inode *, int);
2290 extern int __check_sticky(struct inode *dir, struct inode *inode);
2291
2292 static inline bool execute_ok(struct inode *inode)
2293 {
2294 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2295 }
2296
2297 static inline void file_start_write(struct file *file)
2298 {
2299 if (!S_ISREG(file_inode(file)->i_mode))
2300 return;
2301 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2302 }
2303
2304 static inline bool file_start_write_trylock(struct file *file)
2305 {
2306 if (!S_ISREG(file_inode(file)->i_mode))
2307 return true;
2308 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2309 }
2310
2311 static inline void file_end_write(struct file *file)
2312 {
2313 if (!S_ISREG(file_inode(file)->i_mode))
2314 return;
2315 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2316 }
2317
2318 /*
2319 * get_write_access() gets write permission for a file.
2320 * put_write_access() releases this write permission.
2321 * This is used for regular files.
2322 * We cannot support write (and maybe mmap read-write shared) accesses and
2323 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2324 * can have the following values:
2325 * 0: no writers, no VM_DENYWRITE mappings
2326 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2327 * > 0: (i_writecount) users are writing to the file.
2328 *
2329 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2330 * except for the cases where we don't hold i_writecount yet. Then we need to
2331 * use {get,deny}_write_access() - these functions check the sign and refuse
2332 * to do the change if sign is wrong.
2333 */
2334 static inline int get_write_access(struct inode *inode)
2335 {
2336 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2337 }
2338 static inline int deny_write_access(struct file *file)
2339 {
2340 struct inode *inode = file_inode(file);
2341 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2342 }
2343 static inline void put_write_access(struct inode * inode)
2344 {
2345 atomic_dec(&inode->i_writecount);
2346 }
2347 static inline void allow_write_access(struct file *file)
2348 {
2349 if (file)
2350 atomic_inc(&file_inode(file)->i_writecount);
2351 }
2352 static inline bool inode_is_open_for_write(const struct inode *inode)
2353 {
2354 return atomic_read(&inode->i_writecount) > 0;
2355 }
2356
2357 #ifdef CONFIG_IMA
2358 static inline void i_readcount_dec(struct inode *inode)
2359 {
2360 BUG_ON(!atomic_read(&inode->i_readcount));
2361 atomic_dec(&inode->i_readcount);
2362 }
2363 static inline void i_readcount_inc(struct inode *inode)
2364 {
2365 atomic_inc(&inode->i_readcount);
2366 }
2367 #else
2368 static inline void i_readcount_dec(struct inode *inode)
2369 {
2370 return;
2371 }
2372 static inline void i_readcount_inc(struct inode *inode)
2373 {
2374 return;
2375 }
2376 #endif
2377 extern int do_pipe_flags(int *, int);
2378
2379 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2380 extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
2381 extern ssize_t __kernel_write(struct file *, const char *, size_t, loff_t *);
2382 extern struct file * open_exec(const char *);
2383
2384 /* fs/dcache.c -- generic fs support functions */
2385 extern int is_subdir(struct dentry *, struct dentry *);
2386 extern int path_is_under(struct path *, struct path *);
2387
2388 #include <linux/err.h>
2389
2390 /* needed for stackable file system support */
2391 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2392
2393 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2394
2395 extern int inode_init_always(struct super_block *, struct inode *);
2396 extern void inode_init_once(struct inode *);
2397 extern void address_space_init_once(struct address_space *mapping);
2398 extern struct inode * igrab(struct inode *);
2399 extern ino_t iunique(struct super_block *, ino_t);
2400 extern int inode_needs_sync(struct inode *inode);
2401 extern int generic_delete_inode(struct inode *inode);
2402 static inline int generic_drop_inode(struct inode *inode)
2403 {
2404 return !inode->i_nlink || inode_unhashed(inode);
2405 }
2406
2407 extern struct inode *ilookup5_nowait(struct super_block *sb,
2408 unsigned long hashval, int (*test)(struct inode *, void *),
2409 void *data);
2410 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2411 int (*test)(struct inode *, void *), void *data);
2412 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2413
2414 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2415 extern struct inode * iget_locked(struct super_block *, unsigned long);
2416 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2417 extern int insert_inode_locked(struct inode *);
2418 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2419 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2420 #else
2421 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2422 #endif
2423 extern void unlock_new_inode(struct inode *);
2424 extern unsigned int get_next_ino(void);
2425
2426 extern void __iget(struct inode * inode);
2427 extern void iget_failed(struct inode *);
2428 extern void clear_inode(struct inode *);
2429 extern void __destroy_inode(struct inode *);
2430 extern struct inode *new_inode_pseudo(struct super_block *sb);
2431 extern struct inode *new_inode(struct super_block *sb);
2432 extern void free_inode_nonrcu(struct inode *inode);
2433 extern int should_remove_suid(struct dentry *);
2434 extern int file_remove_suid(struct file *);
2435
2436 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2437 static inline void insert_inode_hash(struct inode *inode)
2438 {
2439 __insert_inode_hash(inode, inode->i_ino);
2440 }
2441
2442 extern void __remove_inode_hash(struct inode *);
2443 static inline void remove_inode_hash(struct inode *inode)
2444 {
2445 if (!inode_unhashed(inode))
2446 __remove_inode_hash(inode);
2447 }
2448
2449 extern void inode_sb_list_add(struct inode *inode);
2450
2451 #ifdef CONFIG_BLOCK
2452 extern void submit_bio(int, struct bio *);
2453 extern int bdev_read_only(struct block_device *);
2454 #endif
2455 extern int set_blocksize(struct block_device *, int);
2456 extern int sb_set_blocksize(struct super_block *, int);
2457 extern int sb_min_blocksize(struct super_block *, int);
2458
2459 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2460 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2461 extern int generic_file_remap_pages(struct vm_area_struct *, unsigned long addr,
2462 unsigned long size, pgoff_t pgoff);
2463 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
2464 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2465 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2466 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2467 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *, loff_t);
2468 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2469 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2470 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2471 extern ssize_t new_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2472 extern ssize_t new_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2473
2474 /* fs/block_dev.c */
2475 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
2476 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2477 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2478 int datasync);
2479 extern void block_sync_page(struct page *page);
2480
2481 /* fs/splice.c */
2482 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2483 struct pipe_inode_info *, size_t, unsigned int);
2484 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2485 struct pipe_inode_info *, size_t, unsigned int);
2486 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2487 struct file *, loff_t *, size_t, unsigned int);
2488 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2489 struct file *out, loff_t *, size_t len, unsigned int flags);
2490 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2491 loff_t *opos, size_t len, unsigned int flags);
2492
2493
2494 extern void
2495 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2496 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2497 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2498 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
2499 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2500 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2501 int whence, loff_t maxsize, loff_t eof);
2502 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2503 int whence, loff_t size);
2504 extern int generic_file_open(struct inode * inode, struct file * filp);
2505 extern int nonseekable_open(struct inode * inode, struct file * filp);
2506
2507 #ifdef CONFIG_FS_XIP
2508 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2509 loff_t *ppos);
2510 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
2511 extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2512 size_t len, loff_t *ppos);
2513 extern int xip_truncate_page(struct address_space *mapping, loff_t from);
2514 #else
2515 static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2516 {
2517 return 0;
2518 }
2519 #endif
2520
2521 #ifdef CONFIG_BLOCK
2522 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2523 loff_t file_offset);
2524
2525 enum {
2526 /* need locking between buffered and direct access */
2527 DIO_LOCKING = 0x01,
2528
2529 /* filesystem does not support filling holes */
2530 DIO_SKIP_HOLES = 0x02,
2531
2532 /* filesystem can handle aio writes beyond i_size */
2533 DIO_ASYNC_EXTEND = 0x04,
2534 };
2535
2536 void dio_end_io(struct bio *bio, int error);
2537
2538 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2539 struct block_device *bdev, struct iov_iter *iter, loff_t offset,
2540 get_block_t get_block, dio_iodone_t end_io,
2541 dio_submit_t submit_io, int flags);
2542
2543 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
2544 struct inode *inode, struct iov_iter *iter, loff_t offset,
2545 get_block_t get_block)
2546 {
2547 return __blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iter,
2548 offset, get_block, NULL, NULL,
2549 DIO_LOCKING | DIO_SKIP_HOLES);
2550 }
2551 #endif
2552
2553 void inode_dio_wait(struct inode *inode);
2554 void inode_dio_done(struct inode *inode);
2555
2556 extern void inode_set_flags(struct inode *inode, unsigned int flags,
2557 unsigned int mask);
2558
2559 extern const struct file_operations generic_ro_fops;
2560
2561 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2562
2563 extern int readlink_copy(char __user *, int, const char *);
2564 extern int page_readlink(struct dentry *, char __user *, int);
2565 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2566 extern void page_put_link(struct dentry *, struct nameidata *, void *);
2567 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2568 int nofs);
2569 extern int page_symlink(struct inode *inode, const char *symname, int len);
2570 extern const struct inode_operations page_symlink_inode_operations;
2571 extern void kfree_put_link(struct dentry *, struct nameidata *, void *);
2572 extern int generic_readlink(struct dentry *, char __user *, int);
2573 extern void generic_fillattr(struct inode *, struct kstat *);
2574 int vfs_getattr_nosec(struct path *path, struct kstat *stat);
2575 extern int vfs_getattr(struct path *, struct kstat *);
2576 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2577 void inode_add_bytes(struct inode *inode, loff_t bytes);
2578 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
2579 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2580 loff_t inode_get_bytes(struct inode *inode);
2581 void inode_set_bytes(struct inode *inode, loff_t bytes);
2582
2583 extern int vfs_readdir(struct file *, filldir_t, void *);
2584 extern int iterate_dir(struct file *, struct dir_context *);
2585
2586 extern int vfs_stat(const char __user *, struct kstat *);
2587 extern int vfs_lstat(const char __user *, struct kstat *);
2588 extern int vfs_fstat(unsigned int, struct kstat *);
2589 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2590
2591 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2592 unsigned long arg);
2593 extern int __generic_block_fiemap(struct inode *inode,
2594 struct fiemap_extent_info *fieinfo,
2595 loff_t start, loff_t len,
2596 get_block_t *get_block);
2597 extern int generic_block_fiemap(struct inode *inode,
2598 struct fiemap_extent_info *fieinfo, u64 start,
2599 u64 len, get_block_t *get_block);
2600
2601 extern void get_filesystem(struct file_system_type *fs);
2602 extern void put_filesystem(struct file_system_type *fs);
2603 extern struct file_system_type *get_fs_type(const char *name);
2604 extern struct super_block *get_super(struct block_device *);
2605 extern struct super_block *get_super_thawed(struct block_device *);
2606 extern struct super_block *get_active_super(struct block_device *bdev);
2607 extern void drop_super(struct super_block *sb);
2608 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2609 extern void iterate_supers_type(struct file_system_type *,
2610 void (*)(struct super_block *, void *), void *);
2611
2612 extern int dcache_dir_open(struct inode *, struct file *);
2613 extern int dcache_dir_close(struct inode *, struct file *);
2614 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2615 extern int dcache_readdir(struct file *, struct dir_context *);
2616 extern int simple_setattr(struct dentry *, struct iattr *);
2617 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2618 extern int simple_statfs(struct dentry *, struct kstatfs *);
2619 extern int simple_open(struct inode *inode, struct file *file);
2620 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2621 extern int simple_unlink(struct inode *, struct dentry *);
2622 extern int simple_rmdir(struct inode *, struct dentry *);
2623 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2624 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2625 extern int simple_empty(struct dentry *);
2626 extern int simple_readpage(struct file *file, struct page *page);
2627 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2628 loff_t pos, unsigned len, unsigned flags,
2629 struct page **pagep, void **fsdata);
2630 extern int simple_write_end(struct file *file, struct address_space *mapping,
2631 loff_t pos, unsigned len, unsigned copied,
2632 struct page *page, void *fsdata);
2633 extern int always_delete_dentry(const struct dentry *);
2634 extern struct inode *alloc_anon_inode(struct super_block *);
2635 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
2636 extern const struct dentry_operations simple_dentry_operations;
2637
2638 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
2639 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2640 extern const struct file_operations simple_dir_operations;
2641 extern const struct inode_operations simple_dir_inode_operations;
2642 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2643 struct dentry *d_alloc_name(struct dentry *, const char *);
2644 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2645 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2646 extern void simple_release_fs(struct vfsmount **mount, int *count);
2647
2648 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2649 loff_t *ppos, const void *from, size_t available);
2650 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2651 const void __user *from, size_t count);
2652
2653 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
2654 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2655
2656 extern int generic_check_addressable(unsigned, u64);
2657
2658 #ifdef CONFIG_MIGRATION
2659 extern int buffer_migrate_page(struct address_space *,
2660 struct page *, struct page *,
2661 enum migrate_mode);
2662 #else
2663 #define buffer_migrate_page NULL
2664 #endif
2665
2666 extern int inode_change_ok(const struct inode *, struct iattr *);
2667 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2668 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2669
2670 extern int file_update_time(struct file *file);
2671
2672 extern int generic_show_options(struct seq_file *m, struct dentry *root);
2673 extern void save_mount_options(struct super_block *sb, char *options);
2674 extern void replace_mount_options(struct super_block *sb, char *options);
2675
2676 static inline ino_t parent_ino(struct dentry *dentry)
2677 {
2678 ino_t res;
2679
2680 /*
2681 * Don't strictly need d_lock here? If the parent ino could change
2682 * then surely we'd have a deeper race in the caller?
2683 */
2684 spin_lock(&dentry->d_lock);
2685 res = dentry->d_parent->d_inode->i_ino;
2686 spin_unlock(&dentry->d_lock);
2687 return res;
2688 }
2689
2690 /* Transaction based IO helpers */
2691
2692 /*
2693 * An argresp is stored in an allocated page and holds the
2694 * size of the argument or response, along with its content
2695 */
2696 struct simple_transaction_argresp {
2697 ssize_t size;
2698 char data[0];
2699 };
2700
2701 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2702
2703 char *simple_transaction_get(struct file *file, const char __user *buf,
2704 size_t size);
2705 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2706 size_t size, loff_t *pos);
2707 int simple_transaction_release(struct inode *inode, struct file *file);
2708
2709 void simple_transaction_set(struct file *file, size_t n);
2710
2711 /*
2712 * simple attribute files
2713 *
2714 * These attributes behave similar to those in sysfs:
2715 *
2716 * Writing to an attribute immediately sets a value, an open file can be
2717 * written to multiple times.
2718 *
2719 * Reading from an attribute creates a buffer from the value that might get
2720 * read with multiple read calls. When the attribute has been read
2721 * completely, no further read calls are possible until the file is opened
2722 * again.
2723 *
2724 * All attributes contain a text representation of a numeric value
2725 * that are accessed with the get() and set() functions.
2726 */
2727 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2728 static int __fops ## _open(struct inode *inode, struct file *file) \
2729 { \
2730 __simple_attr_check_format(__fmt, 0ull); \
2731 return simple_attr_open(inode, file, __get, __set, __fmt); \
2732 } \
2733 static const struct file_operations __fops = { \
2734 .owner = THIS_MODULE, \
2735 .open = __fops ## _open, \
2736 .release = simple_attr_release, \
2737 .read = simple_attr_read, \
2738 .write = simple_attr_write, \
2739 .llseek = generic_file_llseek, \
2740 }
2741
2742 static inline __printf(1, 2)
2743 void __simple_attr_check_format(const char *fmt, ...)
2744 {
2745 /* don't do anything, just let the compiler check the arguments; */
2746 }
2747
2748 int simple_attr_open(struct inode *inode, struct file *file,
2749 int (*get)(void *, u64 *), int (*set)(void *, u64),
2750 const char *fmt);
2751 int simple_attr_release(struct inode *inode, struct file *file);
2752 ssize_t simple_attr_read(struct file *file, char __user *buf,
2753 size_t len, loff_t *ppos);
2754 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2755 size_t len, loff_t *ppos);
2756
2757 struct ctl_table;
2758 int proc_nr_files(struct ctl_table *table, int write,
2759 void __user *buffer, size_t *lenp, loff_t *ppos);
2760 int proc_nr_dentry(struct ctl_table *table, int write,
2761 void __user *buffer, size_t *lenp, loff_t *ppos);
2762 int proc_nr_inodes(struct ctl_table *table, int write,
2763 void __user *buffer, size_t *lenp, loff_t *ppos);
2764 int __init get_filesystem_list(char *buf);
2765
2766 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
2767 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
2768
2769 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2770 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2771 (flag & __FMODE_NONOTIFY)))
2772
2773 static inline int is_sxid(umode_t mode)
2774 {
2775 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
2776 }
2777
2778 static inline int check_sticky(struct inode *dir, struct inode *inode)
2779 {
2780 if (!(dir->i_mode & S_ISVTX))
2781 return 0;
2782
2783 return __check_sticky(dir, inode);
2784 }
2785
2786 static inline void inode_has_no_xattr(struct inode *inode)
2787 {
2788 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
2789 inode->i_flags |= S_NOSEC;
2790 }
2791
2792 static inline bool is_root_inode(struct inode *inode)
2793 {
2794 return inode == inode->i_sb->s_root->d_inode;
2795 }
2796
2797 static inline bool dir_emit(struct dir_context *ctx,
2798 const char *name, int namelen,
2799 u64 ino, unsigned type)
2800 {
2801 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
2802 }
2803 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
2804 {
2805 return ctx->actor(ctx, ".", 1, ctx->pos,
2806 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
2807 }
2808 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
2809 {
2810 return ctx->actor(ctx, "..", 2, ctx->pos,
2811 parent_ino(file->f_path.dentry), DT_DIR) == 0;
2812 }
2813 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
2814 {
2815 if (ctx->pos == 0) {
2816 if (!dir_emit_dot(file, ctx))
2817 return false;
2818 ctx->pos = 1;
2819 }
2820 if (ctx->pos == 1) {
2821 if (!dir_emit_dotdot(file, ctx))
2822 return false;
2823 ctx->pos = 2;
2824 }
2825 return true;
2826 }
2827 static inline bool dir_relax(struct inode *inode)
2828 {
2829 mutex_unlock(&inode->i_mutex);
2830 mutex_lock(&inode->i_mutex);
2831 return !IS_DEADDIR(inode);
2832 }
2833
2834 #endif /* _LINUX_FS_H */