Version:  2.0.40 2.2.26 2.4.37 2.6.39 3.0 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 3.10 3.11 3.12 3.13 3.14 3.15

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

This page was automatically generated by LXR 0.3.1 (source).  •  Linux is a registered trademark of Linus Torvalds  •  Contact us