Version:  2.0.40 2.2.26 2.4.37 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 3.16 3.17 3.18

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

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