Version:  2.0.40 2.2.26 2.4.37 3.0 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 3.10 3.11 3.12 3.13 3.14 3.15 3.16

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

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