Version:  2.0.40 2.2.26 2.4.37 3.4 3.5 3.6 3.7 3.8 3.9 3.10 3.11 3.12 3.13 3.14 3.15 3.16 3.17 3.18 3.19 4.0

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/rwsem.h>
 22 #include <linux/capability.h>
 23 #include <linux/semaphore.h>
 24 #include <linux/fiemap.h>
 25 #include <linux/rculist_bl.h>
 26 #include <linux/atomic.h>
 27 #include <linux/shrinker.h>
 28 #include <linux/migrate_mode.h>
 29 #include <linux/uidgid.h>
 30 #include <linux/lockdep.h>
 31 #include <linux/percpu-rwsem.h>
 32 #include <linux/blk_types.h>
 33 
 34 #include <asm/byteorder.h>
 35 #include <uapi/linux/fs.h>
 36 
 37 struct backing_dev_info;
 38 struct export_operations;
 39 struct hd_geometry;
 40 struct iovec;
 41 struct nameidata;
 42 struct kiocb;
 43 struct kobject;
 44 struct pipe_inode_info;
 45 struct poll_table_struct;
 46 struct kstatfs;
 47 struct vm_area_struct;
 48 struct vfsmount;
 49 struct cred;
 50 struct swap_info_struct;
 51 struct seq_file;
 52 struct workqueue_struct;
 53 struct iov_iter;
 54 struct vm_fault;
 55 
 56 extern void __init inode_init(void);
 57 extern void __init inode_init_early(void);
 58 extern void __init files_init(unsigned long);
 59 
 60 extern struct files_stat_struct files_stat;
 61 extern unsigned long get_max_files(void);
 62 extern int sysctl_nr_open;
 63 extern struct inodes_stat_t inodes_stat;
 64 extern int leases_enable, lease_break_time;
 65 extern int sysctl_protected_symlinks;
 66 extern int sysctl_protected_hardlinks;
 67 
 68 struct buffer_head;
 69 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
 70                         struct buffer_head *bh_result, int create);
 71 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
 72                         ssize_t bytes, void *private);
 73 
 74 #define MAY_EXEC                0x00000001
 75 #define MAY_WRITE               0x00000002
 76 #define MAY_READ                0x00000004
 77 #define MAY_APPEND              0x00000008
 78 #define MAY_ACCESS              0x00000010
 79 #define MAY_OPEN                0x00000020
 80 #define MAY_CHDIR               0x00000040
 81 /* called from RCU mode, don't block */
 82 #define MAY_NOT_BLOCK           0x00000080
 83 
 84 /*
 85  * flags in file.f_mode.  Note that FMODE_READ and FMODE_WRITE must correspond
 86  * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
 87  */
 88 
 89 /* file is open for reading */
 90 #define FMODE_READ              ((__force fmode_t)0x1)
 91 /* file is open for writing */
 92 #define FMODE_WRITE             ((__force fmode_t)0x2)
 93 /* file is seekable */
 94 #define FMODE_LSEEK             ((__force fmode_t)0x4)
 95 /* file can be accessed using pread */
 96 #define FMODE_PREAD             ((__force fmode_t)0x8)
 97 /* file can be accessed using pwrite */
 98 #define FMODE_PWRITE            ((__force fmode_t)0x10)
 99 /* File is opened for execution with sys_execve / sys_uselib */
100 #define FMODE_EXEC              ((__force fmode_t)0x20)
101 /* File is opened with O_NDELAY (only set for block devices) */
102 #define FMODE_NDELAY            ((__force fmode_t)0x40)
103 /* File is opened with O_EXCL (only set for block devices) */
104 #define FMODE_EXCL              ((__force fmode_t)0x80)
105 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
106    (specialy hack for floppy.c) */
107 #define FMODE_WRITE_IOCTL       ((__force fmode_t)0x100)
108 /* 32bit hashes as llseek() offset (for directories) */
109 #define FMODE_32BITHASH         ((__force fmode_t)0x200)
110 /* 64bit hashes as llseek() offset (for directories) */
111 #define FMODE_64BITHASH         ((__force fmode_t)0x400)
112 
113 /*
114  * Don't update ctime and mtime.
115  *
116  * Currently a special hack for the XFS open_by_handle ioctl, but we'll
117  * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
118  */
119 #define FMODE_NOCMTIME          ((__force fmode_t)0x800)
120 
121 /* Expect random access pattern */
122 #define FMODE_RANDOM            ((__force fmode_t)0x1000)
123 
124 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
125 #define FMODE_UNSIGNED_OFFSET   ((__force fmode_t)0x2000)
126 
127 /* File is opened with O_PATH; almost nothing can be done with it */
128 #define FMODE_PATH              ((__force fmode_t)0x4000)
129 
130 /* File needs atomic accesses to f_pos */
131 #define FMODE_ATOMIC_POS        ((__force fmode_t)0x8000)
132 /* Write access to underlying fs */
133 #define FMODE_WRITER            ((__force fmode_t)0x10000)
134 /* Has read method(s) */
135 #define FMODE_CAN_READ          ((__force fmode_t)0x20000)
136 /* Has write method(s) */
137 #define FMODE_CAN_WRITE         ((__force fmode_t)0x40000)
138 
139 /* File was opened by fanotify and shouldn't generate fanotify events */
140 #define FMODE_NONOTIFY          ((__force fmode_t)0x4000000)
141 
142 /*
143  * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
144  * that indicates that they should check the contents of the iovec are
145  * valid, but not check the memory that the iovec elements
146  * points too.
147  */
148 #define CHECK_IOVEC_ONLY -1
149 
150 /*
151  * The below are the various read and write types that we support. Some of
152  * them include behavioral modifiers that send information down to the
153  * block layer and IO scheduler. Terminology:
154  *
155  *      The block layer uses device plugging to defer IO a little bit, in
156  *      the hope that we will see more IO very shortly. This increases
157  *      coalescing of adjacent IO and thus reduces the number of IOs we
158  *      have to send to the device. It also allows for better queuing,
159  *      if the IO isn't mergeable. If the caller is going to be waiting
160  *      for the IO, then he must ensure that the device is unplugged so
161  *      that the IO is dispatched to the driver.
162  *
163  *      All IO is handled async in Linux. This is fine for background
164  *      writes, but for reads or writes that someone waits for completion
165  *      on, we want to notify the block layer and IO scheduler so that they
166  *      know about it. That allows them to make better scheduling
167  *      decisions. So when the below references 'sync' and 'async', it
168  *      is referencing this priority hint.
169  *
170  * With that in mind, the available types are:
171  *
172  * READ                 A normal read operation. Device will be plugged.
173  * READ_SYNC            A synchronous read. Device is not plugged, caller can
174  *                      immediately wait on this read without caring about
175  *                      unplugging.
176  * READA                Used for read-ahead operations. Lower priority, and the
177  *                      block layer could (in theory) choose to ignore this
178  *                      request if it runs into resource problems.
179  * WRITE                A normal async write. Device will be plugged.
180  * WRITE_SYNC           Synchronous write. Identical to WRITE, but passes down
181  *                      the hint that someone will be waiting on this IO
182  *                      shortly. The write equivalent of READ_SYNC.
183  * WRITE_ODIRECT        Special case write for O_DIRECT only.
184  * WRITE_FLUSH          Like WRITE_SYNC but with preceding cache flush.
185  * WRITE_FUA            Like WRITE_SYNC but data is guaranteed to be on
186  *                      non-volatile media on completion.
187  * WRITE_FLUSH_FUA      Combination of WRITE_FLUSH and FUA. The IO is preceded
188  *                      by a cache flush and data is guaranteed to be on
189  *                      non-volatile media on completion.
190  *
191  */
192 #define RW_MASK                 REQ_WRITE
193 #define RWA_MASK                REQ_RAHEAD
194 
195 #define READ                    0
196 #define WRITE                   RW_MASK
197 #define READA                   RWA_MASK
198 
199 #define READ_SYNC               (READ | REQ_SYNC)
200 #define WRITE_SYNC              (WRITE | REQ_SYNC | REQ_NOIDLE)
201 #define WRITE_ODIRECT           (WRITE | REQ_SYNC)
202 #define WRITE_FLUSH             (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
203 #define WRITE_FUA               (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
204 #define WRITE_FLUSH_FUA         (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
205 
206 /*
207  * Attribute flags.  These should be or-ed together to figure out what
208  * has been changed!
209  */
210 #define ATTR_MODE       (1 << 0)
211 #define ATTR_UID        (1 << 1)
212 #define ATTR_GID        (1 << 2)
213 #define ATTR_SIZE       (1 << 3)
214 #define ATTR_ATIME      (1 << 4)
215 #define ATTR_MTIME      (1 << 5)
216 #define ATTR_CTIME      (1 << 6)
217 #define ATTR_ATIME_SET  (1 << 7)
218 #define ATTR_MTIME_SET  (1 << 8)
219 #define ATTR_FORCE      (1 << 9) /* Not a change, but a change it */
220 #define ATTR_ATTR_FLAG  (1 << 10)
221 #define ATTR_KILL_SUID  (1 << 11)
222 #define ATTR_KILL_SGID  (1 << 12)
223 #define ATTR_FILE       (1 << 13)
224 #define ATTR_KILL_PRIV  (1 << 14)
225 #define ATTR_OPEN       (1 << 15) /* Truncating from open(O_TRUNC) */
226 #define ATTR_TIMES_SET  (1 << 16)
227 
228 /*
229  * Whiteout is represented by a char device.  The following constants define the
230  * mode and device number to use.
231  */
232 #define WHITEOUT_MODE 0
233 #define WHITEOUT_DEV 0
234 
235 /*
236  * This is the Inode Attributes structure, used for notify_change().  It
237  * uses the above definitions as flags, to know which values have changed.
238  * Also, in this manner, a Filesystem can look at only the values it cares
239  * about.  Basically, these are the attributes that the VFS layer can
240  * request to change from the FS layer.
241  *
242  * Derek Atkins <warlord@MIT.EDU> 94-10-20
243  */
244 struct iattr {
245         unsigned int    ia_valid;
246         umode_t         ia_mode;
247         kuid_t          ia_uid;
248         kgid_t          ia_gid;
249         loff_t          ia_size;
250         struct timespec ia_atime;
251         struct timespec ia_mtime;
252         struct timespec ia_ctime;
253 
254         /*
255          * Not an attribute, but an auxiliary info for filesystems wanting to
256          * implement an ftruncate() like method.  NOTE: filesystem should
257          * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
258          */
259         struct file     *ia_file;
260 };
261 
262 /*
263  * Includes for diskquotas.
264  */
265 #include <linux/quota.h>
266 
267 /*
268  * Maximum number of layers of fs stack.  Needs to be limited to
269  * prevent kernel stack overflow
270  */
271 #define FILESYSTEM_MAX_STACK_DEPTH 2
272 
273 /** 
274  * enum positive_aop_returns - aop return codes with specific semantics
275  *
276  * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
277  *                          completed, that the page is still locked, and
278  *                          should be considered active.  The VM uses this hint
279  *                          to return the page to the active list -- it won't
280  *                          be a candidate for writeback again in the near
281  *                          future.  Other callers must be careful to unlock
282  *                          the page if they get this return.  Returned by
283  *                          writepage(); 
284  *
285  * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
286  *                      unlocked it and the page might have been truncated.
287  *                      The caller should back up to acquiring a new page and
288  *                      trying again.  The aop will be taking reasonable
289  *                      precautions not to livelock.  If the caller held a page
290  *                      reference, it should drop it before retrying.  Returned
291  *                      by readpage().
292  *
293  * address_space_operation functions return these large constants to indicate
294  * special semantics to the caller.  These are much larger than the bytes in a
295  * page to allow for functions that return the number of bytes operated on in a
296  * given page.
297  */
298 
299 enum positive_aop_returns {
300         AOP_WRITEPAGE_ACTIVATE  = 0x80000,
301         AOP_TRUNCATED_PAGE      = 0x80001,
302 };
303 
304 #define AOP_FLAG_UNINTERRUPTIBLE        0x0001 /* will not do a short write */
305 #define AOP_FLAG_CONT_EXPAND            0x0002 /* called from cont_expand */
306 #define AOP_FLAG_NOFS                   0x0004 /* used by filesystem to direct
307                                                 * helper code (eg buffer layer)
308                                                 * to clear GFP_FS from alloc */
309 
310 /*
311  * oh the beauties of C type declarations.
312  */
313 struct page;
314 struct address_space;
315 struct writeback_control;
316 
317 /*
318  * "descriptor" for what we're up to with a read.
319  * This allows us to use the same read code yet
320  * have multiple different users of the data that
321  * we read from a file.
322  *
323  * The simplest case just copies the data to user
324  * mode.
325  */
326 typedef struct {
327         size_t written;
328         size_t count;
329         union {
330                 char __user *buf;
331                 void *data;
332         } arg;
333         int error;
334 } read_descriptor_t;
335 
336 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
337                 unsigned long, unsigned long);
338 
339 struct address_space_operations {
340         int (*writepage)(struct page *page, struct writeback_control *wbc);
341         int (*readpage)(struct file *, struct page *);
342 
343         /* Write back some dirty pages from this mapping. */
344         int (*writepages)(struct address_space *, struct writeback_control *);
345 
346         /* Set a page dirty.  Return true if this dirtied it */
347         int (*set_page_dirty)(struct page *page);
348 
349         int (*readpages)(struct file *filp, struct address_space *mapping,
350                         struct list_head *pages, unsigned nr_pages);
351 
352         int (*write_begin)(struct file *, struct address_space *mapping,
353                                 loff_t pos, unsigned len, unsigned flags,
354                                 struct page **pagep, void **fsdata);
355         int (*write_end)(struct file *, struct address_space *mapping,
356                                 loff_t pos, unsigned len, unsigned copied,
357                                 struct page *page, void *fsdata);
358 
359         /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
360         sector_t (*bmap)(struct address_space *, sector_t);
361         void (*invalidatepage) (struct page *, unsigned int, unsigned int);
362         int (*releasepage) (struct page *, gfp_t);
363         void (*freepage)(struct page *);
364         ssize_t (*direct_IO)(int, struct kiocb *, struct iov_iter *iter, loff_t offset);
365         /*
366          * migrate the contents of a page to the specified target. If
367          * migrate_mode is MIGRATE_ASYNC, it must not block.
368          */
369         int (*migratepage) (struct address_space *,
370                         struct page *, struct page *, enum migrate_mode);
371         int (*launder_page) (struct page *);
372         int (*is_partially_uptodate) (struct page *, unsigned long,
373                                         unsigned long);
374         void (*is_dirty_writeback) (struct page *, bool *, bool *);
375         int (*error_remove_page)(struct address_space *, struct page *);
376 
377         /* swapfile support */
378         int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
379                                 sector_t *span);
380         void (*swap_deactivate)(struct file *file);
381 };
382 
383 extern const struct address_space_operations empty_aops;
384 
385 /*
386  * pagecache_write_begin/pagecache_write_end must be used by general code
387  * to write into the pagecache.
388  */
389 int pagecache_write_begin(struct file *, struct address_space *mapping,
390                                 loff_t pos, unsigned len, unsigned flags,
391                                 struct page **pagep, void **fsdata);
392 
393 int pagecache_write_end(struct file *, struct address_space *mapping,
394                                 loff_t pos, unsigned len, unsigned copied,
395                                 struct page *page, void *fsdata);
396 
397 struct address_space {
398         struct inode            *host;          /* owner: inode, block_device */
399         struct radix_tree_root  page_tree;      /* radix tree of all pages */
400         spinlock_t              tree_lock;      /* and lock protecting it */
401         atomic_t                i_mmap_writable;/* count VM_SHARED mappings */
402         struct rb_root          i_mmap;         /* tree of private and shared mappings */
403         struct rw_semaphore     i_mmap_rwsem;   /* protect tree, count, list */
404         /* Protected by tree_lock together with the radix tree */
405         unsigned long           nrpages;        /* number of total pages */
406         unsigned long           nrshadows;      /* number of shadow entries */
407         pgoff_t                 writeback_index;/* writeback starts here */
408         const struct address_space_operations *a_ops;   /* methods */
409         unsigned long           flags;          /* error bits/gfp mask */
410         spinlock_t              private_lock;   /* for use by the address_space */
411         struct list_head        private_list;   /* ditto */
412         void                    *private_data;  /* ditto */
413 } __attribute__((aligned(sizeof(long))));
414         /*
415          * On most architectures that alignment is already the case; but
416          * must be enforced here for CRIS, to let the least significant bit
417          * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
418          */
419 struct request_queue;
420 
421 struct block_device {
422         dev_t                   bd_dev;  /* not a kdev_t - it's a search key */
423         int                     bd_openers;
424         struct inode *          bd_inode;       /* will die */
425         struct super_block *    bd_super;
426         struct mutex            bd_mutex;       /* open/close mutex */
427         struct list_head        bd_inodes;
428         void *                  bd_claiming;
429         void *                  bd_holder;
430         int                     bd_holders;
431         bool                    bd_write_holder;
432 #ifdef CONFIG_SYSFS
433         struct list_head        bd_holder_disks;
434 #endif
435         struct block_device *   bd_contains;
436         unsigned                bd_block_size;
437         struct hd_struct *      bd_part;
438         /* number of times partitions within this device have been opened. */
439         unsigned                bd_part_count;
440         int                     bd_invalidated;
441         struct gendisk *        bd_disk;
442         struct request_queue *  bd_queue;
443         struct list_head        bd_list;
444         /*
445          * Private data.  You must have bd_claim'ed the block_device
446          * to use this.  NOTE:  bd_claim allows an owner to claim
447          * the same device multiple times, the owner must take special
448          * care to not mess up bd_private for that case.
449          */
450         unsigned long           bd_private;
451 
452         /* The counter of freeze processes */
453         int                     bd_fsfreeze_count;
454         /* Mutex for freeze */
455         struct mutex            bd_fsfreeze_mutex;
456 };
457 
458 /*
459  * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
460  * radix trees
461  */
462 #define PAGECACHE_TAG_DIRTY     0
463 #define PAGECACHE_TAG_WRITEBACK 1
464 #define PAGECACHE_TAG_TOWRITE   2
465 
466 int mapping_tagged(struct address_space *mapping, int tag);
467 
468 static inline void i_mmap_lock_write(struct address_space *mapping)
469 {
470         down_write(&mapping->i_mmap_rwsem);
471 }
472 
473 static inline void i_mmap_unlock_write(struct address_space *mapping)
474 {
475         up_write(&mapping->i_mmap_rwsem);
476 }
477 
478 static inline void i_mmap_lock_read(struct address_space *mapping)
479 {
480         down_read(&mapping->i_mmap_rwsem);
481 }
482 
483 static inline void i_mmap_unlock_read(struct address_space *mapping)
484 {
485         up_read(&mapping->i_mmap_rwsem);
486 }
487 
488 /*
489  * Might pages of this file be mapped into userspace?
490  */
491 static inline int mapping_mapped(struct address_space *mapping)
492 {
493         return  !RB_EMPTY_ROOT(&mapping->i_mmap);
494 }
495 
496 /*
497  * Might pages of this file have been modified in userspace?
498  * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
499  * marks vma as VM_SHARED if it is shared, and the file was opened for
500  * writing i.e. vma may be mprotected writable even if now readonly.
501  *
502  * If i_mmap_writable is negative, no new writable mappings are allowed. You
503  * can only deny writable mappings, if none exists right now.
504  */
505 static inline int mapping_writably_mapped(struct address_space *mapping)
506 {
507         return atomic_read(&mapping->i_mmap_writable) > 0;
508 }
509 
510 static inline int mapping_map_writable(struct address_space *mapping)
511 {
512         return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
513                 0 : -EPERM;
514 }
515 
516 static inline void mapping_unmap_writable(struct address_space *mapping)
517 {
518         atomic_dec(&mapping->i_mmap_writable);
519 }
520 
521 static inline int mapping_deny_writable(struct address_space *mapping)
522 {
523         return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
524                 0 : -EBUSY;
525 }
526 
527 static inline void mapping_allow_writable(struct address_space *mapping)
528 {
529         atomic_inc(&mapping->i_mmap_writable);
530 }
531 
532 /*
533  * Use sequence counter to get consistent i_size on 32-bit processors.
534  */
535 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
536 #include <linux/seqlock.h>
537 #define __NEED_I_SIZE_ORDERED
538 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
539 #else
540 #define i_size_ordered_init(inode) do { } while (0)
541 #endif
542 
543 struct posix_acl;
544 #define ACL_NOT_CACHED ((void *)(-1))
545 
546 #define IOP_FASTPERM    0x0001
547 #define IOP_LOOKUP      0x0002
548 #define IOP_NOFOLLOW    0x0004
549 
550 /*
551  * Keep mostly read-only and often accessed (especially for
552  * the RCU path lookup and 'stat' data) fields at the beginning
553  * of the 'struct inode'
554  */
555 struct inode {
556         umode_t                 i_mode;
557         unsigned short          i_opflags;
558         kuid_t                  i_uid;
559         kgid_t                  i_gid;
560         unsigned int            i_flags;
561 
562 #ifdef CONFIG_FS_POSIX_ACL
563         struct posix_acl        *i_acl;
564         struct posix_acl        *i_default_acl;
565 #endif
566 
567         const struct inode_operations   *i_op;
568         struct super_block      *i_sb;
569         struct address_space    *i_mapping;
570 
571 #ifdef CONFIG_SECURITY
572         void                    *i_security;
573 #endif
574 
575         /* Stat data, not accessed from path walking */
576         unsigned long           i_ino;
577         /*
578          * Filesystems may only read i_nlink directly.  They shall use the
579          * following functions for modification:
580          *
581          *    (set|clear|inc|drop)_nlink
582          *    inode_(inc|dec)_link_count
583          */
584         union {
585                 const unsigned int i_nlink;
586                 unsigned int __i_nlink;
587         };
588         dev_t                   i_rdev;
589         loff_t                  i_size;
590         struct timespec         i_atime;
591         struct timespec         i_mtime;
592         struct timespec         i_ctime;
593         spinlock_t              i_lock; /* i_blocks, i_bytes, maybe i_size */
594         unsigned short          i_bytes;
595         unsigned int            i_blkbits;
596         blkcnt_t                i_blocks;
597 
598 #ifdef __NEED_I_SIZE_ORDERED
599         seqcount_t              i_size_seqcount;
600 #endif
601 
602         /* Misc */
603         unsigned long           i_state;
604         struct mutex            i_mutex;
605 
606         unsigned long           dirtied_when;   /* jiffies of first dirtying */
607         unsigned long           dirtied_time_when;
608 
609         struct hlist_node       i_hash;
610         struct list_head        i_wb_list;      /* backing dev IO list */
611         struct list_head        i_lru;          /* inode LRU list */
612         struct list_head        i_sb_list;
613         union {
614                 struct hlist_head       i_dentry;
615                 struct rcu_head         i_rcu;
616         };
617         u64                     i_version;
618         atomic_t                i_count;
619         atomic_t                i_dio_count;
620         atomic_t                i_writecount;
621 #ifdef CONFIG_IMA
622         atomic_t                i_readcount; /* struct files open RO */
623 #endif
624         const struct file_operations    *i_fop; /* former ->i_op->default_file_ops */
625         struct file_lock_context        *i_flctx;
626         struct address_space    i_data;
627         struct list_head        i_devices;
628         union {
629                 struct pipe_inode_info  *i_pipe;
630                 struct block_device     *i_bdev;
631                 struct cdev             *i_cdev;
632         };
633 
634         __u32                   i_generation;
635 
636 #ifdef CONFIG_FSNOTIFY
637         __u32                   i_fsnotify_mask; /* all events this inode cares about */
638         struct hlist_head       i_fsnotify_marks;
639 #endif
640 
641         void                    *i_private; /* fs or device private pointer */
642 };
643 
644 static inline int inode_unhashed(struct inode *inode)
645 {
646         return hlist_unhashed(&inode->i_hash);
647 }
648 
649 /*
650  * inode->i_mutex nesting subclasses for the lock validator:
651  *
652  * 0: the object of the current VFS operation
653  * 1: parent
654  * 2: child/target
655  * 3: xattr
656  * 4: second non-directory
657  * 5: second parent (when locking independent directories in rename)
658  *
659  * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
660  * non-directories at once.
661  *
662  * The locking order between these classes is
663  * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
664  */
665 enum inode_i_mutex_lock_class
666 {
667         I_MUTEX_NORMAL,
668         I_MUTEX_PARENT,
669         I_MUTEX_CHILD,
670         I_MUTEX_XATTR,
671         I_MUTEX_NONDIR2,
672         I_MUTEX_PARENT2,
673 };
674 
675 void lock_two_nondirectories(struct inode *, struct inode*);
676 void unlock_two_nondirectories(struct inode *, struct inode*);
677 
678 /*
679  * NOTE: in a 32bit arch with a preemptable kernel and
680  * an UP compile the i_size_read/write must be atomic
681  * with respect to the local cpu (unlike with preempt disabled),
682  * but they don't need to be atomic with respect to other cpus like in
683  * true SMP (so they need either to either locally disable irq around
684  * the read or for example on x86 they can be still implemented as a
685  * cmpxchg8b without the need of the lock prefix). For SMP compiles
686  * and 64bit archs it makes no difference if preempt is enabled or not.
687  */
688 static inline loff_t i_size_read(const struct inode *inode)
689 {
690 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
691         loff_t i_size;
692         unsigned int seq;
693 
694         do {
695                 seq = read_seqcount_begin(&inode->i_size_seqcount);
696                 i_size = inode->i_size;
697         } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
698         return i_size;
699 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
700         loff_t i_size;
701 
702         preempt_disable();
703         i_size = inode->i_size;
704         preempt_enable();
705         return i_size;
706 #else
707         return inode->i_size;
708 #endif
709 }
710 
711 /*
712  * NOTE: unlike i_size_read(), i_size_write() does need locking around it
713  * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
714  * can be lost, resulting in subsequent i_size_read() calls spinning forever.
715  */
716 static inline void i_size_write(struct inode *inode, loff_t i_size)
717 {
718 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
719         preempt_disable();
720         write_seqcount_begin(&inode->i_size_seqcount);
721         inode->i_size = i_size;
722         write_seqcount_end(&inode->i_size_seqcount);
723         preempt_enable();
724 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
725         preempt_disable();
726         inode->i_size = i_size;
727         preempt_enable();
728 #else
729         inode->i_size = i_size;
730 #endif
731 }
732 
733 /* Helper functions so that in most cases filesystems will
734  * not need to deal directly with kuid_t and kgid_t and can
735  * instead deal with the raw numeric values that are stored
736  * in the filesystem.
737  */
738 static inline uid_t i_uid_read(const struct inode *inode)
739 {
740         return from_kuid(&init_user_ns, inode->i_uid);
741 }
742 
743 static inline gid_t i_gid_read(const struct inode *inode)
744 {
745         return from_kgid(&init_user_ns, inode->i_gid);
746 }
747 
748 static inline void i_uid_write(struct inode *inode, uid_t uid)
749 {
750         inode->i_uid = make_kuid(&init_user_ns, uid);
751 }
752 
753 static inline void i_gid_write(struct inode *inode, gid_t gid)
754 {
755         inode->i_gid = make_kgid(&init_user_ns, gid);
756 }
757 
758 static inline unsigned iminor(const struct inode *inode)
759 {
760         return MINOR(inode->i_rdev);
761 }
762 
763 static inline unsigned imajor(const struct inode *inode)
764 {
765         return MAJOR(inode->i_rdev);
766 }
767 
768 extern struct block_device *I_BDEV(struct inode *inode);
769 
770 struct fown_struct {
771         rwlock_t lock;          /* protects pid, uid, euid fields */
772         struct pid *pid;        /* pid or -pgrp where SIGIO should be sent */
773         enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
774         kuid_t uid, euid;       /* uid/euid of process setting the owner */
775         int signum;             /* posix.1b rt signal to be delivered on IO */
776 };
777 
778 /*
779  * Track a single file's readahead state
780  */
781 struct file_ra_state {
782         pgoff_t start;                  /* where readahead started */
783         unsigned int size;              /* # of readahead pages */
784         unsigned int async_size;        /* do asynchronous readahead when
785                                            there are only # of pages ahead */
786 
787         unsigned int ra_pages;          /* Maximum readahead window */
788         unsigned int mmap_miss;         /* Cache miss stat for mmap accesses */
789         loff_t prev_pos;                /* Cache last read() position */
790 };
791 
792 /*
793  * Check if @index falls in the readahead windows.
794  */
795 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
796 {
797         return (index >= ra->start &&
798                 index <  ra->start + ra->size);
799 }
800 
801 struct file {
802         union {
803                 struct llist_node       fu_llist;
804                 struct rcu_head         fu_rcuhead;
805         } f_u;
806         struct path             f_path;
807         struct inode            *f_inode;       /* cached value */
808         const struct file_operations    *f_op;
809 
810         /*
811          * Protects f_ep_links, f_flags.
812          * Must not be taken from IRQ context.
813          */
814         spinlock_t              f_lock;
815         atomic_long_t           f_count;
816         unsigned int            f_flags;
817         fmode_t                 f_mode;
818         struct mutex            f_pos_lock;
819         loff_t                  f_pos;
820         struct fown_struct      f_owner;
821         const struct cred       *f_cred;
822         struct file_ra_state    f_ra;
823 
824         u64                     f_version;
825 #ifdef CONFIG_SECURITY
826         void                    *f_security;
827 #endif
828         /* needed for tty driver, and maybe others */
829         void                    *private_data;
830 
831 #ifdef CONFIG_EPOLL
832         /* Used by fs/eventpoll.c to link all the hooks to this file */
833         struct list_head        f_ep_links;
834         struct list_head        f_tfile_llink;
835 #endif /* #ifdef CONFIG_EPOLL */
836         struct address_space    *f_mapping;
837 } __attribute__((aligned(4)));  /* lest something weird decides that 2 is OK */
838 
839 struct file_handle {
840         __u32 handle_bytes;
841         int handle_type;
842         /* file identifier */
843         unsigned char f_handle[0];
844 };
845 
846 static inline struct file *get_file(struct file *f)
847 {
848         atomic_long_inc(&f->f_count);
849         return f;
850 }
851 #define fput_atomic(x)  atomic_long_add_unless(&(x)->f_count, -1, 1)
852 #define file_count(x)   atomic_long_read(&(x)->f_count)
853 
854 #define MAX_NON_LFS     ((1UL<<31) - 1)
855 
856 /* Page cache limit. The filesystems should put that into their s_maxbytes 
857    limits, otherwise bad things can happen in VM. */ 
858 #if BITS_PER_LONG==32
859 #define MAX_LFS_FILESIZE        (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1) 
860 #elif BITS_PER_LONG==64
861 #define MAX_LFS_FILESIZE        ((loff_t)0x7fffffffffffffffLL)
862 #endif
863 
864 #define FL_POSIX        1
865 #define FL_FLOCK        2
866 #define FL_DELEG        4       /* NFSv4 delegation */
867 #define FL_ACCESS       8       /* not trying to lock, just looking */
868 #define FL_EXISTS       16      /* when unlocking, test for existence */
869 #define FL_LEASE        32      /* lease held on this file */
870 #define FL_CLOSE        64      /* unlock on close */
871 #define FL_SLEEP        128     /* A blocking lock */
872 #define FL_DOWNGRADE_PENDING    256 /* Lease is being downgraded */
873 #define FL_UNLOCK_PENDING       512 /* Lease is being broken */
874 #define FL_OFDLCK       1024    /* lock is "owned" by struct file */
875 #define FL_LAYOUT       2048    /* outstanding pNFS layout */
876 
877 /*
878  * Special return value from posix_lock_file() and vfs_lock_file() for
879  * asynchronous locking.
880  */
881 #define FILE_LOCK_DEFERRED 1
882 
883 /* legacy typedef, should eventually be removed */
884 typedef void *fl_owner_t;
885 
886 struct file_lock;
887 
888 struct file_lock_operations {
889         void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
890         void (*fl_release_private)(struct file_lock *);
891 };
892 
893 struct lock_manager_operations {
894         int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
895         unsigned long (*lm_owner_key)(struct file_lock *);
896         void (*lm_get_owner)(struct file_lock *, struct file_lock *);
897         void (*lm_put_owner)(struct file_lock *);
898         void (*lm_notify)(struct file_lock *);  /* unblock callback */
899         int (*lm_grant)(struct file_lock *, int);
900         bool (*lm_break)(struct file_lock *);
901         int (*lm_change)(struct file_lock *, int, struct list_head *);
902         void (*lm_setup)(struct file_lock *, void **);
903 };
904 
905 struct lock_manager {
906         struct list_head list;
907 };
908 
909 struct net;
910 void locks_start_grace(struct net *, struct lock_manager *);
911 void locks_end_grace(struct lock_manager *);
912 int locks_in_grace(struct net *);
913 
914 /* that will die - we need it for nfs_lock_info */
915 #include <linux/nfs_fs_i.h>
916 
917 /*
918  * struct file_lock represents a generic "file lock". It's used to represent
919  * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
920  * note that the same struct is used to represent both a request for a lock and
921  * the lock itself, but the same object is never used for both.
922  *
923  * FIXME: should we create a separate "struct lock_request" to help distinguish
924  * these two uses?
925  *
926  * The varous i_flctx lists are ordered by:
927  *
928  * 1) lock owner
929  * 2) lock range start
930  * 3) lock range end
931  *
932  * Obviously, the last two criteria only matter for POSIX locks.
933  */
934 struct file_lock {
935         struct file_lock *fl_next;      /* singly linked list for this inode  */
936         struct list_head fl_list;       /* link into file_lock_context */
937         struct hlist_node fl_link;      /* node in global lists */
938         struct list_head fl_block;      /* circular list of blocked processes */
939         fl_owner_t fl_owner;
940         unsigned int fl_flags;
941         unsigned char fl_type;
942         unsigned int fl_pid;
943         int fl_link_cpu;                /* what cpu's list is this on? */
944         struct pid *fl_nspid;
945         wait_queue_head_t fl_wait;
946         struct file *fl_file;
947         loff_t fl_start;
948         loff_t fl_end;
949 
950         struct fasync_struct *  fl_fasync; /* for lease break notifications */
951         /* for lease breaks: */
952         unsigned long fl_break_time;
953         unsigned long fl_downgrade_time;
954 
955         const struct file_lock_operations *fl_ops;      /* Callbacks for filesystems */
956         const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
957         union {
958                 struct nfs_lock_info    nfs_fl;
959                 struct nfs4_lock_info   nfs4_fl;
960                 struct {
961                         struct list_head link;  /* link in AFS vnode's pending_locks list */
962                         int state;              /* state of grant or error if -ve */
963                 } afs;
964         } fl_u;
965 };
966 
967 struct file_lock_context {
968         spinlock_t              flc_lock;
969         struct list_head        flc_flock;
970         struct list_head        flc_posix;
971         struct list_head        flc_lease;
972 };
973 
974 /* The following constant reflects the upper bound of the file/locking space */
975 #ifndef OFFSET_MAX
976 #define INT_LIMIT(x)    (~((x)1 << (sizeof(x)*8 - 1)))
977 #define OFFSET_MAX      INT_LIMIT(loff_t)
978 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
979 #endif
980 
981 #include <linux/fcntl.h>
982 
983 extern void send_sigio(struct fown_struct *fown, int fd, int band);
984 
985 #ifdef CONFIG_FILE_LOCKING
986 extern int fcntl_getlk(struct file *, unsigned int, struct flock __user *);
987 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
988                         struct flock __user *);
989 
990 #if BITS_PER_LONG == 32
991 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 __user *);
992 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
993                         struct flock64 __user *);
994 #endif
995 
996 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
997 extern int fcntl_getlease(struct file *filp);
998 
999 /* fs/locks.c */
1000 void locks_free_lock_context(struct file_lock_context *ctx);
1001 void locks_free_lock(struct file_lock *fl);
1002 extern void locks_init_lock(struct file_lock *);
1003 extern struct file_lock * locks_alloc_lock(void);
1004 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1005 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1006 extern void locks_remove_posix(struct file *, fl_owner_t);
1007 extern void locks_remove_file(struct file *);
1008 extern void locks_release_private(struct file_lock *);
1009 extern void posix_test_lock(struct file *, struct file_lock *);
1010 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1011 extern int posix_lock_file_wait(struct file *, struct file_lock *);
1012 extern int posix_unblock_lock(struct file_lock *);
1013 extern int vfs_test_lock(struct file *, struct file_lock *);
1014 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1015 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1016 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
1017 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1018 extern void lease_get_mtime(struct inode *, struct timespec *time);
1019 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1020 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1021 extern int lease_modify(struct file_lock *, int, struct list_head *);
1022 #else /* !CONFIG_FILE_LOCKING */
1023 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1024                               struct flock __user *user)
1025 {
1026         return -EINVAL;
1027 }
1028 
1029 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1030                               unsigned int cmd, struct flock __user *user)
1031 {
1032         return -EACCES;
1033 }
1034 
1035 #if BITS_PER_LONG == 32
1036 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1037                                 struct flock64 __user *user)
1038 {
1039         return -EINVAL;
1040 }
1041 
1042 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1043                                 unsigned int cmd, struct flock64 __user *user)
1044 {
1045         return -EACCES;
1046 }
1047 #endif
1048 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1049 {
1050         return -EINVAL;
1051 }
1052 
1053 static inline int fcntl_getlease(struct file *filp)
1054 {
1055         return F_UNLCK;
1056 }
1057 
1058 static inline void
1059 locks_free_lock_context(struct file_lock_context *ctx)
1060 {
1061 }
1062 
1063 static inline void locks_init_lock(struct file_lock *fl)
1064 {
1065         return;
1066 }
1067 
1068 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1069 {
1070         return;
1071 }
1072 
1073 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1074 {
1075         return;
1076 }
1077 
1078 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1079 {
1080         return;
1081 }
1082 
1083 static inline void locks_remove_file(struct file *filp)
1084 {
1085         return;
1086 }
1087 
1088 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1089 {
1090         return;
1091 }
1092 
1093 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1094                                   struct file_lock *conflock)
1095 {
1096         return -ENOLCK;
1097 }
1098 
1099 static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1100 {
1101         return -ENOLCK;
1102 }
1103 
1104 static inline int posix_unblock_lock(struct file_lock *waiter)
1105 {
1106         return -ENOENT;
1107 }
1108 
1109 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1110 {
1111         return 0;
1112 }
1113 
1114 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1115                                 struct file_lock *fl, struct file_lock *conf)
1116 {
1117         return -ENOLCK;
1118 }
1119 
1120 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1121 {
1122         return 0;
1123 }
1124 
1125 static inline int flock_lock_file_wait(struct file *filp,
1126                                        struct file_lock *request)
1127 {
1128         return -ENOLCK;
1129 }
1130 
1131 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1132 {
1133         return 0;
1134 }
1135 
1136 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1137 {
1138         return;
1139 }
1140 
1141 static inline int generic_setlease(struct file *filp, long arg,
1142                                     struct file_lock **flp, void **priv)
1143 {
1144         return -EINVAL;
1145 }
1146 
1147 static inline int vfs_setlease(struct file *filp, long arg,
1148                                struct file_lock **lease, void **priv)
1149 {
1150         return -EINVAL;
1151 }
1152 
1153 static inline int lease_modify(struct file_lock *fl, int arg,
1154                                struct list_head *dispose)
1155 {
1156         return -EINVAL;
1157 }
1158 #endif /* !CONFIG_FILE_LOCKING */
1159 
1160 
1161 struct fasync_struct {
1162         spinlock_t              fa_lock;
1163         int                     magic;
1164         int                     fa_fd;
1165         struct fasync_struct    *fa_next; /* singly linked list */
1166         struct file             *fa_file;
1167         struct rcu_head         fa_rcu;
1168 };
1169 
1170 #define FASYNC_MAGIC 0x4601
1171 
1172 /* SMP safe fasync helpers: */
1173 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1174 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1175 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1176 extern struct fasync_struct *fasync_alloc(void);
1177 extern void fasync_free(struct fasync_struct *);
1178 
1179 /* can be called from interrupts */
1180 extern void kill_fasync(struct fasync_struct **, int, int);
1181 
1182 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1183 extern void f_setown(struct file *filp, unsigned long arg, int force);
1184 extern void f_delown(struct file *filp);
1185 extern pid_t f_getown(struct file *filp);
1186 extern int send_sigurg(struct fown_struct *fown);
1187 
1188 struct mm_struct;
1189 
1190 /*
1191  *      Umount options
1192  */
1193 
1194 #define MNT_FORCE       0x00000001      /* Attempt to forcibily umount */
1195 #define MNT_DETACH      0x00000002      /* Just detach from the tree */
1196 #define MNT_EXPIRE      0x00000004      /* Mark for expiry */
1197 #define UMOUNT_NOFOLLOW 0x00000008      /* Don't follow symlink on umount */
1198 #define UMOUNT_UNUSED   0x80000000      /* Flag guaranteed to be unused */
1199 
1200 
1201 /* Possible states of 'frozen' field */
1202 enum {
1203         SB_UNFROZEN = 0,                /* FS is unfrozen */
1204         SB_FREEZE_WRITE = 1,            /* Writes, dir ops, ioctls frozen */
1205         SB_FREEZE_PAGEFAULT = 2,        /* Page faults stopped as well */
1206         SB_FREEZE_FS = 3,               /* For internal FS use (e.g. to stop
1207                                          * internal threads if needed) */
1208         SB_FREEZE_COMPLETE = 4,         /* ->freeze_fs finished successfully */
1209 };
1210 
1211 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1212 
1213 struct sb_writers {
1214         /* Counters for counting writers at each level */
1215         struct percpu_counter   counter[SB_FREEZE_LEVELS];
1216         wait_queue_head_t       wait;           /* queue for waiting for
1217                                                    writers / faults to finish */
1218         int                     frozen;         /* Is sb frozen? */
1219         wait_queue_head_t       wait_unfrozen;  /* queue for waiting for
1220                                                    sb to be thawed */
1221 #ifdef CONFIG_DEBUG_LOCK_ALLOC
1222         struct lockdep_map      lock_map[SB_FREEZE_LEVELS];
1223 #endif
1224 };
1225 
1226 struct super_block {
1227         struct list_head        s_list;         /* Keep this first */
1228         dev_t                   s_dev;          /* search index; _not_ kdev_t */
1229         unsigned char           s_blocksize_bits;
1230         unsigned long           s_blocksize;
1231         loff_t                  s_maxbytes;     /* Max file size */
1232         struct file_system_type *s_type;
1233         const struct super_operations   *s_op;
1234         const struct dquot_operations   *dq_op;
1235         const struct quotactl_ops       *s_qcop;
1236         const struct export_operations *s_export_op;
1237         unsigned long           s_flags;
1238         unsigned long           s_magic;
1239         struct dentry           *s_root;
1240         struct rw_semaphore     s_umount;
1241         int                     s_count;
1242         atomic_t                s_active;
1243 #ifdef CONFIG_SECURITY
1244         void                    *s_security;
1245 #endif
1246         const struct xattr_handler **s_xattr;
1247 
1248         struct list_head        s_inodes;       /* all inodes */
1249         struct hlist_bl_head    s_anon;         /* anonymous dentries for (nfs) exporting */
1250         struct list_head        s_mounts;       /* list of mounts; _not_ for fs use */
1251         struct block_device     *s_bdev;
1252         struct backing_dev_info *s_bdi;
1253         struct mtd_info         *s_mtd;
1254         struct hlist_node       s_instances;
1255         unsigned int            s_quota_types;  /* Bitmask of supported quota types */
1256         struct quota_info       s_dquot;        /* Diskquota specific options */
1257 
1258         struct sb_writers       s_writers;
1259 
1260         char s_id[32];                          /* Informational name */
1261         u8 s_uuid[16];                          /* UUID */
1262 
1263         void                    *s_fs_info;     /* Filesystem private info */
1264         unsigned int            s_max_links;
1265         fmode_t                 s_mode;
1266 
1267         /* Granularity of c/m/atime in ns.
1268            Cannot be worse than a second */
1269         u32                s_time_gran;
1270 
1271         /*
1272          * The next field is for VFS *only*. No filesystems have any business
1273          * even looking at it. You had been warned.
1274          */
1275         struct mutex s_vfs_rename_mutex;        /* Kludge */
1276 
1277         /*
1278          * Filesystem subtype.  If non-empty the filesystem type field
1279          * in /proc/mounts will be "type.subtype"
1280          */
1281         char *s_subtype;
1282 
1283         /*
1284          * Saved mount options for lazy filesystems using
1285          * generic_show_options()
1286          */
1287         char __rcu *s_options;
1288         const struct dentry_operations *s_d_op; /* default d_op for dentries */
1289 
1290         /*
1291          * Saved pool identifier for cleancache (-1 means none)
1292          */
1293         int cleancache_poolid;
1294 
1295         struct shrinker s_shrink;       /* per-sb shrinker handle */
1296 
1297         /* Number of inodes with nlink == 0 but still referenced */
1298         atomic_long_t s_remove_count;
1299 
1300         /* Being remounted read-only */
1301         int s_readonly_remount;
1302 
1303         /* AIO completions deferred from interrupt context */
1304         struct workqueue_struct *s_dio_done_wq;
1305         struct hlist_head s_pins;
1306 
1307         /*
1308          * Keep the lru lists last in the structure so they always sit on their
1309          * own individual cachelines.
1310          */
1311         struct list_lru         s_dentry_lru ____cacheline_aligned_in_smp;
1312         struct list_lru         s_inode_lru ____cacheline_aligned_in_smp;
1313         struct rcu_head         rcu;
1314 
1315         /*
1316          * Indicates how deep in a filesystem stack this SB is
1317          */
1318         int s_stack_depth;
1319 };
1320 
1321 extern struct timespec current_fs_time(struct super_block *sb);
1322 
1323 /*
1324  * Snapshotting support.
1325  */
1326 
1327 void __sb_end_write(struct super_block *sb, int level);
1328 int __sb_start_write(struct super_block *sb, int level, bool wait);
1329 
1330 /**
1331  * sb_end_write - drop write access to a superblock
1332  * @sb: the super we wrote to
1333  *
1334  * Decrement number of writers to the filesystem. Wake up possible waiters
1335  * wanting to freeze the filesystem.
1336  */
1337 static inline void sb_end_write(struct super_block *sb)
1338 {
1339         __sb_end_write(sb, SB_FREEZE_WRITE);
1340 }
1341 
1342 /**
1343  * sb_end_pagefault - drop write access to a superblock from a page fault
1344  * @sb: the super we wrote to
1345  *
1346  * Decrement number of processes handling write page fault to the filesystem.
1347  * Wake up possible waiters wanting to freeze the filesystem.
1348  */
1349 static inline void sb_end_pagefault(struct super_block *sb)
1350 {
1351         __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1352 }
1353 
1354 /**
1355  * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1356  * @sb: the super we wrote to
1357  *
1358  * Decrement fs-internal number of writers to the filesystem.  Wake up possible
1359  * waiters wanting to freeze the filesystem.
1360  */
1361 static inline void sb_end_intwrite(struct super_block *sb)
1362 {
1363         __sb_end_write(sb, SB_FREEZE_FS);
1364 }
1365 
1366 /**
1367  * sb_start_write - get write access to a superblock
1368  * @sb: the super we write to
1369  *
1370  * When a process wants to write data or metadata to a file system (i.e. dirty
1371  * a page or an inode), it should embed the operation in a sb_start_write() -
1372  * sb_end_write() pair to get exclusion against file system freezing. This
1373  * function increments number of writers preventing freezing. If the file
1374  * system is already frozen, the function waits until the file system is
1375  * thawed.
1376  *
1377  * Since freeze protection behaves as a lock, users have to preserve
1378  * ordering of freeze protection and other filesystem locks. Generally,
1379  * freeze protection should be the outermost lock. In particular, we have:
1380  *
1381  * sb_start_write
1382  *   -> i_mutex                 (write path, truncate, directory ops, ...)
1383  *   -> s_umount                (freeze_super, thaw_super)
1384  */
1385 static inline void sb_start_write(struct super_block *sb)
1386 {
1387         __sb_start_write(sb, SB_FREEZE_WRITE, true);
1388 }
1389 
1390 static inline int sb_start_write_trylock(struct super_block *sb)
1391 {
1392         return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1393 }
1394 
1395 /**
1396  * sb_start_pagefault - get write access to a superblock from a page fault
1397  * @sb: the super we write to
1398  *
1399  * When a process starts handling write page fault, it should embed the
1400  * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1401  * exclusion against file system freezing. This is needed since the page fault
1402  * is going to dirty a page. This function increments number of running page
1403  * faults preventing freezing. If the file system is already frozen, the
1404  * function waits until the file system is thawed.
1405  *
1406  * Since page fault freeze protection behaves as a lock, users have to preserve
1407  * ordering of freeze protection and other filesystem locks. It is advised to
1408  * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1409  * handling code implies lock dependency:
1410  *
1411  * mmap_sem
1412  *   -> sb_start_pagefault
1413  */
1414 static inline void sb_start_pagefault(struct super_block *sb)
1415 {
1416         __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1417 }
1418 
1419 /*
1420  * sb_start_intwrite - get write access to a superblock for internal fs purposes
1421  * @sb: the super we write to
1422  *
1423  * This is the third level of protection against filesystem freezing. It is
1424  * free for use by a filesystem. The only requirement is that it must rank
1425  * below sb_start_pagefault.
1426  *
1427  * For example filesystem can call sb_start_intwrite() when starting a
1428  * transaction which somewhat eases handling of freezing for internal sources
1429  * of filesystem changes (internal fs threads, discarding preallocation on file
1430  * close, etc.).
1431  */
1432 static inline void sb_start_intwrite(struct super_block *sb)
1433 {
1434         __sb_start_write(sb, SB_FREEZE_FS, true);
1435 }
1436 
1437 
1438 extern bool inode_owner_or_capable(const struct inode *inode);
1439 
1440 /*
1441  * VFS helper functions..
1442  */
1443 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1444 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1445 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1446 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1447 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1448 extern int vfs_rmdir(struct inode *, struct dentry *);
1449 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1450 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1451 extern int vfs_whiteout(struct inode *, struct dentry *);
1452 
1453 /*
1454  * VFS dentry helper functions.
1455  */
1456 extern void dentry_unhash(struct dentry *dentry);
1457 
1458 /*
1459  * VFS file helper functions.
1460  */
1461 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1462                         umode_t mode);
1463 /*
1464  * VFS FS_IOC_FIEMAP helper definitions.
1465  */
1466 struct fiemap_extent_info {
1467         unsigned int fi_flags;          /* Flags as passed from user */
1468         unsigned int fi_extents_mapped; /* Number of mapped extents */
1469         unsigned int fi_extents_max;    /* Size of fiemap_extent array */
1470         struct fiemap_extent __user *fi_extents_start; /* Start of
1471                                                         fiemap_extent array */
1472 };
1473 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1474                             u64 phys, u64 len, u32 flags);
1475 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1476 
1477 /*
1478  * File types
1479  *
1480  * NOTE! These match bits 12..15 of stat.st_mode
1481  * (ie "(i_mode >> 12) & 15").
1482  */
1483 #define DT_UNKNOWN      0
1484 #define DT_FIFO         1
1485 #define DT_CHR          2
1486 #define DT_DIR          4
1487 #define DT_BLK          6
1488 #define DT_REG          8
1489 #define DT_LNK          10
1490 #define DT_SOCK         12
1491 #define DT_WHT          14
1492 
1493 /*
1494  * This is the "filldir" function type, used by readdir() to let
1495  * the kernel specify what kind of dirent layout it wants to have.
1496  * This allows the kernel to read directories into kernel space or
1497  * to have different dirent layouts depending on the binary type.
1498  */
1499 struct dir_context;
1500 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1501                          unsigned);
1502 
1503 struct dir_context {
1504         const filldir_t actor;
1505         loff_t pos;
1506 };
1507 
1508 struct block_device_operations;
1509 
1510 /* These macros are for out of kernel modules to test that
1511  * the kernel supports the unlocked_ioctl and compat_ioctl
1512  * fields in struct file_operations. */
1513 #define HAVE_COMPAT_IOCTL 1
1514 #define HAVE_UNLOCKED_IOCTL 1
1515 
1516 /*
1517  * These flags let !MMU mmap() govern direct device mapping vs immediate
1518  * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1519  *
1520  * NOMMU_MAP_COPY:      Copy can be mapped (MAP_PRIVATE)
1521  * NOMMU_MAP_DIRECT:    Can be mapped directly (MAP_SHARED)
1522  * NOMMU_MAP_READ:      Can be mapped for reading
1523  * NOMMU_MAP_WRITE:     Can be mapped for writing
1524  * NOMMU_MAP_EXEC:      Can be mapped for execution
1525  */
1526 #define NOMMU_MAP_COPY          0x00000001
1527 #define NOMMU_MAP_DIRECT        0x00000008
1528 #define NOMMU_MAP_READ          VM_MAYREAD
1529 #define NOMMU_MAP_WRITE         VM_MAYWRITE
1530 #define NOMMU_MAP_EXEC          VM_MAYEXEC
1531 
1532 #define NOMMU_VMFLAGS \
1533         (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1534 
1535 
1536 struct iov_iter;
1537 
1538 struct file_operations {
1539         struct module *owner;
1540         loff_t (*llseek) (struct file *, loff_t, int);
1541         ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1542         ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1543         ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1544         ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1545         ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1546         ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1547         int (*iterate) (struct file *, struct dir_context *);
1548         unsigned int (*poll) (struct file *, struct poll_table_struct *);
1549         long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1550         long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1551         int (*mmap) (struct file *, struct vm_area_struct *);
1552         int (*mremap)(struct file *, struct vm_area_struct *);
1553         int (*open) (struct inode *, struct file *);
1554         int (*flush) (struct file *, fl_owner_t id);
1555         int (*release) (struct inode *, struct file *);
1556         int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1557         int (*aio_fsync) (struct kiocb *, int datasync);
1558         int (*fasync) (int, struct file *, int);
1559         int (*lock) (struct file *, int, struct file_lock *);
1560         ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1561         unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1562         int (*check_flags)(int);
1563         int (*flock) (struct file *, int, struct file_lock *);
1564         ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1565         ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1566         int (*setlease)(struct file *, long, struct file_lock **, void **);
1567         long (*fallocate)(struct file *file, int mode, loff_t offset,
1568                           loff_t len);
1569         void (*show_fdinfo)(struct seq_file *m, struct file *f);
1570 #ifndef CONFIG_MMU
1571         unsigned (*mmap_capabilities)(struct file *);
1572 #endif
1573 };
1574 
1575 struct inode_operations {
1576         struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1577         void * (*follow_link) (struct dentry *, struct nameidata *);
1578         int (*permission) (struct inode *, int);
1579         struct posix_acl * (*get_acl)(struct inode *, int);
1580 
1581         int (*readlink) (struct dentry *, char __user *,int);
1582         void (*put_link) (struct dentry *, struct nameidata *, void *);
1583 
1584         int (*create) (struct inode *,struct dentry *, umode_t, bool);
1585         int (*link) (struct dentry *,struct inode *,struct dentry *);
1586         int (*unlink) (struct inode *,struct dentry *);
1587         int (*symlink) (struct inode *,struct dentry *,const char *);
1588         int (*mkdir) (struct inode *,struct dentry *,umode_t);
1589         int (*rmdir) (struct inode *,struct dentry *);
1590         int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1591         int (*rename) (struct inode *, struct dentry *,
1592                         struct inode *, struct dentry *);
1593         int (*rename2) (struct inode *, struct dentry *,
1594                         struct inode *, struct dentry *, unsigned int);
1595         int (*setattr) (struct dentry *, struct iattr *);
1596         int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1597         int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1598         ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1599         ssize_t (*listxattr) (struct dentry *, char *, size_t);
1600         int (*removexattr) (struct dentry *, const char *);
1601         int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1602                       u64 len);
1603         int (*update_time)(struct inode *, struct timespec *, int);
1604         int (*atomic_open)(struct inode *, struct dentry *,
1605                            struct file *, unsigned open_flag,
1606                            umode_t create_mode, int *opened);
1607         int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1608         int (*set_acl)(struct inode *, struct posix_acl *, int);
1609 
1610         /* WARNING: probably going away soon, do not use! */
1611         int (*dentry_open)(struct dentry *, struct file *, const struct cred *);
1612 } ____cacheline_aligned;
1613 
1614 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1615                               unsigned long nr_segs, unsigned long fast_segs,
1616                               struct iovec *fast_pointer,
1617                               struct iovec **ret_pointer);
1618 
1619 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1620 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1621 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1622 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1623                 unsigned long, loff_t *);
1624 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1625                 unsigned long, loff_t *);
1626 
1627 struct super_operations {
1628         struct inode *(*alloc_inode)(struct super_block *sb);
1629         void (*destroy_inode)(struct inode *);
1630 
1631         void (*dirty_inode) (struct inode *, int flags);
1632         int (*write_inode) (struct inode *, struct writeback_control *wbc);
1633         int (*drop_inode) (struct inode *);
1634         void (*evict_inode) (struct inode *);
1635         void (*put_super) (struct super_block *);
1636         int (*sync_fs)(struct super_block *sb, int wait);
1637         int (*freeze_super) (struct super_block *);
1638         int (*freeze_fs) (struct super_block *);
1639         int (*thaw_super) (struct super_block *);
1640         int (*unfreeze_fs) (struct super_block *);
1641         int (*statfs) (struct dentry *, struct kstatfs *);
1642         int (*remount_fs) (struct super_block *, int *, char *);
1643         void (*umount_begin) (struct super_block *);
1644 
1645         int (*show_options)(struct seq_file *, struct dentry *);
1646         int (*show_devname)(struct seq_file *, struct dentry *);
1647         int (*show_path)(struct seq_file *, struct dentry *);
1648         int (*show_stats)(struct seq_file *, struct dentry *);
1649 #ifdef CONFIG_QUOTA
1650         ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1651         ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1652         struct dquot **(*get_dquots)(struct inode *);
1653 #endif
1654         int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1655         long (*nr_cached_objects)(struct super_block *,
1656                                   struct shrink_control *);
1657         long (*free_cached_objects)(struct super_block *,
1658                                     struct shrink_control *);
1659 };
1660 
1661 /*
1662  * Inode flags - they have no relation to superblock flags now
1663  */
1664 #define S_SYNC          1       /* Writes are synced at once */
1665 #define S_NOATIME       2       /* Do not update access times */
1666 #define S_APPEND        4       /* Append-only file */
1667 #define S_IMMUTABLE     8       /* Immutable file */
1668 #define S_DEAD          16      /* removed, but still open directory */
1669 #define S_NOQUOTA       32      /* Inode is not counted to quota */
1670 #define S_DIRSYNC       64      /* Directory modifications are synchronous */
1671 #define S_NOCMTIME      128     /* Do not update file c/mtime */
1672 #define S_SWAPFILE      256     /* Do not truncate: swapon got its bmaps */
1673 #define S_PRIVATE       512     /* Inode is fs-internal */
1674 #define S_IMA           1024    /* Inode has an associated IMA struct */
1675 #define S_AUTOMOUNT     2048    /* Automount/referral quasi-directory */
1676 #define S_NOSEC         4096    /* no suid or xattr security attributes */
1677 #ifdef CONFIG_FS_DAX
1678 #define S_DAX           8192    /* Direct Access, avoiding the page cache */
1679 #else
1680 #define S_DAX           0       /* Make all the DAX code disappear */
1681 #endif
1682 
1683 /*
1684  * Note that nosuid etc flags are inode-specific: setting some file-system
1685  * flags just means all the inodes inherit those flags by default. It might be
1686  * possible to override it selectively if you really wanted to with some
1687  * ioctl() that is not currently implemented.
1688  *
1689  * Exception: MS_RDONLY is always applied to the entire file system.
1690  *
1691  * Unfortunately, it is possible to change a filesystems flags with it mounted
1692  * with files in use.  This means that all of the inodes will not have their
1693  * i_flags updated.  Hence, i_flags no longer inherit the superblock mount
1694  * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1695  */
1696 #define __IS_FLG(inode, flg)    ((inode)->i_sb->s_flags & (flg))
1697 
1698 #define IS_RDONLY(inode)        ((inode)->i_sb->s_flags & MS_RDONLY)
1699 #define IS_SYNC(inode)          (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1700                                         ((inode)->i_flags & S_SYNC))
1701 #define IS_DIRSYNC(inode)       (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1702                                         ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1703 #define IS_MANDLOCK(inode)      __IS_FLG(inode, MS_MANDLOCK)
1704 #define IS_NOATIME(inode)       __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1705 #define IS_I_VERSION(inode)     __IS_FLG(inode, MS_I_VERSION)
1706 
1707 #define IS_NOQUOTA(inode)       ((inode)->i_flags & S_NOQUOTA)
1708 #define IS_APPEND(inode)        ((inode)->i_flags & S_APPEND)
1709 #define IS_IMMUTABLE(inode)     ((inode)->i_flags & S_IMMUTABLE)
1710 #define IS_POSIXACL(inode)      __IS_FLG(inode, MS_POSIXACL)
1711 
1712 #define IS_DEADDIR(inode)       ((inode)->i_flags & S_DEAD)
1713 #define IS_NOCMTIME(inode)      ((inode)->i_flags & S_NOCMTIME)
1714 #define IS_SWAPFILE(inode)      ((inode)->i_flags & S_SWAPFILE)
1715 #define IS_PRIVATE(inode)       ((inode)->i_flags & S_PRIVATE)
1716 #define IS_IMA(inode)           ((inode)->i_flags & S_IMA)
1717 #define IS_AUTOMOUNT(inode)     ((inode)->i_flags & S_AUTOMOUNT)
1718 #define IS_NOSEC(inode)         ((inode)->i_flags & S_NOSEC)
1719 #define IS_DAX(inode)           ((inode)->i_flags & S_DAX)
1720 
1721 #define IS_WHITEOUT(inode)      (S_ISCHR(inode->i_mode) && \
1722                                  (inode)->i_rdev == WHITEOUT_DEV)
1723 
1724 /*
1725  * Inode state bits.  Protected by inode->i_lock
1726  *
1727  * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1728  * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1729  *
1730  * Four bits define the lifetime of an inode.  Initially, inodes are I_NEW,
1731  * until that flag is cleared.  I_WILL_FREE, I_FREEING and I_CLEAR are set at
1732  * various stages of removing an inode.
1733  *
1734  * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1735  *
1736  * I_DIRTY_SYNC         Inode is dirty, but doesn't have to be written on
1737  *                      fdatasync().  i_atime is the usual cause.
1738  * I_DIRTY_DATASYNC     Data-related inode changes pending. We keep track of
1739  *                      these changes separately from I_DIRTY_SYNC so that we
1740  *                      don't have to write inode on fdatasync() when only
1741  *                      mtime has changed in it.
1742  * I_DIRTY_PAGES        Inode has dirty pages.  Inode itself may be clean.
1743  * I_NEW                Serves as both a mutex and completion notification.
1744  *                      New inodes set I_NEW.  If two processes both create
1745  *                      the same inode, one of them will release its inode and
1746  *                      wait for I_NEW to be released before returning.
1747  *                      Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1748  *                      also cause waiting on I_NEW, without I_NEW actually
1749  *                      being set.  find_inode() uses this to prevent returning
1750  *                      nearly-dead inodes.
1751  * I_WILL_FREE          Must be set when calling write_inode_now() if i_count
1752  *                      is zero.  I_FREEING must be set when I_WILL_FREE is
1753  *                      cleared.
1754  * I_FREEING            Set when inode is about to be freed but still has dirty
1755  *                      pages or buffers attached or the inode itself is still
1756  *                      dirty.
1757  * I_CLEAR              Added by clear_inode().  In this state the inode is
1758  *                      clean and can be destroyed.  Inode keeps I_FREEING.
1759  *
1760  *                      Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1761  *                      prohibited for many purposes.  iget() must wait for
1762  *                      the inode to be completely released, then create it
1763  *                      anew.  Other functions will just ignore such inodes,
1764  *                      if appropriate.  I_NEW is used for waiting.
1765  *
1766  * I_SYNC               Writeback of inode is running. The bit is set during
1767  *                      data writeback, and cleared with a wakeup on the bit
1768  *                      address once it is done. The bit is also used to pin
1769  *                      the inode in memory for flusher thread.
1770  *
1771  * I_REFERENCED         Marks the inode as recently references on the LRU list.
1772  *
1773  * I_DIO_WAKEUP         Never set.  Only used as a key for wait_on_bit().
1774  *
1775  * Q: What is the difference between I_WILL_FREE and I_FREEING?
1776  */
1777 #define I_DIRTY_SYNC            (1 << 0)
1778 #define I_DIRTY_DATASYNC        (1 << 1)
1779 #define I_DIRTY_PAGES           (1 << 2)
1780 #define __I_NEW                 3
1781 #define I_NEW                   (1 << __I_NEW)
1782 #define I_WILL_FREE             (1 << 4)
1783 #define I_FREEING               (1 << 5)
1784 #define I_CLEAR                 (1 << 6)
1785 #define __I_SYNC                7
1786 #define I_SYNC                  (1 << __I_SYNC)
1787 #define I_REFERENCED            (1 << 8)
1788 #define __I_DIO_WAKEUP          9
1789 #define I_DIO_WAKEUP            (1 << I_DIO_WAKEUP)
1790 #define I_LINKABLE              (1 << 10)
1791 #define I_DIRTY_TIME            (1 << 11)
1792 #define __I_DIRTY_TIME_EXPIRED  12
1793 #define I_DIRTY_TIME_EXPIRED    (1 << __I_DIRTY_TIME_EXPIRED)
1794 
1795 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1796 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1797 
1798 extern void __mark_inode_dirty(struct inode *, int);
1799 static inline void mark_inode_dirty(struct inode *inode)
1800 {
1801         __mark_inode_dirty(inode, I_DIRTY);
1802 }
1803 
1804 static inline void mark_inode_dirty_sync(struct inode *inode)
1805 {
1806         __mark_inode_dirty(inode, I_DIRTY_SYNC);
1807 }
1808 
1809 extern void inc_nlink(struct inode *inode);
1810 extern void drop_nlink(struct inode *inode);
1811 extern void clear_nlink(struct inode *inode);
1812 extern void set_nlink(struct inode *inode, unsigned int nlink);
1813 
1814 static inline void inode_inc_link_count(struct inode *inode)
1815 {
1816         inc_nlink(inode);
1817         mark_inode_dirty(inode);
1818 }
1819 
1820 static inline void inode_dec_link_count(struct inode *inode)
1821 {
1822         drop_nlink(inode);
1823         mark_inode_dirty(inode);
1824 }
1825 
1826 /**
1827  * inode_inc_iversion - increments i_version
1828  * @inode: inode that need to be updated
1829  *
1830  * Every time the inode is modified, the i_version field will be incremented.
1831  * The filesystem has to be mounted with i_version flag
1832  */
1833 
1834 static inline void inode_inc_iversion(struct inode *inode)
1835 {
1836        spin_lock(&inode->i_lock);
1837        inode->i_version++;
1838        spin_unlock(&inode->i_lock);
1839 }
1840 
1841 enum file_time_flags {
1842         S_ATIME = 1,
1843         S_MTIME = 2,
1844         S_CTIME = 4,
1845         S_VERSION = 8,
1846 };
1847 
1848 extern void touch_atime(const struct path *);
1849 static inline void file_accessed(struct file *file)
1850 {
1851         if (!(file->f_flags & O_NOATIME))
1852                 touch_atime(&file->f_path);
1853 }
1854 
1855 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1856 int sync_inode_metadata(struct inode *inode, int wait);
1857 
1858 struct file_system_type {
1859         const char *name;
1860         int fs_flags;
1861 #define FS_REQUIRES_DEV         1 
1862 #define FS_BINARY_MOUNTDATA     2
1863 #define FS_HAS_SUBTYPE          4
1864 #define FS_USERNS_MOUNT         8       /* Can be mounted by userns root */
1865 #define FS_USERNS_DEV_MOUNT     16 /* A userns mount does not imply MNT_NODEV */
1866 #define FS_RENAME_DOES_D_MOVE   32768   /* FS will handle d_move() during rename() internally. */
1867         struct dentry *(*mount) (struct file_system_type *, int,
1868                        const char *, void *);
1869         void (*kill_sb) (struct super_block *);
1870         struct module *owner;
1871         struct file_system_type * next;
1872         struct hlist_head fs_supers;
1873 
1874         struct lock_class_key s_lock_key;
1875         struct lock_class_key s_umount_key;
1876         struct lock_class_key s_vfs_rename_key;
1877         struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
1878 
1879         struct lock_class_key i_lock_key;
1880         struct lock_class_key i_mutex_key;
1881         struct lock_class_key i_mutex_dir_key;
1882 };
1883 
1884 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
1885 
1886 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1887         void *data, int (*fill_super)(struct super_block *, void *, int));
1888 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1889         int flags, const char *dev_name, void *data,
1890         int (*fill_super)(struct super_block *, void *, int));
1891 extern struct dentry *mount_single(struct file_system_type *fs_type,
1892         int flags, void *data,
1893         int (*fill_super)(struct super_block *, void *, int));
1894 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1895         int flags, void *data,
1896         int (*fill_super)(struct super_block *, void *, int));
1897 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
1898 void generic_shutdown_super(struct super_block *sb);
1899 void kill_block_super(struct super_block *sb);
1900 void kill_anon_super(struct super_block *sb);
1901 void kill_litter_super(struct super_block *sb);
1902 void deactivate_super(struct super_block *sb);
1903 void deactivate_locked_super(struct super_block *sb);
1904 int set_anon_super(struct super_block *s, void *data);
1905 int get_anon_bdev(dev_t *);
1906 void free_anon_bdev(dev_t);
1907 struct super_block *sget(struct file_system_type *type,
1908                         int (*test)(struct super_block *,void *),
1909                         int (*set)(struct super_block *,void *),
1910                         int flags, void *data);
1911 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
1912         const struct super_operations *ops,
1913         const struct dentry_operations *dops,
1914         unsigned long);
1915 
1916 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1917 #define fops_get(fops) \
1918         (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1919 #define fops_put(fops) \
1920         do { if (fops) module_put((fops)->owner); } while(0)
1921 /*
1922  * This one is to be used *ONLY* from ->open() instances.
1923  * fops must be non-NULL, pinned down *and* module dependencies
1924  * should be sufficient to pin the caller down as well.
1925  */
1926 #define replace_fops(f, fops) \
1927         do {    \
1928                 struct file *__file = (f); \
1929                 fops_put(__file->f_op); \
1930                 BUG_ON(!(__file->f_op = (fops))); \
1931         } while(0)
1932 
1933 extern int register_filesystem(struct file_system_type *);
1934 extern int unregister_filesystem(struct file_system_type *);
1935 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1936 #define kern_mount(type) kern_mount_data(type, NULL)
1937 extern void kern_unmount(struct vfsmount *mnt);
1938 extern int may_umount_tree(struct vfsmount *);
1939 extern int may_umount(struct vfsmount *);
1940 extern long do_mount(const char *, const char __user *,
1941                      const char *, unsigned long, void *);
1942 extern struct vfsmount *collect_mounts(struct path *);
1943 extern void drop_collected_mounts(struct vfsmount *);
1944 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1945                           struct vfsmount *);
1946 extern int vfs_statfs(struct path *, struct kstatfs *);
1947 extern int user_statfs(const char __user *, struct kstatfs *);
1948 extern int fd_statfs(int, struct kstatfs *);
1949 extern int vfs_ustat(dev_t, struct kstatfs *);
1950 extern int freeze_super(struct super_block *super);
1951 extern int thaw_super(struct super_block *super);
1952 extern bool our_mnt(struct vfsmount *mnt);
1953 extern bool fs_fully_visible(struct file_system_type *);
1954 
1955 extern int current_umask(void);
1956 
1957 extern void ihold(struct inode * inode);
1958 extern void iput(struct inode *);
1959 extern int generic_update_time(struct inode *, struct timespec *, int);
1960 
1961 static inline struct inode *file_inode(const struct file *f)
1962 {
1963         return f->f_inode;
1964 }
1965 
1966 /* /sys/fs */
1967 extern struct kobject *fs_kobj;
1968 
1969 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
1970 
1971 #define FLOCK_VERIFY_READ  1
1972 #define FLOCK_VERIFY_WRITE 2
1973 
1974 #ifdef CONFIG_FILE_LOCKING
1975 extern int locks_mandatory_locked(struct file *);
1976 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1977 
1978 /*
1979  * Candidates for mandatory locking have the setgid bit set
1980  * but no group execute bit -  an otherwise meaningless combination.
1981  */
1982 
1983 static inline int __mandatory_lock(struct inode *ino)
1984 {
1985         return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1986 }
1987 
1988 /*
1989  * ... and these candidates should be on MS_MANDLOCK mounted fs,
1990  * otherwise these will be advisory locks
1991  */
1992 
1993 static inline int mandatory_lock(struct inode *ino)
1994 {
1995         return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1996 }
1997 
1998 static inline int locks_verify_locked(struct file *file)
1999 {
2000         if (mandatory_lock(file_inode(file)))
2001                 return locks_mandatory_locked(file);
2002         return 0;
2003 }
2004 
2005 static inline int locks_verify_truncate(struct inode *inode,
2006                                     struct file *filp,
2007                                     loff_t size)
2008 {
2009         if (inode->i_flctx && mandatory_lock(inode))
2010                 return locks_mandatory_area(
2011                         FLOCK_VERIFY_WRITE, inode, filp,
2012                         size < inode->i_size ? size : inode->i_size,
2013                         (size < inode->i_size ? inode->i_size - size
2014                          : size - inode->i_size)
2015                 );
2016         return 0;
2017 }
2018 
2019 static inline int break_lease(struct inode *inode, unsigned int mode)
2020 {
2021         /*
2022          * Since this check is lockless, we must ensure that any refcounts
2023          * taken are done before checking i_flctx->flc_lease. Otherwise, we
2024          * could end up racing with tasks trying to set a new lease on this
2025          * file.
2026          */
2027         smp_mb();
2028         if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2029                 return __break_lease(inode, mode, FL_LEASE);
2030         return 0;
2031 }
2032 
2033 static inline int break_deleg(struct inode *inode, unsigned int mode)
2034 {
2035         /*
2036          * Since this check is lockless, we must ensure that any refcounts
2037          * taken are done before checking i_flctx->flc_lease. Otherwise, we
2038          * could end up racing with tasks trying to set a new lease on this
2039          * file.
2040          */
2041         smp_mb();
2042         if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2043                 return __break_lease(inode, mode, FL_DELEG);
2044         return 0;
2045 }
2046 
2047 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2048 {
2049         int ret;
2050 
2051         ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2052         if (ret == -EWOULDBLOCK && delegated_inode) {
2053                 *delegated_inode = inode;
2054                 ihold(inode);
2055         }
2056         return ret;
2057 }
2058 
2059 static inline int break_deleg_wait(struct inode **delegated_inode)
2060 {
2061         int ret;
2062 
2063         ret = break_deleg(*delegated_inode, O_WRONLY);
2064         iput(*delegated_inode);
2065         *delegated_inode = NULL;
2066         return ret;
2067 }
2068 
2069 static inline int break_layout(struct inode *inode, bool wait)
2070 {
2071         smp_mb();
2072         if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2073                 return __break_lease(inode,
2074                                 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2075                                 FL_LAYOUT);
2076         return 0;
2077 }
2078 
2079 #else /* !CONFIG_FILE_LOCKING */
2080 static inline int locks_mandatory_locked(struct file *file)
2081 {
2082         return 0;
2083 }
2084 
2085 static inline int locks_mandatory_area(int rw, struct inode *inode,
2086                                        struct file *filp, loff_t offset,
2087                                        size_t count)
2088 {
2089         return 0;
2090 }
2091 
2092 static inline int __mandatory_lock(struct inode *inode)
2093 {
2094         return 0;
2095 }
2096 
2097 static inline int mandatory_lock(struct inode *inode)
2098 {
2099         return 0;
2100 }
2101 
2102 static inline int locks_verify_locked(struct file *file)
2103 {
2104         return 0;
2105 }
2106 
2107 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2108                                         size_t size)
2109 {
2110         return 0;
2111 }
2112 
2113 static inline int break_lease(struct inode *inode, unsigned int mode)
2114 {
2115         return 0;
2116 }
2117 
2118 static inline int break_deleg(struct inode *inode, unsigned int mode)
2119 {
2120         return 0;
2121 }
2122 
2123 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2124 {
2125         return 0;
2126 }
2127 
2128 static inline int break_deleg_wait(struct inode **delegated_inode)
2129 {
2130         BUG();
2131         return 0;
2132 }
2133 
2134 static inline int break_layout(struct inode *inode, bool wait)
2135 {
2136         return 0;
2137 }
2138 
2139 #endif /* CONFIG_FILE_LOCKING */
2140 
2141 /* fs/open.c */
2142 struct audit_names;
2143 struct filename {
2144         const char              *name;  /* pointer to actual string */
2145         const __user char       *uptr;  /* original userland pointer */
2146         struct audit_names      *aname;
2147         int                     refcnt;
2148         bool                    separate; /* should "name" be freed? */
2149 };
2150 
2151 extern long vfs_truncate(struct path *, loff_t);
2152 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2153                        struct file *filp);
2154 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2155                         loff_t len);
2156 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2157                         umode_t mode);
2158 extern struct file *file_open_name(struct filename *, int, umode_t);
2159 extern struct file *filp_open(const char *, int, umode_t);
2160 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2161                                    const char *, int);
2162 extern int vfs_open(const struct path *, struct file *, const struct cred *);
2163 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2164 extern int filp_close(struct file *, fl_owner_t id);
2165 
2166 extern struct filename *getname_flags(const char __user *, int, int *);
2167 extern struct filename *getname(const char __user *);
2168 extern struct filename *getname_kernel(const char *);
2169 extern void putname(struct filename *name);
2170 
2171 enum {
2172         FILE_CREATED = 1,
2173         FILE_OPENED = 2
2174 };
2175 extern int finish_open(struct file *file, struct dentry *dentry,
2176                         int (*open)(struct inode *, struct file *),
2177                         int *opened);
2178 extern int finish_no_open(struct file *file, struct dentry *dentry);
2179 
2180 /* fs/ioctl.c */
2181 
2182 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2183 
2184 /* fs/dcache.c */
2185 extern void __init vfs_caches_init_early(void);
2186 extern void __init vfs_caches_init(unsigned long);
2187 
2188 extern struct kmem_cache *names_cachep;
2189 
2190 #define __getname()             kmem_cache_alloc(names_cachep, GFP_KERNEL)
2191 #define __putname(name)         kmem_cache_free(names_cachep, (void *)(name))
2192 
2193 #ifdef CONFIG_BLOCK
2194 extern int register_blkdev(unsigned int, const char *);
2195 extern void unregister_blkdev(unsigned int, const char *);
2196 extern struct block_device *bdget(dev_t);
2197 extern struct block_device *bdgrab(struct block_device *bdev);
2198 extern void bd_set_size(struct block_device *, loff_t size);
2199 extern void bd_forget(struct inode *inode);
2200 extern void bdput(struct block_device *);
2201 extern void invalidate_bdev(struct block_device *);
2202 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2203 extern int sync_blockdev(struct block_device *bdev);
2204 extern void kill_bdev(struct block_device *);
2205 extern struct super_block *freeze_bdev(struct block_device *);
2206 extern void emergency_thaw_all(void);
2207 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2208 extern int fsync_bdev(struct block_device *);
2209 extern int sb_is_blkdev_sb(struct super_block *sb);
2210 #else
2211 static inline void bd_forget(struct inode *inode) {}
2212 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2213 static inline void kill_bdev(struct block_device *bdev) {}
2214 static inline void invalidate_bdev(struct block_device *bdev) {}
2215 
2216 static inline struct super_block *freeze_bdev(struct block_device *sb)
2217 {
2218         return NULL;
2219 }
2220 
2221 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2222 {
2223         return 0;
2224 }
2225 
2226 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2227 {
2228 }
2229 
2230 static inline int sb_is_blkdev_sb(struct super_block *sb)
2231 {
2232         return 0;
2233 }
2234 #endif
2235 extern int sync_filesystem(struct super_block *);
2236 extern const struct file_operations def_blk_fops;
2237 extern const struct file_operations def_chr_fops;
2238 #ifdef CONFIG_BLOCK
2239 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2240 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2241 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2242 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2243 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2244                                                void *holder);
2245 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2246                                               void *holder);
2247 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2248 #ifdef CONFIG_SYSFS
2249 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2250 extern void bd_unlink_disk_holder(struct block_device *bdev,
2251                                   struct gendisk *disk);
2252 #else
2253 static inline int bd_link_disk_holder(struct block_device *bdev,
2254                                       struct gendisk *disk)
2255 {
2256         return 0;
2257 }
2258 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2259                                          struct gendisk *disk)
2260 {
2261 }
2262 #endif
2263 #endif
2264 
2265 /* fs/char_dev.c */
2266 #define CHRDEV_MAJOR_HASH_SIZE  255
2267 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2268 extern int register_chrdev_region(dev_t, unsigned, const char *);
2269 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2270                              unsigned int count, const char *name,
2271                              const struct file_operations *fops);
2272 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2273                                 unsigned int count, const char *name);
2274 extern void unregister_chrdev_region(dev_t, unsigned);
2275 extern void chrdev_show(struct seq_file *,off_t);
2276 
2277 static inline int register_chrdev(unsigned int major, const char *name,
2278                                   const struct file_operations *fops)
2279 {
2280         return __register_chrdev(major, 0, 256, name, fops);
2281 }
2282 
2283 static inline void unregister_chrdev(unsigned int major, const char *name)
2284 {
2285         __unregister_chrdev(major, 0, 256, name);
2286 }
2287 
2288 /* fs/block_dev.c */
2289 #define BDEVNAME_SIZE   32      /* Largest string for a blockdev identifier */
2290 #define BDEVT_SIZE      10      /* Largest string for MAJ:MIN for blkdev */
2291 
2292 #ifdef CONFIG_BLOCK
2293 #define BLKDEV_MAJOR_HASH_SIZE  255
2294 extern const char *__bdevname(dev_t, char *buffer);
2295 extern const char *bdevname(struct block_device *bdev, char *buffer);
2296 extern struct block_device *lookup_bdev(const char *);
2297 extern void blkdev_show(struct seq_file *,off_t);
2298 
2299 #else
2300 #define BLKDEV_MAJOR_HASH_SIZE  0
2301 #endif
2302 
2303 extern void init_special_inode(struct inode *, umode_t, dev_t);
2304 
2305 /* Invalid inode operations -- fs/bad_inode.c */
2306 extern void make_bad_inode(struct inode *);
2307 extern int is_bad_inode(struct inode *);
2308 
2309 #ifdef CONFIG_BLOCK
2310 /*
2311  * return READ, READA, or WRITE
2312  */
2313 #define bio_rw(bio)             ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2314 
2315 /*
2316  * return data direction, READ or WRITE
2317  */
2318 #define bio_data_dir(bio)       ((bio)->bi_rw & 1)
2319 
2320 extern void check_disk_size_change(struct gendisk *disk,
2321                                    struct block_device *bdev);
2322 extern int revalidate_disk(struct gendisk *);
2323 extern int check_disk_change(struct block_device *);
2324 extern int __invalidate_device(struct block_device *, bool);
2325 extern int invalidate_partition(struct gendisk *, int);
2326 #endif
2327 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2328                                         pgoff_t start, pgoff_t end);
2329 
2330 static inline void invalidate_remote_inode(struct inode *inode)
2331 {
2332         if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2333             S_ISLNK(inode->i_mode))
2334                 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2335 }
2336 extern int invalidate_inode_pages2(struct address_space *mapping);
2337 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2338                                          pgoff_t start, pgoff_t end);
2339 extern int write_inode_now(struct inode *, int);
2340 extern int filemap_fdatawrite(struct address_space *);
2341 extern int filemap_flush(struct address_space *);
2342 extern int filemap_fdatawait(struct address_space *);
2343 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2344                                    loff_t lend);
2345 extern int filemap_write_and_wait(struct address_space *mapping);
2346 extern int filemap_write_and_wait_range(struct address_space *mapping,
2347                                         loff_t lstart, loff_t lend);
2348 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2349                                 loff_t start, loff_t end, int sync_mode);
2350 extern int filemap_fdatawrite_range(struct address_space *mapping,
2351                                 loff_t start, loff_t end);
2352 
2353 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2354                            int datasync);
2355 extern int vfs_fsync(struct file *file, int datasync);
2356 static inline int generic_write_sync(struct file *file, loff_t pos, loff_t count)
2357 {
2358         if (!(file->f_flags & O_DSYNC) && !IS_SYNC(file->f_mapping->host))
2359                 return 0;
2360         return vfs_fsync_range(file, pos, pos + count - 1,
2361                                (file->f_flags & __O_SYNC) ? 0 : 1);
2362 }
2363 extern void emergency_sync(void);
2364 extern void emergency_remount(void);
2365 #ifdef CONFIG_BLOCK
2366 extern sector_t bmap(struct inode *, sector_t);
2367 #endif
2368 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2369 extern int inode_permission(struct inode *, int);
2370 extern int __inode_permission(struct inode *, int);
2371 extern int generic_permission(struct inode *, int);
2372 extern int __check_sticky(struct inode *dir, struct inode *inode);
2373 
2374 static inline bool execute_ok(struct inode *inode)
2375 {
2376         return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2377 }
2378 
2379 static inline void file_start_write(struct file *file)
2380 {
2381         if (!S_ISREG(file_inode(file)->i_mode))
2382                 return;
2383         __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2384 }
2385 
2386 static inline bool file_start_write_trylock(struct file *file)
2387 {
2388         if (!S_ISREG(file_inode(file)->i_mode))
2389                 return true;
2390         return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2391 }
2392 
2393 static inline void file_end_write(struct file *file)
2394 {
2395         if (!S_ISREG(file_inode(file)->i_mode))
2396                 return;
2397         __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2398 }
2399 
2400 /*
2401  * get_write_access() gets write permission for a file.
2402  * put_write_access() releases this write permission.
2403  * This is used for regular files.
2404  * We cannot support write (and maybe mmap read-write shared) accesses and
2405  * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2406  * can have the following values:
2407  * 0: no writers, no VM_DENYWRITE mappings
2408  * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2409  * > 0: (i_writecount) users are writing to the file.
2410  *
2411  * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2412  * except for the cases where we don't hold i_writecount yet. Then we need to
2413  * use {get,deny}_write_access() - these functions check the sign and refuse
2414  * to do the change if sign is wrong.
2415  */
2416 static inline int get_write_access(struct inode *inode)
2417 {
2418         return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2419 }
2420 static inline int deny_write_access(struct file *file)
2421 {
2422         struct inode *inode = file_inode(file);
2423         return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2424 }
2425 static inline void put_write_access(struct inode * inode)
2426 {
2427         atomic_dec(&inode->i_writecount);
2428 }
2429 static inline void allow_write_access(struct file *file)
2430 {
2431         if (file)
2432                 atomic_inc(&file_inode(file)->i_writecount);
2433 }
2434 static inline bool inode_is_open_for_write(const struct inode *inode)
2435 {
2436         return atomic_read(&inode->i_writecount) > 0;
2437 }
2438 
2439 #ifdef CONFIG_IMA
2440 static inline void i_readcount_dec(struct inode *inode)
2441 {
2442         BUG_ON(!atomic_read(&inode->i_readcount));
2443         atomic_dec(&inode->i_readcount);
2444 }
2445 static inline void i_readcount_inc(struct inode *inode)
2446 {
2447         atomic_inc(&inode->i_readcount);
2448 }
2449 #else
2450 static inline void i_readcount_dec(struct inode *inode)
2451 {
2452         return;
2453 }
2454 static inline void i_readcount_inc(struct inode *inode)
2455 {
2456         return;
2457 }
2458 #endif
2459 extern int do_pipe_flags(int *, int);
2460 
2461 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2462 extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
2463 extern ssize_t __kernel_write(struct file *, const char *, size_t, loff_t *);
2464 extern struct file * open_exec(const char *);
2465  
2466 /* fs/dcache.c -- generic fs support functions */
2467 extern int is_subdir(struct dentry *, struct dentry *);
2468 extern int path_is_under(struct path *, struct path *);
2469 
2470 #include <linux/err.h>
2471 
2472 /* needed for stackable file system support */
2473 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2474 
2475 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2476 
2477 extern int inode_init_always(struct super_block *, struct inode *);
2478 extern void inode_init_once(struct inode *);
2479 extern void address_space_init_once(struct address_space *mapping);
2480 extern struct inode * igrab(struct inode *);
2481 extern ino_t iunique(struct super_block *, ino_t);
2482 extern int inode_needs_sync(struct inode *inode);
2483 extern int generic_delete_inode(struct inode *inode);
2484 static inline int generic_drop_inode(struct inode *inode)
2485 {
2486         return !inode->i_nlink || inode_unhashed(inode);
2487 }
2488 
2489 extern struct inode *ilookup5_nowait(struct super_block *sb,
2490                 unsigned long hashval, int (*test)(struct inode *, void *),
2491                 void *data);
2492 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2493                 int (*test)(struct inode *, void *), void *data);
2494 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2495 
2496 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2497 extern struct inode * iget_locked(struct super_block *, unsigned long);
2498 extern struct inode *find_inode_nowait(struct super_block *,
2499                                        unsigned long,
2500                                        int (*match)(struct inode *,
2501                                                     unsigned long, void *),
2502                                        void *data);
2503 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2504 extern int insert_inode_locked(struct inode *);
2505 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2506 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2507 #else
2508 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2509 #endif
2510 extern void unlock_new_inode(struct inode *);
2511 extern unsigned int get_next_ino(void);
2512 
2513 extern void __iget(struct inode * inode);
2514 extern void iget_failed(struct inode *);
2515 extern void clear_inode(struct inode *);
2516 extern void __destroy_inode(struct inode *);
2517 extern struct inode *new_inode_pseudo(struct super_block *sb);
2518 extern struct inode *new_inode(struct super_block *sb);
2519 extern void free_inode_nonrcu(struct inode *inode);
2520 extern int should_remove_suid(struct dentry *);
2521 extern int file_remove_suid(struct file *);
2522 
2523 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2524 static inline void insert_inode_hash(struct inode *inode)
2525 {
2526         __insert_inode_hash(inode, inode->i_ino);
2527 }
2528 
2529 extern void __remove_inode_hash(struct inode *);
2530 static inline void remove_inode_hash(struct inode *inode)
2531 {
2532         if (!inode_unhashed(inode))
2533                 __remove_inode_hash(inode);
2534 }
2535 
2536 extern void inode_sb_list_add(struct inode *inode);
2537 
2538 #ifdef CONFIG_BLOCK
2539 extern void submit_bio(int, struct bio *);
2540 extern int bdev_read_only(struct block_device *);
2541 #endif
2542 extern int set_blocksize(struct block_device *, int);
2543 extern int sb_set_blocksize(struct super_block *, int);
2544 extern int sb_min_blocksize(struct super_block *, int);
2545 
2546 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2547 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2548 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
2549 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2550 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2551 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2552 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *, loff_t);
2553 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2554 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2555 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2556 extern ssize_t new_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2557 extern ssize_t new_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2558 
2559 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos);
2560 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos);
2561 
2562 /* fs/block_dev.c */
2563 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
2564 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2565 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2566                         int datasync);
2567 extern void block_sync_page(struct page *page);
2568 
2569 /* fs/splice.c */
2570 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2571                 struct pipe_inode_info *, size_t, unsigned int);
2572 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2573                 struct pipe_inode_info *, size_t, unsigned int);
2574 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2575                 struct file *, loff_t *, size_t, unsigned int);
2576 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2577                 struct file *out, loff_t *, size_t len, unsigned int flags);
2578 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2579                 loff_t *opos, size_t len, unsigned int flags);
2580 
2581 
2582 extern void
2583 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2584 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2585 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2586 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
2587 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2588 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2589                 int whence, loff_t maxsize, loff_t eof);
2590 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2591                 int whence, loff_t size);
2592 extern int generic_file_open(struct inode * inode, struct file * filp);
2593 extern int nonseekable_open(struct inode * inode, struct file * filp);
2594 
2595 ssize_t dax_do_io(int rw, struct kiocb *, struct inode *, struct iov_iter *,
2596                 loff_t, get_block_t, dio_iodone_t, int flags);
2597 int dax_clear_blocks(struct inode *, sector_t block, long size);
2598 int dax_zero_page_range(struct inode *, loff_t from, unsigned len, get_block_t);
2599 int dax_truncate_page(struct inode *, loff_t from, get_block_t);
2600 int dax_fault(struct vm_area_struct *, struct vm_fault *, get_block_t);
2601 #define dax_mkwrite(vma, vmf, gb)       dax_fault(vma, vmf, gb)
2602 
2603 #ifdef CONFIG_BLOCK
2604 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2605                             loff_t file_offset);
2606 
2607 enum {
2608         /* need locking between buffered and direct access */
2609         DIO_LOCKING     = 0x01,
2610 
2611         /* filesystem does not support filling holes */
2612         DIO_SKIP_HOLES  = 0x02,
2613 
2614         /* filesystem can handle aio writes beyond i_size */
2615         DIO_ASYNC_EXTEND = 0x04,
2616 };
2617 
2618 void dio_end_io(struct bio *bio, int error);
2619 
2620 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2621         struct block_device *bdev, struct iov_iter *iter, loff_t offset,
2622         get_block_t get_block, dio_iodone_t end_io,
2623         dio_submit_t submit_io, int flags);
2624 
2625 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
2626                 struct inode *inode, struct iov_iter *iter, loff_t offset,
2627                 get_block_t get_block)
2628 {
2629         return __blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iter,
2630                                     offset, get_block, NULL, NULL,
2631                                     DIO_LOCKING | DIO_SKIP_HOLES);
2632 }
2633 #endif
2634 
2635 void inode_dio_wait(struct inode *inode);
2636 void inode_dio_done(struct inode *inode);
2637 
2638 extern void inode_set_flags(struct inode *inode, unsigned int flags,
2639                             unsigned int mask);
2640 
2641 extern const struct file_operations generic_ro_fops;
2642 
2643 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2644 
2645 extern int readlink_copy(char __user *, int, const char *);
2646 extern int page_readlink(struct dentry *, char __user *, int);
2647 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2648 extern void page_put_link(struct dentry *, struct nameidata *, void *);
2649 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2650                 int nofs);
2651 extern int page_symlink(struct inode *inode, const char *symname, int len);
2652 extern const struct inode_operations page_symlink_inode_operations;
2653 extern void kfree_put_link(struct dentry *, struct nameidata *, void *);
2654 extern int generic_readlink(struct dentry *, char __user *, int);
2655 extern void generic_fillattr(struct inode *, struct kstat *);
2656 int vfs_getattr_nosec(struct path *path, struct kstat *stat);
2657 extern int vfs_getattr(struct path *, struct kstat *);
2658 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2659 void inode_add_bytes(struct inode *inode, loff_t bytes);
2660 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
2661 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2662 loff_t inode_get_bytes(struct inode *inode);
2663 void inode_set_bytes(struct inode *inode, loff_t bytes);
2664 
2665 extern int vfs_readdir(struct file *, filldir_t, void *);
2666 extern int iterate_dir(struct file *, struct dir_context *);
2667 
2668 extern int vfs_stat(const char __user *, struct kstat *);
2669 extern int vfs_lstat(const char __user *, struct kstat *);
2670 extern int vfs_fstat(unsigned int, struct kstat *);
2671 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2672 
2673 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2674                     unsigned long arg);
2675 extern int __generic_block_fiemap(struct inode *inode,
2676                                   struct fiemap_extent_info *fieinfo,
2677                                   loff_t start, loff_t len,
2678                                   get_block_t *get_block);
2679 extern int generic_block_fiemap(struct inode *inode,
2680                                 struct fiemap_extent_info *fieinfo, u64 start,
2681                                 u64 len, get_block_t *get_block);
2682 
2683 extern void get_filesystem(struct file_system_type *fs);
2684 extern void put_filesystem(struct file_system_type *fs);
2685 extern struct file_system_type *get_fs_type(const char *name);
2686 extern struct super_block *get_super(struct block_device *);
2687 extern struct super_block *get_super_thawed(struct block_device *);
2688 extern struct super_block *get_active_super(struct block_device *bdev);
2689 extern void drop_super(struct super_block *sb);
2690 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2691 extern void iterate_supers_type(struct file_system_type *,
2692                                 void (*)(struct super_block *, void *), void *);
2693 
2694 extern int dcache_dir_open(struct inode *, struct file *);
2695 extern int dcache_dir_close(struct inode *, struct file *);
2696 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2697 extern int dcache_readdir(struct file *, struct dir_context *);
2698 extern int simple_setattr(struct dentry *, struct iattr *);
2699 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2700 extern int simple_statfs(struct dentry *, struct kstatfs *);
2701 extern int simple_open(struct inode *inode, struct file *file);
2702 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2703 extern int simple_unlink(struct inode *, struct dentry *);
2704 extern int simple_rmdir(struct inode *, struct dentry *);
2705 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2706 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2707 extern int simple_empty(struct dentry *);
2708 extern int simple_readpage(struct file *file, struct page *page);
2709 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2710                         loff_t pos, unsigned len, unsigned flags,
2711                         struct page **pagep, void **fsdata);
2712 extern int simple_write_end(struct file *file, struct address_space *mapping,
2713                         loff_t pos, unsigned len, unsigned copied,
2714                         struct page *page, void *fsdata);
2715 extern int always_delete_dentry(const struct dentry *);
2716 extern struct inode *alloc_anon_inode(struct super_block *);
2717 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
2718 extern const struct dentry_operations simple_dentry_operations;
2719 
2720 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
2721 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2722 extern const struct file_operations simple_dir_operations;
2723 extern const struct inode_operations simple_dir_inode_operations;
2724 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2725 struct dentry *d_alloc_name(struct dentry *, const char *);
2726 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2727 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2728 extern void simple_release_fs(struct vfsmount **mount, int *count);
2729 
2730 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2731                         loff_t *ppos, const void *from, size_t available);
2732 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2733                 const void __user *from, size_t count);
2734 
2735 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
2736 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2737 
2738 extern int generic_check_addressable(unsigned, u64);
2739 
2740 #ifdef CONFIG_MIGRATION
2741 extern int buffer_migrate_page(struct address_space *,
2742                                 struct page *, struct page *,
2743                                 enum migrate_mode);
2744 #else
2745 #define buffer_migrate_page NULL
2746 #endif
2747 
2748 extern int inode_change_ok(const struct inode *, struct iattr *);
2749 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2750 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2751 
2752 extern int file_update_time(struct file *file);
2753 
2754 extern int generic_show_options(struct seq_file *m, struct dentry *root);
2755 extern void save_mount_options(struct super_block *sb, char *options);
2756 extern void replace_mount_options(struct super_block *sb, char *options);
2757 
2758 static inline bool io_is_direct(struct file *filp)
2759 {
2760         return (filp->f_flags & O_DIRECT) || IS_DAX(file_inode(filp));
2761 }
2762 
2763 static inline ino_t parent_ino(struct dentry *dentry)
2764 {
2765         ino_t res;
2766 
2767         /*
2768          * Don't strictly need d_lock here? If the parent ino could change
2769          * then surely we'd have a deeper race in the caller?
2770          */
2771         spin_lock(&dentry->d_lock);
2772         res = dentry->d_parent->d_inode->i_ino;
2773         spin_unlock(&dentry->d_lock);
2774         return res;
2775 }
2776 
2777 /* Transaction based IO helpers */
2778 
2779 /*
2780  * An argresp is stored in an allocated page and holds the
2781  * size of the argument or response, along with its content
2782  */
2783 struct simple_transaction_argresp {
2784         ssize_t size;
2785         char data[0];
2786 };
2787 
2788 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2789 
2790 char *simple_transaction_get(struct file *file, const char __user *buf,
2791                                 size_t size);
2792 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2793                                 size_t size, loff_t *pos);
2794 int simple_transaction_release(struct inode *inode, struct file *file);
2795 
2796 void simple_transaction_set(struct file *file, size_t n);
2797 
2798 /*
2799  * simple attribute files
2800  *
2801  * These attributes behave similar to those in sysfs:
2802  *
2803  * Writing to an attribute immediately sets a value, an open file can be
2804  * written to multiple times.
2805  *
2806  * Reading from an attribute creates a buffer from the value that might get
2807  * read with multiple read calls. When the attribute has been read
2808  * completely, no further read calls are possible until the file is opened
2809  * again.
2810  *
2811  * All attributes contain a text representation of a numeric value
2812  * that are accessed with the get() and set() functions.
2813  */
2814 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt)            \
2815 static int __fops ## _open(struct inode *inode, struct file *file)      \
2816 {                                                                       \
2817         __simple_attr_check_format(__fmt, 0ull);                        \
2818         return simple_attr_open(inode, file, __get, __set, __fmt);      \
2819 }                                                                       \
2820 static const struct file_operations __fops = {                          \
2821         .owner   = THIS_MODULE,                                         \
2822         .open    = __fops ## _open,                                     \
2823         .release = simple_attr_release,                                 \
2824         .read    = simple_attr_read,                                    \
2825         .write   = simple_attr_write,                                   \
2826         .llseek  = generic_file_llseek,                                 \
2827 }
2828 
2829 static inline __printf(1, 2)
2830 void __simple_attr_check_format(const char *fmt, ...)
2831 {
2832         /* don't do anything, just let the compiler check the arguments; */
2833 }
2834 
2835 int simple_attr_open(struct inode *inode, struct file *file,
2836                      int (*get)(void *, u64 *), int (*set)(void *, u64),
2837                      const char *fmt);
2838 int simple_attr_release(struct inode *inode, struct file *file);
2839 ssize_t simple_attr_read(struct file *file, char __user *buf,
2840                          size_t len, loff_t *ppos);
2841 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2842                           size_t len, loff_t *ppos);
2843 
2844 struct ctl_table;
2845 int proc_nr_files(struct ctl_table *table, int write,
2846                   void __user *buffer, size_t *lenp, loff_t *ppos);
2847 int proc_nr_dentry(struct ctl_table *table, int write,
2848                   void __user *buffer, size_t *lenp, loff_t *ppos);
2849 int proc_nr_inodes(struct ctl_table *table, int write,
2850                    void __user *buffer, size_t *lenp, loff_t *ppos);
2851 int __init get_filesystem_list(char *buf);
2852 
2853 #define __FMODE_EXEC            ((__force int) FMODE_EXEC)
2854 #define __FMODE_NONOTIFY        ((__force int) FMODE_NONOTIFY)
2855 
2856 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2857 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2858                                             (flag & __FMODE_NONOTIFY)))
2859 
2860 static inline int is_sxid(umode_t mode)
2861 {
2862         return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
2863 }
2864 
2865 static inline int check_sticky(struct inode *dir, struct inode *inode)
2866 {
2867         if (!(dir->i_mode & S_ISVTX))
2868                 return 0;
2869 
2870         return __check_sticky(dir, inode);
2871 }
2872 
2873 static inline void inode_has_no_xattr(struct inode *inode)
2874 {
2875         if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
2876                 inode->i_flags |= S_NOSEC;
2877 }
2878 
2879 static inline bool is_root_inode(struct inode *inode)
2880 {
2881         return inode == inode->i_sb->s_root->d_inode;
2882 }
2883 
2884 static inline bool dir_emit(struct dir_context *ctx,
2885                             const char *name, int namelen,
2886                             u64 ino, unsigned type)
2887 {
2888         return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
2889 }
2890 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
2891 {
2892         return ctx->actor(ctx, ".", 1, ctx->pos,
2893                           file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
2894 }
2895 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
2896 {
2897         return ctx->actor(ctx, "..", 2, ctx->pos,
2898                           parent_ino(file->f_path.dentry), DT_DIR) == 0;
2899 }
2900 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
2901 {
2902         if (ctx->pos == 0) {
2903                 if (!dir_emit_dot(file, ctx))
2904                         return false;
2905                 ctx->pos = 1;
2906         }
2907         if (ctx->pos == 1) {
2908                 if (!dir_emit_dotdot(file, ctx))
2909                         return false;
2910                 ctx->pos = 2;
2911         }
2912         return true;
2913 }
2914 static inline bool dir_relax(struct inode *inode)
2915 {
2916         mutex_unlock(&inode->i_mutex);
2917         mutex_lock(&inode->i_mutex);
2918         return !IS_DEADDIR(inode);
2919 }
2920 
2921 #endif /* _LINUX_FS_H */
2922 

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