Version:  2.0.40 2.2.26 2.4.37 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 4.1

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 #define IOCB_EVENTFD            (1 << 0)
318 #define IOCB_APPEND             (1 << 1)
319 #define IOCB_DIRECT             (1 << 2)
320 
321 struct kiocb {
322         struct file             *ki_filp;
323         loff_t                  ki_pos;
324         void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
325         void                    *private;
326         int                     ki_flags;
327 };
328 
329 static inline bool is_sync_kiocb(struct kiocb *kiocb)
330 {
331         return kiocb->ki_complete == NULL;
332 }
333 
334 static inline int iocb_flags(struct file *file);
335 
336 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
337 {
338         *kiocb = (struct kiocb) {
339                 .ki_filp = filp,
340                 .ki_flags = iocb_flags(filp),
341         };
342 }
343 
344 /*
345  * "descriptor" for what we're up to with a read.
346  * This allows us to use the same read code yet
347  * have multiple different users of the data that
348  * we read from a file.
349  *
350  * The simplest case just copies the data to user
351  * mode.
352  */
353 typedef struct {
354         size_t written;
355         size_t count;
356         union {
357                 char __user *buf;
358                 void *data;
359         } arg;
360         int error;
361 } read_descriptor_t;
362 
363 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
364                 unsigned long, unsigned long);
365 
366 struct address_space_operations {
367         int (*writepage)(struct page *page, struct writeback_control *wbc);
368         int (*readpage)(struct file *, struct page *);
369 
370         /* Write back some dirty pages from this mapping. */
371         int (*writepages)(struct address_space *, struct writeback_control *);
372 
373         /* Set a page dirty.  Return true if this dirtied it */
374         int (*set_page_dirty)(struct page *page);
375 
376         int (*readpages)(struct file *filp, struct address_space *mapping,
377                         struct list_head *pages, unsigned nr_pages);
378 
379         int (*write_begin)(struct file *, struct address_space *mapping,
380                                 loff_t pos, unsigned len, unsigned flags,
381                                 struct page **pagep, void **fsdata);
382         int (*write_end)(struct file *, struct address_space *mapping,
383                                 loff_t pos, unsigned len, unsigned copied,
384                                 struct page *page, void *fsdata);
385 
386         /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
387         sector_t (*bmap)(struct address_space *, sector_t);
388         void (*invalidatepage) (struct page *, unsigned int, unsigned int);
389         int (*releasepage) (struct page *, gfp_t);
390         void (*freepage)(struct page *);
391         ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter, loff_t offset);
392         /*
393          * migrate the contents of a page to the specified target. If
394          * migrate_mode is MIGRATE_ASYNC, it must not block.
395          */
396         int (*migratepage) (struct address_space *,
397                         struct page *, struct page *, enum migrate_mode);
398         int (*launder_page) (struct page *);
399         int (*is_partially_uptodate) (struct page *, unsigned long,
400                                         unsigned long);
401         void (*is_dirty_writeback) (struct page *, bool *, bool *);
402         int (*error_remove_page)(struct address_space *, struct page *);
403 
404         /* swapfile support */
405         int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
406                                 sector_t *span);
407         void (*swap_deactivate)(struct file *file);
408 };
409 
410 extern const struct address_space_operations empty_aops;
411 
412 /*
413  * pagecache_write_begin/pagecache_write_end must be used by general code
414  * to write into the pagecache.
415  */
416 int pagecache_write_begin(struct file *, struct address_space *mapping,
417                                 loff_t pos, unsigned len, unsigned flags,
418                                 struct page **pagep, void **fsdata);
419 
420 int pagecache_write_end(struct file *, struct address_space *mapping,
421                                 loff_t pos, unsigned len, unsigned copied,
422                                 struct page *page, void *fsdata);
423 
424 struct address_space {
425         struct inode            *host;          /* owner: inode, block_device */
426         struct radix_tree_root  page_tree;      /* radix tree of all pages */
427         spinlock_t              tree_lock;      /* and lock protecting it */
428         atomic_t                i_mmap_writable;/* count VM_SHARED mappings */
429         struct rb_root          i_mmap;         /* tree of private and shared mappings */
430         struct rw_semaphore     i_mmap_rwsem;   /* protect tree, count, list */
431         /* Protected by tree_lock together with the radix tree */
432         unsigned long           nrpages;        /* number of total pages */
433         unsigned long           nrshadows;      /* number of shadow entries */
434         pgoff_t                 writeback_index;/* writeback starts here */
435         const struct address_space_operations *a_ops;   /* methods */
436         unsigned long           flags;          /* error bits/gfp mask */
437         spinlock_t              private_lock;   /* for use by the address_space */
438         struct list_head        private_list;   /* ditto */
439         void                    *private_data;  /* ditto */
440 } __attribute__((aligned(sizeof(long))));
441         /*
442          * On most architectures that alignment is already the case; but
443          * must be enforced here for CRIS, to let the least significant bit
444          * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
445          */
446 struct request_queue;
447 
448 struct block_device {
449         dev_t                   bd_dev;  /* not a kdev_t - it's a search key */
450         int                     bd_openers;
451         struct inode *          bd_inode;       /* will die */
452         struct super_block *    bd_super;
453         struct mutex            bd_mutex;       /* open/close mutex */
454         struct list_head        bd_inodes;
455         void *                  bd_claiming;
456         void *                  bd_holder;
457         int                     bd_holders;
458         bool                    bd_write_holder;
459 #ifdef CONFIG_SYSFS
460         struct list_head        bd_holder_disks;
461 #endif
462         struct block_device *   bd_contains;
463         unsigned                bd_block_size;
464         struct hd_struct *      bd_part;
465         /* number of times partitions within this device have been opened. */
466         unsigned                bd_part_count;
467         int                     bd_invalidated;
468         struct gendisk *        bd_disk;
469         struct request_queue *  bd_queue;
470         struct list_head        bd_list;
471         /*
472          * Private data.  You must have bd_claim'ed the block_device
473          * to use this.  NOTE:  bd_claim allows an owner to claim
474          * the same device multiple times, the owner must take special
475          * care to not mess up bd_private for that case.
476          */
477         unsigned long           bd_private;
478 
479         /* The counter of freeze processes */
480         int                     bd_fsfreeze_count;
481         /* Mutex for freeze */
482         struct mutex            bd_fsfreeze_mutex;
483 };
484 
485 /*
486  * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
487  * radix trees
488  */
489 #define PAGECACHE_TAG_DIRTY     0
490 #define PAGECACHE_TAG_WRITEBACK 1
491 #define PAGECACHE_TAG_TOWRITE   2
492 
493 int mapping_tagged(struct address_space *mapping, int tag);
494 
495 static inline void i_mmap_lock_write(struct address_space *mapping)
496 {
497         down_write(&mapping->i_mmap_rwsem);
498 }
499 
500 static inline void i_mmap_unlock_write(struct address_space *mapping)
501 {
502         up_write(&mapping->i_mmap_rwsem);
503 }
504 
505 static inline void i_mmap_lock_read(struct address_space *mapping)
506 {
507         down_read(&mapping->i_mmap_rwsem);
508 }
509 
510 static inline void i_mmap_unlock_read(struct address_space *mapping)
511 {
512         up_read(&mapping->i_mmap_rwsem);
513 }
514 
515 /*
516  * Might pages of this file be mapped into userspace?
517  */
518 static inline int mapping_mapped(struct address_space *mapping)
519 {
520         return  !RB_EMPTY_ROOT(&mapping->i_mmap);
521 }
522 
523 /*
524  * Might pages of this file have been modified in userspace?
525  * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
526  * marks vma as VM_SHARED if it is shared, and the file was opened for
527  * writing i.e. vma may be mprotected writable even if now readonly.
528  *
529  * If i_mmap_writable is negative, no new writable mappings are allowed. You
530  * can only deny writable mappings, if none exists right now.
531  */
532 static inline int mapping_writably_mapped(struct address_space *mapping)
533 {
534         return atomic_read(&mapping->i_mmap_writable) > 0;
535 }
536 
537 static inline int mapping_map_writable(struct address_space *mapping)
538 {
539         return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
540                 0 : -EPERM;
541 }
542 
543 static inline void mapping_unmap_writable(struct address_space *mapping)
544 {
545         atomic_dec(&mapping->i_mmap_writable);
546 }
547 
548 static inline int mapping_deny_writable(struct address_space *mapping)
549 {
550         return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
551                 0 : -EBUSY;
552 }
553 
554 static inline void mapping_allow_writable(struct address_space *mapping)
555 {
556         atomic_inc(&mapping->i_mmap_writable);
557 }
558 
559 /*
560  * Use sequence counter to get consistent i_size on 32-bit processors.
561  */
562 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
563 #include <linux/seqlock.h>
564 #define __NEED_I_SIZE_ORDERED
565 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
566 #else
567 #define i_size_ordered_init(inode) do { } while (0)
568 #endif
569 
570 struct posix_acl;
571 #define ACL_NOT_CACHED ((void *)(-1))
572 
573 #define IOP_FASTPERM    0x0001
574 #define IOP_LOOKUP      0x0002
575 #define IOP_NOFOLLOW    0x0004
576 
577 /*
578  * Keep mostly read-only and often accessed (especially for
579  * the RCU path lookup and 'stat' data) fields at the beginning
580  * of the 'struct inode'
581  */
582 struct inode {
583         umode_t                 i_mode;
584         unsigned short          i_opflags;
585         kuid_t                  i_uid;
586         kgid_t                  i_gid;
587         unsigned int            i_flags;
588 
589 #ifdef CONFIG_FS_POSIX_ACL
590         struct posix_acl        *i_acl;
591         struct posix_acl        *i_default_acl;
592 #endif
593 
594         const struct inode_operations   *i_op;
595         struct super_block      *i_sb;
596         struct address_space    *i_mapping;
597 
598 #ifdef CONFIG_SECURITY
599         void                    *i_security;
600 #endif
601 
602         /* Stat data, not accessed from path walking */
603         unsigned long           i_ino;
604         /*
605          * Filesystems may only read i_nlink directly.  They shall use the
606          * following functions for modification:
607          *
608          *    (set|clear|inc|drop)_nlink
609          *    inode_(inc|dec)_link_count
610          */
611         union {
612                 const unsigned int i_nlink;
613                 unsigned int __i_nlink;
614         };
615         dev_t                   i_rdev;
616         loff_t                  i_size;
617         struct timespec         i_atime;
618         struct timespec         i_mtime;
619         struct timespec         i_ctime;
620         spinlock_t              i_lock; /* i_blocks, i_bytes, maybe i_size */
621         unsigned short          i_bytes;
622         unsigned int            i_blkbits;
623         blkcnt_t                i_blocks;
624 
625 #ifdef __NEED_I_SIZE_ORDERED
626         seqcount_t              i_size_seqcount;
627 #endif
628 
629         /* Misc */
630         unsigned long           i_state;
631         struct mutex            i_mutex;
632 
633         unsigned long           dirtied_when;   /* jiffies of first dirtying */
634         unsigned long           dirtied_time_when;
635 
636         struct hlist_node       i_hash;
637         struct list_head        i_wb_list;      /* backing dev IO list */
638         struct list_head        i_lru;          /* inode LRU list */
639         struct list_head        i_sb_list;
640         union {
641                 struct hlist_head       i_dentry;
642                 struct rcu_head         i_rcu;
643         };
644         u64                     i_version;
645         atomic_t                i_count;
646         atomic_t                i_dio_count;
647         atomic_t                i_writecount;
648 #ifdef CONFIG_IMA
649         atomic_t                i_readcount; /* struct files open RO */
650 #endif
651         const struct file_operations    *i_fop; /* former ->i_op->default_file_ops */
652         struct file_lock_context        *i_flctx;
653         struct address_space    i_data;
654         struct list_head        i_devices;
655         union {
656                 struct pipe_inode_info  *i_pipe;
657                 struct block_device     *i_bdev;
658                 struct cdev             *i_cdev;
659         };
660 
661         __u32                   i_generation;
662 
663 #ifdef CONFIG_FSNOTIFY
664         __u32                   i_fsnotify_mask; /* all events this inode cares about */
665         struct hlist_head       i_fsnotify_marks;
666 #endif
667 
668         void                    *i_private; /* fs or device private pointer */
669 };
670 
671 static inline int inode_unhashed(struct inode *inode)
672 {
673         return hlist_unhashed(&inode->i_hash);
674 }
675 
676 /*
677  * inode->i_mutex nesting subclasses for the lock validator:
678  *
679  * 0: the object of the current VFS operation
680  * 1: parent
681  * 2: child/target
682  * 3: xattr
683  * 4: second non-directory
684  * 5: second parent (when locking independent directories in rename)
685  *
686  * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
687  * non-directories at once.
688  *
689  * The locking order between these classes is
690  * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
691  */
692 enum inode_i_mutex_lock_class
693 {
694         I_MUTEX_NORMAL,
695         I_MUTEX_PARENT,
696         I_MUTEX_CHILD,
697         I_MUTEX_XATTR,
698         I_MUTEX_NONDIR2,
699         I_MUTEX_PARENT2,
700 };
701 
702 void lock_two_nondirectories(struct inode *, struct inode*);
703 void unlock_two_nondirectories(struct inode *, struct inode*);
704 
705 /*
706  * NOTE: in a 32bit arch with a preemptable kernel and
707  * an UP compile the i_size_read/write must be atomic
708  * with respect to the local cpu (unlike with preempt disabled),
709  * but they don't need to be atomic with respect to other cpus like in
710  * true SMP (so they need either to either locally disable irq around
711  * the read or for example on x86 they can be still implemented as a
712  * cmpxchg8b without the need of the lock prefix). For SMP compiles
713  * and 64bit archs it makes no difference if preempt is enabled or not.
714  */
715 static inline loff_t i_size_read(const struct inode *inode)
716 {
717 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
718         loff_t i_size;
719         unsigned int seq;
720 
721         do {
722                 seq = read_seqcount_begin(&inode->i_size_seqcount);
723                 i_size = inode->i_size;
724         } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
725         return i_size;
726 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
727         loff_t i_size;
728 
729         preempt_disable();
730         i_size = inode->i_size;
731         preempt_enable();
732         return i_size;
733 #else
734         return inode->i_size;
735 #endif
736 }
737 
738 /*
739  * NOTE: unlike i_size_read(), i_size_write() does need locking around it
740  * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
741  * can be lost, resulting in subsequent i_size_read() calls spinning forever.
742  */
743 static inline void i_size_write(struct inode *inode, loff_t i_size)
744 {
745 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
746         preempt_disable();
747         write_seqcount_begin(&inode->i_size_seqcount);
748         inode->i_size = i_size;
749         write_seqcount_end(&inode->i_size_seqcount);
750         preempt_enable();
751 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
752         preempt_disable();
753         inode->i_size = i_size;
754         preempt_enable();
755 #else
756         inode->i_size = i_size;
757 #endif
758 }
759 
760 /* Helper functions so that in most cases filesystems will
761  * not need to deal directly with kuid_t and kgid_t and can
762  * instead deal with the raw numeric values that are stored
763  * in the filesystem.
764  */
765 static inline uid_t i_uid_read(const struct inode *inode)
766 {
767         return from_kuid(&init_user_ns, inode->i_uid);
768 }
769 
770 static inline gid_t i_gid_read(const struct inode *inode)
771 {
772         return from_kgid(&init_user_ns, inode->i_gid);
773 }
774 
775 static inline void i_uid_write(struct inode *inode, uid_t uid)
776 {
777         inode->i_uid = make_kuid(&init_user_ns, uid);
778 }
779 
780 static inline void i_gid_write(struct inode *inode, gid_t gid)
781 {
782         inode->i_gid = make_kgid(&init_user_ns, gid);
783 }
784 
785 static inline unsigned iminor(const struct inode *inode)
786 {
787         return MINOR(inode->i_rdev);
788 }
789 
790 static inline unsigned imajor(const struct inode *inode)
791 {
792         return MAJOR(inode->i_rdev);
793 }
794 
795 extern struct block_device *I_BDEV(struct inode *inode);
796 
797 struct fown_struct {
798         rwlock_t lock;          /* protects pid, uid, euid fields */
799         struct pid *pid;        /* pid or -pgrp where SIGIO should be sent */
800         enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
801         kuid_t uid, euid;       /* uid/euid of process setting the owner */
802         int signum;             /* posix.1b rt signal to be delivered on IO */
803 };
804 
805 /*
806  * Track a single file's readahead state
807  */
808 struct file_ra_state {
809         pgoff_t start;                  /* where readahead started */
810         unsigned int size;              /* # of readahead pages */
811         unsigned int async_size;        /* do asynchronous readahead when
812                                            there are only # of pages ahead */
813 
814         unsigned int ra_pages;          /* Maximum readahead window */
815         unsigned int mmap_miss;         /* Cache miss stat for mmap accesses */
816         loff_t prev_pos;                /* Cache last read() position */
817 };
818 
819 /*
820  * Check if @index falls in the readahead windows.
821  */
822 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
823 {
824         return (index >= ra->start &&
825                 index <  ra->start + ra->size);
826 }
827 
828 struct file {
829         union {
830                 struct llist_node       fu_llist;
831                 struct rcu_head         fu_rcuhead;
832         } f_u;
833         struct path             f_path;
834         struct inode            *f_inode;       /* cached value */
835         const struct file_operations    *f_op;
836 
837         /*
838          * Protects f_ep_links, f_flags.
839          * Must not be taken from IRQ context.
840          */
841         spinlock_t              f_lock;
842         atomic_long_t           f_count;
843         unsigned int            f_flags;
844         fmode_t                 f_mode;
845         struct mutex            f_pos_lock;
846         loff_t                  f_pos;
847         struct fown_struct      f_owner;
848         const struct cred       *f_cred;
849         struct file_ra_state    f_ra;
850 
851         u64                     f_version;
852 #ifdef CONFIG_SECURITY
853         void                    *f_security;
854 #endif
855         /* needed for tty driver, and maybe others */
856         void                    *private_data;
857 
858 #ifdef CONFIG_EPOLL
859         /* Used by fs/eventpoll.c to link all the hooks to this file */
860         struct list_head        f_ep_links;
861         struct list_head        f_tfile_llink;
862 #endif /* #ifdef CONFIG_EPOLL */
863         struct address_space    *f_mapping;
864 } __attribute__((aligned(4)));  /* lest something weird decides that 2 is OK */
865 
866 struct file_handle {
867         __u32 handle_bytes;
868         int handle_type;
869         /* file identifier */
870         unsigned char f_handle[0];
871 };
872 
873 static inline struct file *get_file(struct file *f)
874 {
875         atomic_long_inc(&f->f_count);
876         return f;
877 }
878 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
879 #define fput_atomic(x)  atomic_long_add_unless(&(x)->f_count, -1, 1)
880 #define file_count(x)   atomic_long_read(&(x)->f_count)
881 
882 #define MAX_NON_LFS     ((1UL<<31) - 1)
883 
884 /* Page cache limit. The filesystems should put that into their s_maxbytes 
885    limits, otherwise bad things can happen in VM. */ 
886 #if BITS_PER_LONG==32
887 #define MAX_LFS_FILESIZE        (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1) 
888 #elif BITS_PER_LONG==64
889 #define MAX_LFS_FILESIZE        ((loff_t)0x7fffffffffffffffLL)
890 #endif
891 
892 #define FL_POSIX        1
893 #define FL_FLOCK        2
894 #define FL_DELEG        4       /* NFSv4 delegation */
895 #define FL_ACCESS       8       /* not trying to lock, just looking */
896 #define FL_EXISTS       16      /* when unlocking, test for existence */
897 #define FL_LEASE        32      /* lease held on this file */
898 #define FL_CLOSE        64      /* unlock on close */
899 #define FL_SLEEP        128     /* A blocking lock */
900 #define FL_DOWNGRADE_PENDING    256 /* Lease is being downgraded */
901 #define FL_UNLOCK_PENDING       512 /* Lease is being broken */
902 #define FL_OFDLCK       1024    /* lock is "owned" by struct file */
903 #define FL_LAYOUT       2048    /* outstanding pNFS layout */
904 
905 /*
906  * Special return value from posix_lock_file() and vfs_lock_file() for
907  * asynchronous locking.
908  */
909 #define FILE_LOCK_DEFERRED 1
910 
911 /* legacy typedef, should eventually be removed */
912 typedef void *fl_owner_t;
913 
914 struct file_lock;
915 
916 struct file_lock_operations {
917         void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
918         void (*fl_release_private)(struct file_lock *);
919 };
920 
921 struct lock_manager_operations {
922         int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
923         unsigned long (*lm_owner_key)(struct file_lock *);
924         fl_owner_t (*lm_get_owner)(fl_owner_t);
925         void (*lm_put_owner)(fl_owner_t);
926         void (*lm_notify)(struct file_lock *);  /* unblock callback */
927         int (*lm_grant)(struct file_lock *, int);
928         bool (*lm_break)(struct file_lock *);
929         int (*lm_change)(struct file_lock *, int, struct list_head *);
930         void (*lm_setup)(struct file_lock *, void **);
931 };
932 
933 struct lock_manager {
934         struct list_head list;
935 };
936 
937 struct net;
938 void locks_start_grace(struct net *, struct lock_manager *);
939 void locks_end_grace(struct lock_manager *);
940 int locks_in_grace(struct net *);
941 
942 /* that will die - we need it for nfs_lock_info */
943 #include <linux/nfs_fs_i.h>
944 
945 /*
946  * struct file_lock represents a generic "file lock". It's used to represent
947  * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
948  * note that the same struct is used to represent both a request for a lock and
949  * the lock itself, but the same object is never used for both.
950  *
951  * FIXME: should we create a separate "struct lock_request" to help distinguish
952  * these two uses?
953  *
954  * The varous i_flctx lists are ordered by:
955  *
956  * 1) lock owner
957  * 2) lock range start
958  * 3) lock range end
959  *
960  * Obviously, the last two criteria only matter for POSIX locks.
961  */
962 struct file_lock {
963         struct file_lock *fl_next;      /* singly linked list for this inode  */
964         struct list_head fl_list;       /* link into file_lock_context */
965         struct hlist_node fl_link;      /* node in global lists */
966         struct list_head fl_block;      /* circular list of blocked processes */
967         fl_owner_t fl_owner;
968         unsigned int fl_flags;
969         unsigned char fl_type;
970         unsigned int fl_pid;
971         int fl_link_cpu;                /* what cpu's list is this on? */
972         struct pid *fl_nspid;
973         wait_queue_head_t fl_wait;
974         struct file *fl_file;
975         loff_t fl_start;
976         loff_t fl_end;
977 
978         struct fasync_struct *  fl_fasync; /* for lease break notifications */
979         /* for lease breaks: */
980         unsigned long fl_break_time;
981         unsigned long fl_downgrade_time;
982 
983         const struct file_lock_operations *fl_ops;      /* Callbacks for filesystems */
984         const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
985         union {
986                 struct nfs_lock_info    nfs_fl;
987                 struct nfs4_lock_info   nfs4_fl;
988                 struct {
989                         struct list_head link;  /* link in AFS vnode's pending_locks list */
990                         int state;              /* state of grant or error if -ve */
991                 } afs;
992         } fl_u;
993 };
994 
995 struct file_lock_context {
996         spinlock_t              flc_lock;
997         struct list_head        flc_flock;
998         struct list_head        flc_posix;
999         struct list_head        flc_lease;
1000 };
1001 
1002 /* The following constant reflects the upper bound of the file/locking space */
1003 #ifndef OFFSET_MAX
1004 #define INT_LIMIT(x)    (~((x)1 << (sizeof(x)*8 - 1)))
1005 #define OFFSET_MAX      INT_LIMIT(loff_t)
1006 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1007 #endif
1008 
1009 #include <linux/fcntl.h>
1010 
1011 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1012 
1013 #ifdef CONFIG_FILE_LOCKING
1014 extern int fcntl_getlk(struct file *, unsigned int, struct flock __user *);
1015 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1016                         struct flock __user *);
1017 
1018 #if BITS_PER_LONG == 32
1019 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 __user *);
1020 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1021                         struct flock64 __user *);
1022 #endif
1023 
1024 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1025 extern int fcntl_getlease(struct file *filp);
1026 
1027 /* fs/locks.c */
1028 void locks_free_lock_context(struct file_lock_context *ctx);
1029 void locks_free_lock(struct file_lock *fl);
1030 extern void locks_init_lock(struct file_lock *);
1031 extern struct file_lock * locks_alloc_lock(void);
1032 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1033 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1034 extern void locks_remove_posix(struct file *, fl_owner_t);
1035 extern void locks_remove_file(struct file *);
1036 extern void locks_release_private(struct file_lock *);
1037 extern void posix_test_lock(struct file *, struct file_lock *);
1038 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1039 extern int posix_lock_file_wait(struct file *, struct file_lock *);
1040 extern int posix_unblock_lock(struct file_lock *);
1041 extern int vfs_test_lock(struct file *, struct file_lock *);
1042 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1043 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1044 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
1045 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1046 extern void lease_get_mtime(struct inode *, struct timespec *time);
1047 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1048 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1049 extern int lease_modify(struct file_lock *, int, struct list_head *);
1050 struct files_struct;
1051 extern void show_fd_locks(struct seq_file *f,
1052                          struct file *filp, struct files_struct *files);
1053 #else /* !CONFIG_FILE_LOCKING */
1054 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1055                               struct flock __user *user)
1056 {
1057         return -EINVAL;
1058 }
1059 
1060 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1061                               unsigned int cmd, struct flock __user *user)
1062 {
1063         return -EACCES;
1064 }
1065 
1066 #if BITS_PER_LONG == 32
1067 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1068                                 struct flock64 __user *user)
1069 {
1070         return -EINVAL;
1071 }
1072 
1073 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1074                                 unsigned int cmd, struct flock64 __user *user)
1075 {
1076         return -EACCES;
1077 }
1078 #endif
1079 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1080 {
1081         return -EINVAL;
1082 }
1083 
1084 static inline int fcntl_getlease(struct file *filp)
1085 {
1086         return F_UNLCK;
1087 }
1088 
1089 static inline void
1090 locks_free_lock_context(struct file_lock_context *ctx)
1091 {
1092 }
1093 
1094 static inline void locks_init_lock(struct file_lock *fl)
1095 {
1096         return;
1097 }
1098 
1099 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1100 {
1101         return;
1102 }
1103 
1104 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1105 {
1106         return;
1107 }
1108 
1109 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1110 {
1111         return;
1112 }
1113 
1114 static inline void locks_remove_file(struct file *filp)
1115 {
1116         return;
1117 }
1118 
1119 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1120 {
1121         return;
1122 }
1123 
1124 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1125                                   struct file_lock *conflock)
1126 {
1127         return -ENOLCK;
1128 }
1129 
1130 static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1131 {
1132         return -ENOLCK;
1133 }
1134 
1135 static inline int posix_unblock_lock(struct file_lock *waiter)
1136 {
1137         return -ENOENT;
1138 }
1139 
1140 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1141 {
1142         return 0;
1143 }
1144 
1145 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1146                                 struct file_lock *fl, struct file_lock *conf)
1147 {
1148         return -ENOLCK;
1149 }
1150 
1151 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1152 {
1153         return 0;
1154 }
1155 
1156 static inline int flock_lock_file_wait(struct file *filp,
1157                                        struct file_lock *request)
1158 {
1159         return -ENOLCK;
1160 }
1161 
1162 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1163 {
1164         return 0;
1165 }
1166 
1167 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1168 {
1169         return;
1170 }
1171 
1172 static inline int generic_setlease(struct file *filp, long arg,
1173                                     struct file_lock **flp, void **priv)
1174 {
1175         return -EINVAL;
1176 }
1177 
1178 static inline int vfs_setlease(struct file *filp, long arg,
1179                                struct file_lock **lease, void **priv)
1180 {
1181         return -EINVAL;
1182 }
1183 
1184 static inline int lease_modify(struct file_lock *fl, int arg,
1185                                struct list_head *dispose)
1186 {
1187         return -EINVAL;
1188 }
1189 
1190 struct files_struct;
1191 static inline void show_fd_locks(struct seq_file *f,
1192                         struct file *filp, struct files_struct *files) {}
1193 #endif /* !CONFIG_FILE_LOCKING */
1194 
1195 
1196 struct fasync_struct {
1197         spinlock_t              fa_lock;
1198         int                     magic;
1199         int                     fa_fd;
1200         struct fasync_struct    *fa_next; /* singly linked list */
1201         struct file             *fa_file;
1202         struct rcu_head         fa_rcu;
1203 };
1204 
1205 #define FASYNC_MAGIC 0x4601
1206 
1207 /* SMP safe fasync helpers: */
1208 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1209 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1210 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1211 extern struct fasync_struct *fasync_alloc(void);
1212 extern void fasync_free(struct fasync_struct *);
1213 
1214 /* can be called from interrupts */
1215 extern void kill_fasync(struct fasync_struct **, int, int);
1216 
1217 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1218 extern void f_setown(struct file *filp, unsigned long arg, int force);
1219 extern void f_delown(struct file *filp);
1220 extern pid_t f_getown(struct file *filp);
1221 extern int send_sigurg(struct fown_struct *fown);
1222 
1223 struct mm_struct;
1224 
1225 /*
1226  *      Umount options
1227  */
1228 
1229 #define MNT_FORCE       0x00000001      /* Attempt to forcibily umount */
1230 #define MNT_DETACH      0x00000002      /* Just detach from the tree */
1231 #define MNT_EXPIRE      0x00000004      /* Mark for expiry */
1232 #define UMOUNT_NOFOLLOW 0x00000008      /* Don't follow symlink on umount */
1233 #define UMOUNT_UNUSED   0x80000000      /* Flag guaranteed to be unused */
1234 
1235 
1236 /* Possible states of 'frozen' field */
1237 enum {
1238         SB_UNFROZEN = 0,                /* FS is unfrozen */
1239         SB_FREEZE_WRITE = 1,            /* Writes, dir ops, ioctls frozen */
1240         SB_FREEZE_PAGEFAULT = 2,        /* Page faults stopped as well */
1241         SB_FREEZE_FS = 3,               /* For internal FS use (e.g. to stop
1242                                          * internal threads if needed) */
1243         SB_FREEZE_COMPLETE = 4,         /* ->freeze_fs finished successfully */
1244 };
1245 
1246 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1247 
1248 struct sb_writers {
1249         /* Counters for counting writers at each level */
1250         struct percpu_counter   counter[SB_FREEZE_LEVELS];
1251         wait_queue_head_t       wait;           /* queue for waiting for
1252                                                    writers / faults to finish */
1253         int                     frozen;         /* Is sb frozen? */
1254         wait_queue_head_t       wait_unfrozen;  /* queue for waiting for
1255                                                    sb to be thawed */
1256 #ifdef CONFIG_DEBUG_LOCK_ALLOC
1257         struct lockdep_map      lock_map[SB_FREEZE_LEVELS];
1258 #endif
1259 };
1260 
1261 struct super_block {
1262         struct list_head        s_list;         /* Keep this first */
1263         dev_t                   s_dev;          /* search index; _not_ kdev_t */
1264         unsigned char           s_blocksize_bits;
1265         unsigned long           s_blocksize;
1266         loff_t                  s_maxbytes;     /* Max file size */
1267         struct file_system_type *s_type;
1268         const struct super_operations   *s_op;
1269         const struct dquot_operations   *dq_op;
1270         const struct quotactl_ops       *s_qcop;
1271         const struct export_operations *s_export_op;
1272         unsigned long           s_flags;
1273         unsigned long           s_magic;
1274         struct dentry           *s_root;
1275         struct rw_semaphore     s_umount;
1276         int                     s_count;
1277         atomic_t                s_active;
1278 #ifdef CONFIG_SECURITY
1279         void                    *s_security;
1280 #endif
1281         const struct xattr_handler **s_xattr;
1282 
1283         struct list_head        s_inodes;       /* all inodes */
1284         struct hlist_bl_head    s_anon;         /* anonymous dentries for (nfs) exporting */
1285         struct list_head        s_mounts;       /* list of mounts; _not_ for fs use */
1286         struct block_device     *s_bdev;
1287         struct backing_dev_info *s_bdi;
1288         struct mtd_info         *s_mtd;
1289         struct hlist_node       s_instances;
1290         unsigned int            s_quota_types;  /* Bitmask of supported quota types */
1291         struct quota_info       s_dquot;        /* Diskquota specific options */
1292 
1293         struct sb_writers       s_writers;
1294 
1295         char s_id[32];                          /* Informational name */
1296         u8 s_uuid[16];                          /* UUID */
1297 
1298         void                    *s_fs_info;     /* Filesystem private info */
1299         unsigned int            s_max_links;
1300         fmode_t                 s_mode;
1301 
1302         /* Granularity of c/m/atime in ns.
1303            Cannot be worse than a second */
1304         u32                s_time_gran;
1305 
1306         /*
1307          * The next field is for VFS *only*. No filesystems have any business
1308          * even looking at it. You had been warned.
1309          */
1310         struct mutex s_vfs_rename_mutex;        /* Kludge */
1311 
1312         /*
1313          * Filesystem subtype.  If non-empty the filesystem type field
1314          * in /proc/mounts will be "type.subtype"
1315          */
1316         char *s_subtype;
1317 
1318         /*
1319          * Saved mount options for lazy filesystems using
1320          * generic_show_options()
1321          */
1322         char __rcu *s_options;
1323         const struct dentry_operations *s_d_op; /* default d_op for dentries */
1324 
1325         /*
1326          * Saved pool identifier for cleancache (-1 means none)
1327          */
1328         int cleancache_poolid;
1329 
1330         struct shrinker s_shrink;       /* per-sb shrinker handle */
1331 
1332         /* Number of inodes with nlink == 0 but still referenced */
1333         atomic_long_t s_remove_count;
1334 
1335         /* Being remounted read-only */
1336         int s_readonly_remount;
1337 
1338         /* AIO completions deferred from interrupt context */
1339         struct workqueue_struct *s_dio_done_wq;
1340         struct hlist_head s_pins;
1341 
1342         /*
1343          * Keep the lru lists last in the structure so they always sit on their
1344          * own individual cachelines.
1345          */
1346         struct list_lru         s_dentry_lru ____cacheline_aligned_in_smp;
1347         struct list_lru         s_inode_lru ____cacheline_aligned_in_smp;
1348         struct rcu_head         rcu;
1349 
1350         /*
1351          * Indicates how deep in a filesystem stack this SB is
1352          */
1353         int s_stack_depth;
1354 };
1355 
1356 extern struct timespec current_fs_time(struct super_block *sb);
1357 
1358 /*
1359  * Snapshotting support.
1360  */
1361 
1362 void __sb_end_write(struct super_block *sb, int level);
1363 int __sb_start_write(struct super_block *sb, int level, bool wait);
1364 
1365 /**
1366  * sb_end_write - drop write access to a superblock
1367  * @sb: the super we wrote to
1368  *
1369  * Decrement number of writers to the filesystem. Wake up possible waiters
1370  * wanting to freeze the filesystem.
1371  */
1372 static inline void sb_end_write(struct super_block *sb)
1373 {
1374         __sb_end_write(sb, SB_FREEZE_WRITE);
1375 }
1376 
1377 /**
1378  * sb_end_pagefault - drop write access to a superblock from a page fault
1379  * @sb: the super we wrote to
1380  *
1381  * Decrement number of processes handling write page fault to the filesystem.
1382  * Wake up possible waiters wanting to freeze the filesystem.
1383  */
1384 static inline void sb_end_pagefault(struct super_block *sb)
1385 {
1386         __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1387 }
1388 
1389 /**
1390  * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1391  * @sb: the super we wrote to
1392  *
1393  * Decrement fs-internal number of writers to the filesystem.  Wake up possible
1394  * waiters wanting to freeze the filesystem.
1395  */
1396 static inline void sb_end_intwrite(struct super_block *sb)
1397 {
1398         __sb_end_write(sb, SB_FREEZE_FS);
1399 }
1400 
1401 /**
1402  * sb_start_write - get write access to a superblock
1403  * @sb: the super we write to
1404  *
1405  * When a process wants to write data or metadata to a file system (i.e. dirty
1406  * a page or an inode), it should embed the operation in a sb_start_write() -
1407  * sb_end_write() pair to get exclusion against file system freezing. This
1408  * function increments number of writers preventing freezing. If the file
1409  * system is already frozen, the function waits until the file system is
1410  * thawed.
1411  *
1412  * Since freeze protection behaves as a lock, users have to preserve
1413  * ordering of freeze protection and other filesystem locks. Generally,
1414  * freeze protection should be the outermost lock. In particular, we have:
1415  *
1416  * sb_start_write
1417  *   -> i_mutex                 (write path, truncate, directory ops, ...)
1418  *   -> s_umount                (freeze_super, thaw_super)
1419  */
1420 static inline void sb_start_write(struct super_block *sb)
1421 {
1422         __sb_start_write(sb, SB_FREEZE_WRITE, true);
1423 }
1424 
1425 static inline int sb_start_write_trylock(struct super_block *sb)
1426 {
1427         return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1428 }
1429 
1430 /**
1431  * sb_start_pagefault - get write access to a superblock from a page fault
1432  * @sb: the super we write to
1433  *
1434  * When a process starts handling write page fault, it should embed the
1435  * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1436  * exclusion against file system freezing. This is needed since the page fault
1437  * is going to dirty a page. This function increments number of running page
1438  * faults preventing freezing. If the file system is already frozen, the
1439  * function waits until the file system is thawed.
1440  *
1441  * Since page fault freeze protection behaves as a lock, users have to preserve
1442  * ordering of freeze protection and other filesystem locks. It is advised to
1443  * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1444  * handling code implies lock dependency:
1445  *
1446  * mmap_sem
1447  *   -> sb_start_pagefault
1448  */
1449 static inline void sb_start_pagefault(struct super_block *sb)
1450 {
1451         __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1452 }
1453 
1454 /*
1455  * sb_start_intwrite - get write access to a superblock for internal fs purposes
1456  * @sb: the super we write to
1457  *
1458  * This is the third level of protection against filesystem freezing. It is
1459  * free for use by a filesystem. The only requirement is that it must rank
1460  * below sb_start_pagefault.
1461  *
1462  * For example filesystem can call sb_start_intwrite() when starting a
1463  * transaction which somewhat eases handling of freezing for internal sources
1464  * of filesystem changes (internal fs threads, discarding preallocation on file
1465  * close, etc.).
1466  */
1467 static inline void sb_start_intwrite(struct super_block *sb)
1468 {
1469         __sb_start_write(sb, SB_FREEZE_FS, true);
1470 }
1471 
1472 
1473 extern bool inode_owner_or_capable(const struct inode *inode);
1474 
1475 /*
1476  * VFS helper functions..
1477  */
1478 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1479 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1480 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1481 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1482 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1483 extern int vfs_rmdir(struct inode *, struct dentry *);
1484 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1485 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1486 extern int vfs_whiteout(struct inode *, struct dentry *);
1487 
1488 /*
1489  * VFS dentry helper functions.
1490  */
1491 extern void dentry_unhash(struct dentry *dentry);
1492 
1493 /*
1494  * VFS file helper functions.
1495  */
1496 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1497                         umode_t mode);
1498 /*
1499  * VFS FS_IOC_FIEMAP helper definitions.
1500  */
1501 struct fiemap_extent_info {
1502         unsigned int fi_flags;          /* Flags as passed from user */
1503         unsigned int fi_extents_mapped; /* Number of mapped extents */
1504         unsigned int fi_extents_max;    /* Size of fiemap_extent array */
1505         struct fiemap_extent __user *fi_extents_start; /* Start of
1506                                                         fiemap_extent array */
1507 };
1508 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1509                             u64 phys, u64 len, u32 flags);
1510 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1511 
1512 /*
1513  * File types
1514  *
1515  * NOTE! These match bits 12..15 of stat.st_mode
1516  * (ie "(i_mode >> 12) & 15").
1517  */
1518 #define DT_UNKNOWN      0
1519 #define DT_FIFO         1
1520 #define DT_CHR          2
1521 #define DT_DIR          4
1522 #define DT_BLK          6
1523 #define DT_REG          8
1524 #define DT_LNK          10
1525 #define DT_SOCK         12
1526 #define DT_WHT          14
1527 
1528 /*
1529  * This is the "filldir" function type, used by readdir() to let
1530  * the kernel specify what kind of dirent layout it wants to have.
1531  * This allows the kernel to read directories into kernel space or
1532  * to have different dirent layouts depending on the binary type.
1533  */
1534 struct dir_context;
1535 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1536                          unsigned);
1537 
1538 struct dir_context {
1539         const filldir_t actor;
1540         loff_t pos;
1541 };
1542 
1543 struct block_device_operations;
1544 
1545 /* These macros are for out of kernel modules to test that
1546  * the kernel supports the unlocked_ioctl and compat_ioctl
1547  * fields in struct file_operations. */
1548 #define HAVE_COMPAT_IOCTL 1
1549 #define HAVE_UNLOCKED_IOCTL 1
1550 
1551 /*
1552  * These flags let !MMU mmap() govern direct device mapping vs immediate
1553  * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1554  *
1555  * NOMMU_MAP_COPY:      Copy can be mapped (MAP_PRIVATE)
1556  * NOMMU_MAP_DIRECT:    Can be mapped directly (MAP_SHARED)
1557  * NOMMU_MAP_READ:      Can be mapped for reading
1558  * NOMMU_MAP_WRITE:     Can be mapped for writing
1559  * NOMMU_MAP_EXEC:      Can be mapped for execution
1560  */
1561 #define NOMMU_MAP_COPY          0x00000001
1562 #define NOMMU_MAP_DIRECT        0x00000008
1563 #define NOMMU_MAP_READ          VM_MAYREAD
1564 #define NOMMU_MAP_WRITE         VM_MAYWRITE
1565 #define NOMMU_MAP_EXEC          VM_MAYEXEC
1566 
1567 #define NOMMU_VMFLAGS \
1568         (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1569 
1570 
1571 struct iov_iter;
1572 
1573 struct file_operations {
1574         struct module *owner;
1575         loff_t (*llseek) (struct file *, loff_t, int);
1576         ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1577         ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1578         ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1579         ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1580         int (*iterate) (struct file *, struct dir_context *);
1581         unsigned int (*poll) (struct file *, struct poll_table_struct *);
1582         long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1583         long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1584         int (*mmap) (struct file *, struct vm_area_struct *);
1585         int (*mremap)(struct file *, struct vm_area_struct *);
1586         int (*open) (struct inode *, struct file *);
1587         int (*flush) (struct file *, fl_owner_t id);
1588         int (*release) (struct inode *, struct file *);
1589         int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1590         int (*aio_fsync) (struct kiocb *, int datasync);
1591         int (*fasync) (int, struct file *, int);
1592         int (*lock) (struct file *, int, struct file_lock *);
1593         ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1594         unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1595         int (*check_flags)(int);
1596         int (*flock) (struct file *, int, struct file_lock *);
1597         ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1598         ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1599         int (*setlease)(struct file *, long, struct file_lock **, void **);
1600         long (*fallocate)(struct file *file, int mode, loff_t offset,
1601                           loff_t len);
1602         void (*show_fdinfo)(struct seq_file *m, struct file *f);
1603 #ifndef CONFIG_MMU
1604         unsigned (*mmap_capabilities)(struct file *);
1605 #endif
1606 };
1607 
1608 struct inode_operations {
1609         struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1610         void * (*follow_link) (struct dentry *, struct nameidata *);
1611         int (*permission) (struct inode *, int);
1612         struct posix_acl * (*get_acl)(struct inode *, int);
1613 
1614         int (*readlink) (struct dentry *, char __user *,int);
1615         void (*put_link) (struct dentry *, struct nameidata *, void *);
1616 
1617         int (*create) (struct inode *,struct dentry *, umode_t, bool);
1618         int (*link) (struct dentry *,struct inode *,struct dentry *);
1619         int (*unlink) (struct inode *,struct dentry *);
1620         int (*symlink) (struct inode *,struct dentry *,const char *);
1621         int (*mkdir) (struct inode *,struct dentry *,umode_t);
1622         int (*rmdir) (struct inode *,struct dentry *);
1623         int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1624         int (*rename) (struct inode *, struct dentry *,
1625                         struct inode *, struct dentry *);
1626         int (*rename2) (struct inode *, struct dentry *,
1627                         struct inode *, struct dentry *, unsigned int);
1628         int (*setattr) (struct dentry *, struct iattr *);
1629         int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1630         int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1631         ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1632         ssize_t (*listxattr) (struct dentry *, char *, size_t);
1633         int (*removexattr) (struct dentry *, const char *);
1634         int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1635                       u64 len);
1636         int (*update_time)(struct inode *, struct timespec *, int);
1637         int (*atomic_open)(struct inode *, struct dentry *,
1638                            struct file *, unsigned open_flag,
1639                            umode_t create_mode, int *opened);
1640         int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1641         int (*set_acl)(struct inode *, struct posix_acl *, int);
1642 
1643         /* WARNING: probably going away soon, do not use! */
1644         int (*dentry_open)(struct dentry *, struct file *, const struct cred *);
1645 } ____cacheline_aligned;
1646 
1647 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1648                               unsigned long nr_segs, unsigned long fast_segs,
1649                               struct iovec *fast_pointer,
1650                               struct iovec **ret_pointer);
1651 
1652 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1653 extern ssize_t __vfs_write(struct file *, const char __user *, size_t, loff_t *);
1654 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1655 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1656 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1657                 unsigned long, loff_t *);
1658 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1659                 unsigned long, loff_t *);
1660 
1661 struct super_operations {
1662         struct inode *(*alloc_inode)(struct super_block *sb);
1663         void (*destroy_inode)(struct inode *);
1664 
1665         void (*dirty_inode) (struct inode *, int flags);
1666         int (*write_inode) (struct inode *, struct writeback_control *wbc);
1667         int (*drop_inode) (struct inode *);
1668         void (*evict_inode) (struct inode *);
1669         void (*put_super) (struct super_block *);
1670         int (*sync_fs)(struct super_block *sb, int wait);
1671         int (*freeze_super) (struct super_block *);
1672         int (*freeze_fs) (struct super_block *);
1673         int (*thaw_super) (struct super_block *);
1674         int (*unfreeze_fs) (struct super_block *);
1675         int (*statfs) (struct dentry *, struct kstatfs *);
1676         int (*remount_fs) (struct super_block *, int *, char *);
1677         void (*umount_begin) (struct super_block *);
1678 
1679         int (*show_options)(struct seq_file *, struct dentry *);
1680         int (*show_devname)(struct seq_file *, struct dentry *);
1681         int (*show_path)(struct seq_file *, struct dentry *);
1682         int (*show_stats)(struct seq_file *, struct dentry *);
1683 #ifdef CONFIG_QUOTA
1684         ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1685         ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1686         struct dquot **(*get_dquots)(struct inode *);
1687 #endif
1688         int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1689         long (*nr_cached_objects)(struct super_block *,
1690                                   struct shrink_control *);
1691         long (*free_cached_objects)(struct super_block *,
1692                                     struct shrink_control *);
1693 };
1694 
1695 /*
1696  * Inode flags - they have no relation to superblock flags now
1697  */
1698 #define S_SYNC          1       /* Writes are synced at once */
1699 #define S_NOATIME       2       /* Do not update access times */
1700 #define S_APPEND        4       /* Append-only file */
1701 #define S_IMMUTABLE     8       /* Immutable file */
1702 #define S_DEAD          16      /* removed, but still open directory */
1703 #define S_NOQUOTA       32      /* Inode is not counted to quota */
1704 #define S_DIRSYNC       64      /* Directory modifications are synchronous */
1705 #define S_NOCMTIME      128     /* Do not update file c/mtime */
1706 #define S_SWAPFILE      256     /* Do not truncate: swapon got its bmaps */
1707 #define S_PRIVATE       512     /* Inode is fs-internal */
1708 #define S_IMA           1024    /* Inode has an associated IMA struct */
1709 #define S_AUTOMOUNT     2048    /* Automount/referral quasi-directory */
1710 #define S_NOSEC         4096    /* no suid or xattr security attributes */
1711 #ifdef CONFIG_FS_DAX
1712 #define S_DAX           8192    /* Direct Access, avoiding the page cache */
1713 #else
1714 #define S_DAX           0       /* Make all the DAX code disappear */
1715 #endif
1716 
1717 /*
1718  * Note that nosuid etc flags are inode-specific: setting some file-system
1719  * flags just means all the inodes inherit those flags by default. It might be
1720  * possible to override it selectively if you really wanted to with some
1721  * ioctl() that is not currently implemented.
1722  *
1723  * Exception: MS_RDONLY is always applied to the entire file system.
1724  *
1725  * Unfortunately, it is possible to change a filesystems flags with it mounted
1726  * with files in use.  This means that all of the inodes will not have their
1727  * i_flags updated.  Hence, i_flags no longer inherit the superblock mount
1728  * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1729  */
1730 #define __IS_FLG(inode, flg)    ((inode)->i_sb->s_flags & (flg))
1731 
1732 #define IS_RDONLY(inode)        ((inode)->i_sb->s_flags & MS_RDONLY)
1733 #define IS_SYNC(inode)          (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1734                                         ((inode)->i_flags & S_SYNC))
1735 #define IS_DIRSYNC(inode)       (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1736                                         ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1737 #define IS_MANDLOCK(inode)      __IS_FLG(inode, MS_MANDLOCK)
1738 #define IS_NOATIME(inode)       __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1739 #define IS_I_VERSION(inode)     __IS_FLG(inode, MS_I_VERSION)
1740 
1741 #define IS_NOQUOTA(inode)       ((inode)->i_flags & S_NOQUOTA)
1742 #define IS_APPEND(inode)        ((inode)->i_flags & S_APPEND)
1743 #define IS_IMMUTABLE(inode)     ((inode)->i_flags & S_IMMUTABLE)
1744 #define IS_POSIXACL(inode)      __IS_FLG(inode, MS_POSIXACL)
1745 
1746 #define IS_DEADDIR(inode)       ((inode)->i_flags & S_DEAD)
1747 #define IS_NOCMTIME(inode)      ((inode)->i_flags & S_NOCMTIME)
1748 #define IS_SWAPFILE(inode)      ((inode)->i_flags & S_SWAPFILE)
1749 #define IS_PRIVATE(inode)       ((inode)->i_flags & S_PRIVATE)
1750 #define IS_IMA(inode)           ((inode)->i_flags & S_IMA)
1751 #define IS_AUTOMOUNT(inode)     ((inode)->i_flags & S_AUTOMOUNT)
1752 #define IS_NOSEC(inode)         ((inode)->i_flags & S_NOSEC)
1753 #define IS_DAX(inode)           ((inode)->i_flags & S_DAX)
1754 
1755 #define IS_WHITEOUT(inode)      (S_ISCHR(inode->i_mode) && \
1756                                  (inode)->i_rdev == WHITEOUT_DEV)
1757 
1758 /*
1759  * Inode state bits.  Protected by inode->i_lock
1760  *
1761  * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1762  * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1763  *
1764  * Four bits define the lifetime of an inode.  Initially, inodes are I_NEW,
1765  * until that flag is cleared.  I_WILL_FREE, I_FREEING and I_CLEAR are set at
1766  * various stages of removing an inode.
1767  *
1768  * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1769  *
1770  * I_DIRTY_SYNC         Inode is dirty, but doesn't have to be written on
1771  *                      fdatasync().  i_atime is the usual cause.
1772  * I_DIRTY_DATASYNC     Data-related inode changes pending. We keep track of
1773  *                      these changes separately from I_DIRTY_SYNC so that we
1774  *                      don't have to write inode on fdatasync() when only
1775  *                      mtime has changed in it.
1776  * I_DIRTY_PAGES        Inode has dirty pages.  Inode itself may be clean.
1777  * I_NEW                Serves as both a mutex and completion notification.
1778  *                      New inodes set I_NEW.  If two processes both create
1779  *                      the same inode, one of them will release its inode and
1780  *                      wait for I_NEW to be released before returning.
1781  *                      Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1782  *                      also cause waiting on I_NEW, without I_NEW actually
1783  *                      being set.  find_inode() uses this to prevent returning
1784  *                      nearly-dead inodes.
1785  * I_WILL_FREE          Must be set when calling write_inode_now() if i_count
1786  *                      is zero.  I_FREEING must be set when I_WILL_FREE is
1787  *                      cleared.
1788  * I_FREEING            Set when inode is about to be freed but still has dirty
1789  *                      pages or buffers attached or the inode itself is still
1790  *                      dirty.
1791  * I_CLEAR              Added by clear_inode().  In this state the inode is
1792  *                      clean and can be destroyed.  Inode keeps I_FREEING.
1793  *
1794  *                      Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1795  *                      prohibited for many purposes.  iget() must wait for
1796  *                      the inode to be completely released, then create it
1797  *                      anew.  Other functions will just ignore such inodes,
1798  *                      if appropriate.  I_NEW is used for waiting.
1799  *
1800  * I_SYNC               Writeback of inode is running. The bit is set during
1801  *                      data writeback, and cleared with a wakeup on the bit
1802  *                      address once it is done. The bit is also used to pin
1803  *                      the inode in memory for flusher thread.
1804  *
1805  * I_REFERENCED         Marks the inode as recently references on the LRU list.
1806  *
1807  * I_DIO_WAKEUP         Never set.  Only used as a key for wait_on_bit().
1808  *
1809  * Q: What is the difference between I_WILL_FREE and I_FREEING?
1810  */
1811 #define I_DIRTY_SYNC            (1 << 0)
1812 #define I_DIRTY_DATASYNC        (1 << 1)
1813 #define I_DIRTY_PAGES           (1 << 2)
1814 #define __I_NEW                 3
1815 #define I_NEW                   (1 << __I_NEW)
1816 #define I_WILL_FREE             (1 << 4)
1817 #define I_FREEING               (1 << 5)
1818 #define I_CLEAR                 (1 << 6)
1819 #define __I_SYNC                7
1820 #define I_SYNC                  (1 << __I_SYNC)
1821 #define I_REFERENCED            (1 << 8)
1822 #define __I_DIO_WAKEUP          9
1823 #define I_DIO_WAKEUP            (1 << __I_DIO_WAKEUP)
1824 #define I_LINKABLE              (1 << 10)
1825 #define I_DIRTY_TIME            (1 << 11)
1826 #define __I_DIRTY_TIME_EXPIRED  12
1827 #define I_DIRTY_TIME_EXPIRED    (1 << __I_DIRTY_TIME_EXPIRED)
1828 
1829 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1830 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1831 
1832 extern void __mark_inode_dirty(struct inode *, int);
1833 static inline void mark_inode_dirty(struct inode *inode)
1834 {
1835         __mark_inode_dirty(inode, I_DIRTY);
1836 }
1837 
1838 static inline void mark_inode_dirty_sync(struct inode *inode)
1839 {
1840         __mark_inode_dirty(inode, I_DIRTY_SYNC);
1841 }
1842 
1843 extern void inc_nlink(struct inode *inode);
1844 extern void drop_nlink(struct inode *inode);
1845 extern void clear_nlink(struct inode *inode);
1846 extern void set_nlink(struct inode *inode, unsigned int nlink);
1847 
1848 static inline void inode_inc_link_count(struct inode *inode)
1849 {
1850         inc_nlink(inode);
1851         mark_inode_dirty(inode);
1852 }
1853 
1854 static inline void inode_dec_link_count(struct inode *inode)
1855 {
1856         drop_nlink(inode);
1857         mark_inode_dirty(inode);
1858 }
1859 
1860 /**
1861  * inode_inc_iversion - increments i_version
1862  * @inode: inode that need to be updated
1863  *
1864  * Every time the inode is modified, the i_version field will be incremented.
1865  * The filesystem has to be mounted with i_version flag
1866  */
1867 
1868 static inline void inode_inc_iversion(struct inode *inode)
1869 {
1870        spin_lock(&inode->i_lock);
1871        inode->i_version++;
1872        spin_unlock(&inode->i_lock);
1873 }
1874 
1875 enum file_time_flags {
1876         S_ATIME = 1,
1877         S_MTIME = 2,
1878         S_CTIME = 4,
1879         S_VERSION = 8,
1880 };
1881 
1882 extern void touch_atime(const struct path *);
1883 static inline void file_accessed(struct file *file)
1884 {
1885         if (!(file->f_flags & O_NOATIME))
1886                 touch_atime(&file->f_path);
1887 }
1888 
1889 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1890 int sync_inode_metadata(struct inode *inode, int wait);
1891 
1892 struct file_system_type {
1893         const char *name;
1894         int fs_flags;
1895 #define FS_REQUIRES_DEV         1 
1896 #define FS_BINARY_MOUNTDATA     2
1897 #define FS_HAS_SUBTYPE          4
1898 #define FS_USERNS_MOUNT         8       /* Can be mounted by userns root */
1899 #define FS_USERNS_DEV_MOUNT     16 /* A userns mount does not imply MNT_NODEV */
1900 #define FS_RENAME_DOES_D_MOVE   32768   /* FS will handle d_move() during rename() internally. */
1901         struct dentry *(*mount) (struct file_system_type *, int,
1902                        const char *, void *);
1903         void (*kill_sb) (struct super_block *);
1904         struct module *owner;
1905         struct file_system_type * next;
1906         struct hlist_head fs_supers;
1907 
1908         struct lock_class_key s_lock_key;
1909         struct lock_class_key s_umount_key;
1910         struct lock_class_key s_vfs_rename_key;
1911         struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
1912 
1913         struct lock_class_key i_lock_key;
1914         struct lock_class_key i_mutex_key;
1915         struct lock_class_key i_mutex_dir_key;
1916 };
1917 
1918 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
1919 
1920 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1921         void *data, int (*fill_super)(struct super_block *, void *, int));
1922 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1923         int flags, const char *dev_name, void *data,
1924         int (*fill_super)(struct super_block *, void *, int));
1925 extern struct dentry *mount_single(struct file_system_type *fs_type,
1926         int flags, void *data,
1927         int (*fill_super)(struct super_block *, void *, int));
1928 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1929         int flags, void *data,
1930         int (*fill_super)(struct super_block *, void *, int));
1931 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
1932 void generic_shutdown_super(struct super_block *sb);
1933 void kill_block_super(struct super_block *sb);
1934 void kill_anon_super(struct super_block *sb);
1935 void kill_litter_super(struct super_block *sb);
1936 void deactivate_super(struct super_block *sb);
1937 void deactivate_locked_super(struct super_block *sb);
1938 int set_anon_super(struct super_block *s, void *data);
1939 int get_anon_bdev(dev_t *);
1940 void free_anon_bdev(dev_t);
1941 struct super_block *sget(struct file_system_type *type,
1942                         int (*test)(struct super_block *,void *),
1943                         int (*set)(struct super_block *,void *),
1944                         int flags, void *data);
1945 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
1946         const struct super_operations *ops,
1947         const struct dentry_operations *dops,
1948         unsigned long);
1949 
1950 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1951 #define fops_get(fops) \
1952         (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1953 #define fops_put(fops) \
1954         do { if (fops) module_put((fops)->owner); } while(0)
1955 /*
1956  * This one is to be used *ONLY* from ->open() instances.
1957  * fops must be non-NULL, pinned down *and* module dependencies
1958  * should be sufficient to pin the caller down as well.
1959  */
1960 #define replace_fops(f, fops) \
1961         do {    \
1962                 struct file *__file = (f); \
1963                 fops_put(__file->f_op); \
1964                 BUG_ON(!(__file->f_op = (fops))); \
1965         } while(0)
1966 
1967 extern int register_filesystem(struct file_system_type *);
1968 extern int unregister_filesystem(struct file_system_type *);
1969 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1970 #define kern_mount(type) kern_mount_data(type, NULL)
1971 extern void kern_unmount(struct vfsmount *mnt);
1972 extern int may_umount_tree(struct vfsmount *);
1973 extern int may_umount(struct vfsmount *);
1974 extern long do_mount(const char *, const char __user *,
1975                      const char *, unsigned long, void *);
1976 extern struct vfsmount *collect_mounts(struct path *);
1977 extern void drop_collected_mounts(struct vfsmount *);
1978 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1979                           struct vfsmount *);
1980 extern int vfs_statfs(struct path *, struct kstatfs *);
1981 extern int user_statfs(const char __user *, struct kstatfs *);
1982 extern int fd_statfs(int, struct kstatfs *);
1983 extern int vfs_ustat(dev_t, struct kstatfs *);
1984 extern int freeze_super(struct super_block *super);
1985 extern int thaw_super(struct super_block *super);
1986 extern bool our_mnt(struct vfsmount *mnt);
1987 extern bool fs_fully_visible(struct file_system_type *);
1988 
1989 extern int current_umask(void);
1990 
1991 extern void ihold(struct inode * inode);
1992 extern void iput(struct inode *);
1993 extern int generic_update_time(struct inode *, struct timespec *, int);
1994 
1995 static inline struct inode *file_inode(const struct file *f)
1996 {
1997         return f->f_inode;
1998 }
1999 
2000 /* /sys/fs */
2001 extern struct kobject *fs_kobj;
2002 
2003 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
2004 
2005 #define FLOCK_VERIFY_READ  1
2006 #define FLOCK_VERIFY_WRITE 2
2007 
2008 #ifdef CONFIG_FILE_LOCKING
2009 extern int locks_mandatory_locked(struct file *);
2010 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
2011 
2012 /*
2013  * Candidates for mandatory locking have the setgid bit set
2014  * but no group execute bit -  an otherwise meaningless combination.
2015  */
2016 
2017 static inline int __mandatory_lock(struct inode *ino)
2018 {
2019         return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2020 }
2021 
2022 /*
2023  * ... and these candidates should be on MS_MANDLOCK mounted fs,
2024  * otherwise these will be advisory locks
2025  */
2026 
2027 static inline int mandatory_lock(struct inode *ino)
2028 {
2029         return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2030 }
2031 
2032 static inline int locks_verify_locked(struct file *file)
2033 {
2034         if (mandatory_lock(file_inode(file)))
2035                 return locks_mandatory_locked(file);
2036         return 0;
2037 }
2038 
2039 static inline int locks_verify_truncate(struct inode *inode,
2040                                     struct file *filp,
2041                                     loff_t size)
2042 {
2043         if (inode->i_flctx && mandatory_lock(inode))
2044                 return locks_mandatory_area(
2045                         FLOCK_VERIFY_WRITE, inode, filp,
2046                         size < inode->i_size ? size : inode->i_size,
2047                         (size < inode->i_size ? inode->i_size - size
2048                          : size - inode->i_size)
2049                 );
2050         return 0;
2051 }
2052 
2053 static inline int break_lease(struct inode *inode, unsigned int mode)
2054 {
2055         /*
2056          * Since this check is lockless, we must ensure that any refcounts
2057          * taken are done before checking i_flctx->flc_lease. Otherwise, we
2058          * could end up racing with tasks trying to set a new lease on this
2059          * file.
2060          */
2061         smp_mb();
2062         if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2063                 return __break_lease(inode, mode, FL_LEASE);
2064         return 0;
2065 }
2066 
2067 static inline int break_deleg(struct inode *inode, unsigned int mode)
2068 {
2069         /*
2070          * Since this check is lockless, we must ensure that any refcounts
2071          * taken are done before checking i_flctx->flc_lease. Otherwise, we
2072          * could end up racing with tasks trying to set a new lease on this
2073          * file.
2074          */
2075         smp_mb();
2076         if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2077                 return __break_lease(inode, mode, FL_DELEG);
2078         return 0;
2079 }
2080 
2081 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2082 {
2083         int ret;
2084 
2085         ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2086         if (ret == -EWOULDBLOCK && delegated_inode) {
2087                 *delegated_inode = inode;
2088                 ihold(inode);
2089         }
2090         return ret;
2091 }
2092 
2093 static inline int break_deleg_wait(struct inode **delegated_inode)
2094 {
2095         int ret;
2096 
2097         ret = break_deleg(*delegated_inode, O_WRONLY);
2098         iput(*delegated_inode);
2099         *delegated_inode = NULL;
2100         return ret;
2101 }
2102 
2103 static inline int break_layout(struct inode *inode, bool wait)
2104 {
2105         smp_mb();
2106         if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2107                 return __break_lease(inode,
2108                                 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2109                                 FL_LAYOUT);
2110         return 0;
2111 }
2112 
2113 #else /* !CONFIG_FILE_LOCKING */
2114 static inline int locks_mandatory_locked(struct file *file)
2115 {
2116         return 0;
2117 }
2118 
2119 static inline int locks_mandatory_area(int rw, struct inode *inode,
2120                                        struct file *filp, loff_t offset,
2121                                        size_t count)
2122 {
2123         return 0;
2124 }
2125 
2126 static inline int __mandatory_lock(struct inode *inode)
2127 {
2128         return 0;
2129 }
2130 
2131 static inline int mandatory_lock(struct inode *inode)
2132 {
2133         return 0;
2134 }
2135 
2136 static inline int locks_verify_locked(struct file *file)
2137 {
2138         return 0;
2139 }
2140 
2141 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2142                                         size_t size)
2143 {
2144         return 0;
2145 }
2146 
2147 static inline int break_lease(struct inode *inode, unsigned int mode)
2148 {
2149         return 0;
2150 }
2151 
2152 static inline int break_deleg(struct inode *inode, unsigned int mode)
2153 {
2154         return 0;
2155 }
2156 
2157 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2158 {
2159         return 0;
2160 }
2161 
2162 static inline int break_deleg_wait(struct inode **delegated_inode)
2163 {
2164         BUG();
2165         return 0;
2166 }
2167 
2168 static inline int break_layout(struct inode *inode, bool wait)
2169 {
2170         return 0;
2171 }
2172 
2173 #endif /* CONFIG_FILE_LOCKING */
2174 
2175 /* fs/open.c */
2176 struct audit_names;
2177 struct filename {
2178         const char              *name;  /* pointer to actual string */
2179         const __user char       *uptr;  /* original userland pointer */
2180         struct audit_names      *aname;
2181         int                     refcnt;
2182         const char              iname[];
2183 };
2184 
2185 extern long vfs_truncate(struct path *, loff_t);
2186 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2187                        struct file *filp);
2188 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2189                         loff_t len);
2190 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2191                         umode_t mode);
2192 extern struct file *file_open_name(struct filename *, int, umode_t);
2193 extern struct file *filp_open(const char *, int, umode_t);
2194 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2195                                    const char *, int);
2196 extern int vfs_open(const struct path *, struct file *, const struct cred *);
2197 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2198 extern int filp_close(struct file *, fl_owner_t id);
2199 
2200 extern struct filename *getname_flags(const char __user *, int, int *);
2201 extern struct filename *getname(const char __user *);
2202 extern struct filename *getname_kernel(const char *);
2203 extern void putname(struct filename *name);
2204 
2205 enum {
2206         FILE_CREATED = 1,
2207         FILE_OPENED = 2
2208 };
2209 extern int finish_open(struct file *file, struct dentry *dentry,
2210                         int (*open)(struct inode *, struct file *),
2211                         int *opened);
2212 extern int finish_no_open(struct file *file, struct dentry *dentry);
2213 
2214 /* fs/ioctl.c */
2215 
2216 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2217 
2218 /* fs/dcache.c */
2219 extern void __init vfs_caches_init_early(void);
2220 extern void __init vfs_caches_init(unsigned long);
2221 
2222 extern struct kmem_cache *names_cachep;
2223 
2224 #define __getname()             kmem_cache_alloc(names_cachep, GFP_KERNEL)
2225 #define __putname(name)         kmem_cache_free(names_cachep, (void *)(name))
2226 
2227 #ifdef CONFIG_BLOCK
2228 extern int register_blkdev(unsigned int, const char *);
2229 extern void unregister_blkdev(unsigned int, const char *);
2230 extern struct block_device *bdget(dev_t);
2231 extern struct block_device *bdgrab(struct block_device *bdev);
2232 extern void bd_set_size(struct block_device *, loff_t size);
2233 extern void bd_forget(struct inode *inode);
2234 extern void bdput(struct block_device *);
2235 extern void invalidate_bdev(struct block_device *);
2236 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2237 extern int sync_blockdev(struct block_device *bdev);
2238 extern void kill_bdev(struct block_device *);
2239 extern struct super_block *freeze_bdev(struct block_device *);
2240 extern void emergency_thaw_all(void);
2241 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2242 extern int fsync_bdev(struct block_device *);
2243 extern int sb_is_blkdev_sb(struct super_block *sb);
2244 #else
2245 static inline void bd_forget(struct inode *inode) {}
2246 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2247 static inline void kill_bdev(struct block_device *bdev) {}
2248 static inline void invalidate_bdev(struct block_device *bdev) {}
2249 
2250 static inline struct super_block *freeze_bdev(struct block_device *sb)
2251 {
2252         return NULL;
2253 }
2254 
2255 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2256 {
2257         return 0;
2258 }
2259 
2260 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2261 {
2262 }
2263 
2264 static inline int sb_is_blkdev_sb(struct super_block *sb)
2265 {
2266         return 0;
2267 }
2268 #endif
2269 extern int sync_filesystem(struct super_block *);
2270 extern const struct file_operations def_blk_fops;
2271 extern const struct file_operations def_chr_fops;
2272 #ifdef CONFIG_BLOCK
2273 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2274 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2275 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2276 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2277 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2278                                                void *holder);
2279 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2280                                               void *holder);
2281 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2282 #ifdef CONFIG_SYSFS
2283 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2284 extern void bd_unlink_disk_holder(struct block_device *bdev,
2285                                   struct gendisk *disk);
2286 #else
2287 static inline int bd_link_disk_holder(struct block_device *bdev,
2288                                       struct gendisk *disk)
2289 {
2290         return 0;
2291 }
2292 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2293                                          struct gendisk *disk)
2294 {
2295 }
2296 #endif
2297 #endif
2298 
2299 /* fs/char_dev.c */
2300 #define CHRDEV_MAJOR_HASH_SIZE  255
2301 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2302 extern int register_chrdev_region(dev_t, unsigned, const char *);
2303 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2304                              unsigned int count, const char *name,
2305                              const struct file_operations *fops);
2306 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2307                                 unsigned int count, const char *name);
2308 extern void unregister_chrdev_region(dev_t, unsigned);
2309 extern void chrdev_show(struct seq_file *,off_t);
2310 
2311 static inline int register_chrdev(unsigned int major, const char *name,
2312                                   const struct file_operations *fops)
2313 {
2314         return __register_chrdev(major, 0, 256, name, fops);
2315 }
2316 
2317 static inline void unregister_chrdev(unsigned int major, const char *name)
2318 {
2319         __unregister_chrdev(major, 0, 256, name);
2320 }
2321 
2322 /* fs/block_dev.c */
2323 #define BDEVNAME_SIZE   32      /* Largest string for a blockdev identifier */
2324 #define BDEVT_SIZE      10      /* Largest string for MAJ:MIN for blkdev */
2325 
2326 #ifdef CONFIG_BLOCK
2327 #define BLKDEV_MAJOR_HASH_SIZE  255
2328 extern const char *__bdevname(dev_t, char *buffer);
2329 extern const char *bdevname(struct block_device *bdev, char *buffer);
2330 extern struct block_device *lookup_bdev(const char *);
2331 extern void blkdev_show(struct seq_file *,off_t);
2332 
2333 #else
2334 #define BLKDEV_MAJOR_HASH_SIZE  0
2335 #endif
2336 
2337 extern void init_special_inode(struct inode *, umode_t, dev_t);
2338 
2339 /* Invalid inode operations -- fs/bad_inode.c */
2340 extern void make_bad_inode(struct inode *);
2341 extern int is_bad_inode(struct inode *);
2342 
2343 #ifdef CONFIG_BLOCK
2344 /*
2345  * return READ, READA, or WRITE
2346  */
2347 #define bio_rw(bio)             ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2348 
2349 /*
2350  * return data direction, READ or WRITE
2351  */
2352 #define bio_data_dir(bio)       ((bio)->bi_rw & 1)
2353 
2354 extern void check_disk_size_change(struct gendisk *disk,
2355                                    struct block_device *bdev);
2356 extern int revalidate_disk(struct gendisk *);
2357 extern int check_disk_change(struct block_device *);
2358 extern int __invalidate_device(struct block_device *, bool);
2359 extern int invalidate_partition(struct gendisk *, int);
2360 #endif
2361 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2362                                         pgoff_t start, pgoff_t end);
2363 
2364 static inline void invalidate_remote_inode(struct inode *inode)
2365 {
2366         if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2367             S_ISLNK(inode->i_mode))
2368                 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2369 }
2370 extern int invalidate_inode_pages2(struct address_space *mapping);
2371 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2372                                          pgoff_t start, pgoff_t end);
2373 extern int write_inode_now(struct inode *, int);
2374 extern int filemap_fdatawrite(struct address_space *);
2375 extern int filemap_flush(struct address_space *);
2376 extern int filemap_fdatawait(struct address_space *);
2377 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2378                                    loff_t lend);
2379 extern int filemap_write_and_wait(struct address_space *mapping);
2380 extern int filemap_write_and_wait_range(struct address_space *mapping,
2381                                         loff_t lstart, loff_t lend);
2382 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2383                                 loff_t start, loff_t end, int sync_mode);
2384 extern int filemap_fdatawrite_range(struct address_space *mapping,
2385                                 loff_t start, loff_t end);
2386 
2387 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2388                            int datasync);
2389 extern int vfs_fsync(struct file *file, int datasync);
2390 static inline int generic_write_sync(struct file *file, loff_t pos, loff_t count)
2391 {
2392         if (!(file->f_flags & O_DSYNC) && !IS_SYNC(file->f_mapping->host))
2393                 return 0;
2394         return vfs_fsync_range(file, pos, pos + count - 1,
2395                                (file->f_flags & __O_SYNC) ? 0 : 1);
2396 }
2397 extern void emergency_sync(void);
2398 extern void emergency_remount(void);
2399 #ifdef CONFIG_BLOCK
2400 extern sector_t bmap(struct inode *, sector_t);
2401 #endif
2402 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2403 extern int inode_permission(struct inode *, int);
2404 extern int __inode_permission(struct inode *, int);
2405 extern int generic_permission(struct inode *, int);
2406 extern int __check_sticky(struct inode *dir, struct inode *inode);
2407 
2408 static inline bool execute_ok(struct inode *inode)
2409 {
2410         return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2411 }
2412 
2413 static inline void file_start_write(struct file *file)
2414 {
2415         if (!S_ISREG(file_inode(file)->i_mode))
2416                 return;
2417         __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2418 }
2419 
2420 static inline bool file_start_write_trylock(struct file *file)
2421 {
2422         if (!S_ISREG(file_inode(file)->i_mode))
2423                 return true;
2424         return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2425 }
2426 
2427 static inline void file_end_write(struct file *file)
2428 {
2429         if (!S_ISREG(file_inode(file)->i_mode))
2430                 return;
2431         __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2432 }
2433 
2434 /*
2435  * get_write_access() gets write permission for a file.
2436  * put_write_access() releases this write permission.
2437  * This is used for regular files.
2438  * We cannot support write (and maybe mmap read-write shared) accesses and
2439  * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2440  * can have the following values:
2441  * 0: no writers, no VM_DENYWRITE mappings
2442  * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2443  * > 0: (i_writecount) users are writing to the file.
2444  *
2445  * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2446  * except for the cases where we don't hold i_writecount yet. Then we need to
2447  * use {get,deny}_write_access() - these functions check the sign and refuse
2448  * to do the change if sign is wrong.
2449  */
2450 static inline int get_write_access(struct inode *inode)
2451 {
2452         return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2453 }
2454 static inline int deny_write_access(struct file *file)
2455 {
2456         struct inode *inode = file_inode(file);
2457         return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2458 }
2459 static inline void put_write_access(struct inode * inode)
2460 {
2461         atomic_dec(&inode->i_writecount);
2462 }
2463 static inline void allow_write_access(struct file *file)
2464 {
2465         if (file)
2466                 atomic_inc(&file_inode(file)->i_writecount);
2467 }
2468 static inline bool inode_is_open_for_write(const struct inode *inode)
2469 {
2470         return atomic_read(&inode->i_writecount) > 0;
2471 }
2472 
2473 #ifdef CONFIG_IMA
2474 static inline void i_readcount_dec(struct inode *inode)
2475 {
2476         BUG_ON(!atomic_read(&inode->i_readcount));
2477         atomic_dec(&inode->i_readcount);
2478 }
2479 static inline void i_readcount_inc(struct inode *inode)
2480 {
2481         atomic_inc(&inode->i_readcount);
2482 }
2483 #else
2484 static inline void i_readcount_dec(struct inode *inode)
2485 {
2486         return;
2487 }
2488 static inline void i_readcount_inc(struct inode *inode)
2489 {
2490         return;
2491 }
2492 #endif
2493 extern int do_pipe_flags(int *, int);
2494 
2495 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2496 extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
2497 extern ssize_t __kernel_write(struct file *, const char *, size_t, loff_t *);
2498 extern struct file * open_exec(const char *);
2499  
2500 /* fs/dcache.c -- generic fs support functions */
2501 extern int is_subdir(struct dentry *, struct dentry *);
2502 extern int path_is_under(struct path *, struct path *);
2503 
2504 #include <linux/err.h>
2505 
2506 /* needed for stackable file system support */
2507 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2508 
2509 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2510 
2511 extern int inode_init_always(struct super_block *, struct inode *);
2512 extern void inode_init_once(struct inode *);
2513 extern void address_space_init_once(struct address_space *mapping);
2514 extern struct inode * igrab(struct inode *);
2515 extern ino_t iunique(struct super_block *, ino_t);
2516 extern int inode_needs_sync(struct inode *inode);
2517 extern int generic_delete_inode(struct inode *inode);
2518 static inline int generic_drop_inode(struct inode *inode)
2519 {
2520         return !inode->i_nlink || inode_unhashed(inode);
2521 }
2522 
2523 extern struct inode *ilookup5_nowait(struct super_block *sb,
2524                 unsigned long hashval, int (*test)(struct inode *, void *),
2525                 void *data);
2526 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2527                 int (*test)(struct inode *, void *), void *data);
2528 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2529 
2530 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2531 extern struct inode * iget_locked(struct super_block *, unsigned long);
2532 extern struct inode *find_inode_nowait(struct super_block *,
2533                                        unsigned long,
2534                                        int (*match)(struct inode *,
2535                                                     unsigned long, void *),
2536                                        void *data);
2537 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2538 extern int insert_inode_locked(struct inode *);
2539 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2540 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2541 #else
2542 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2543 #endif
2544 extern void unlock_new_inode(struct inode *);
2545 extern unsigned int get_next_ino(void);
2546 
2547 extern void __iget(struct inode * inode);
2548 extern void iget_failed(struct inode *);
2549 extern void clear_inode(struct inode *);
2550 extern void __destroy_inode(struct inode *);
2551 extern struct inode *new_inode_pseudo(struct super_block *sb);
2552 extern struct inode *new_inode(struct super_block *sb);
2553 extern void free_inode_nonrcu(struct inode *inode);
2554 extern int should_remove_suid(struct dentry *);
2555 extern int file_remove_suid(struct file *);
2556 
2557 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2558 static inline void insert_inode_hash(struct inode *inode)
2559 {
2560         __insert_inode_hash(inode, inode->i_ino);
2561 }
2562 
2563 extern void __remove_inode_hash(struct inode *);
2564 static inline void remove_inode_hash(struct inode *inode)
2565 {
2566         if (!inode_unhashed(inode))
2567                 __remove_inode_hash(inode);
2568 }
2569 
2570 extern void inode_sb_list_add(struct inode *inode);
2571 
2572 #ifdef CONFIG_BLOCK
2573 extern void submit_bio(int, struct bio *);
2574 extern int bdev_read_only(struct block_device *);
2575 #endif
2576 extern int set_blocksize(struct block_device *, int);
2577 extern int sb_set_blocksize(struct super_block *, int);
2578 extern int sb_min_blocksize(struct super_block *, int);
2579 
2580 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2581 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2582 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2583 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2584 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2585 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2586 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *, loff_t);
2587 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2588 
2589 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos);
2590 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos);
2591 
2592 /* fs/block_dev.c */
2593 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
2594 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2595 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2596                         int datasync);
2597 extern void block_sync_page(struct page *page);
2598 
2599 /* fs/splice.c */
2600 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2601                 struct pipe_inode_info *, size_t, unsigned int);
2602 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2603                 struct pipe_inode_info *, size_t, unsigned int);
2604 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2605                 struct file *, loff_t *, size_t, unsigned int);
2606 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2607                 struct file *out, loff_t *, size_t len, unsigned int flags);
2608 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2609                 loff_t *opos, size_t len, unsigned int flags);
2610 
2611 
2612 extern void
2613 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2614 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2615 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2616 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
2617 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2618 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2619                 int whence, loff_t maxsize, loff_t eof);
2620 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2621                 int whence, loff_t size);
2622 extern int generic_file_open(struct inode * inode, struct file * filp);
2623 extern int nonseekable_open(struct inode * inode, struct file * filp);
2624 
2625 ssize_t dax_do_io(struct kiocb *, struct inode *, struct iov_iter *, loff_t,
2626                   get_block_t, dio_iodone_t, int flags);
2627 int dax_clear_blocks(struct inode *, sector_t block, long size);
2628 int dax_zero_page_range(struct inode *, loff_t from, unsigned len, get_block_t);
2629 int dax_truncate_page(struct inode *, loff_t from, get_block_t);
2630 int dax_fault(struct vm_area_struct *, struct vm_fault *, get_block_t);
2631 int dax_pfn_mkwrite(struct vm_area_struct *, struct vm_fault *);
2632 #define dax_mkwrite(vma, vmf, gb)       dax_fault(vma, vmf, gb)
2633 
2634 #ifdef CONFIG_BLOCK
2635 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2636                             loff_t file_offset);
2637 
2638 enum {
2639         /* need locking between buffered and direct access */
2640         DIO_LOCKING     = 0x01,
2641 
2642         /* filesystem does not support filling holes */
2643         DIO_SKIP_HOLES  = 0x02,
2644 
2645         /* filesystem can handle aio writes beyond i_size */
2646         DIO_ASYNC_EXTEND = 0x04,
2647 
2648         /* inode/fs/bdev does not need truncate protection */
2649         DIO_SKIP_DIO_COUNT = 0x08,
2650 };
2651 
2652 void dio_end_io(struct bio *bio, int error);
2653 
2654 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
2655                              struct block_device *bdev, struct iov_iter *iter,
2656                              loff_t offset, get_block_t get_block,
2657                              dio_iodone_t end_io, dio_submit_t submit_io,
2658                              int flags);
2659 
2660 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
2661                                          struct inode *inode,
2662                                          struct iov_iter *iter, loff_t offset,
2663                                          get_block_t get_block)
2664 {
2665         return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
2666                                     offset, get_block, NULL, NULL,
2667                                     DIO_LOCKING | DIO_SKIP_HOLES);
2668 }
2669 #endif
2670 
2671 void inode_dio_wait(struct inode *inode);
2672 
2673 /*
2674  * inode_dio_begin - signal start of a direct I/O requests
2675  * @inode: inode the direct I/O happens on
2676  *
2677  * This is called once we've finished processing a direct I/O request,
2678  * and is used to wake up callers waiting for direct I/O to be quiesced.
2679  */
2680 static inline void inode_dio_begin(struct inode *inode)
2681 {
2682         atomic_inc(&inode->i_dio_count);
2683 }
2684 
2685 /*
2686  * inode_dio_end - signal finish of a direct I/O requests
2687  * @inode: inode the direct I/O happens on
2688  *
2689  * This is called once we've finished processing a direct I/O request,
2690  * and is used to wake up callers waiting for direct I/O to be quiesced.
2691  */
2692 static inline void inode_dio_end(struct inode *inode)
2693 {
2694         if (atomic_dec_and_test(&inode->i_dio_count))
2695                 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
2696 }
2697 
2698 extern void inode_set_flags(struct inode *inode, unsigned int flags,
2699                             unsigned int mask);
2700 
2701 extern const struct file_operations generic_ro_fops;
2702 
2703 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2704 
2705 extern int readlink_copy(char __user *, int, const char *);
2706 extern int page_readlink(struct dentry *, char __user *, int);
2707 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2708 extern void page_put_link(struct dentry *, struct nameidata *, void *);
2709 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2710                 int nofs);
2711 extern int page_symlink(struct inode *inode, const char *symname, int len);
2712 extern const struct inode_operations page_symlink_inode_operations;
2713 extern void kfree_put_link(struct dentry *, struct nameidata *, void *);
2714 extern int generic_readlink(struct dentry *, char __user *, int);
2715 extern void generic_fillattr(struct inode *, struct kstat *);
2716 int vfs_getattr_nosec(struct path *path, struct kstat *stat);
2717 extern int vfs_getattr(struct path *, struct kstat *);
2718 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2719 void inode_add_bytes(struct inode *inode, loff_t bytes);
2720 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
2721 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2722 loff_t inode_get_bytes(struct inode *inode);
2723 void inode_set_bytes(struct inode *inode, loff_t bytes);
2724 
2725 extern int iterate_dir(struct file *, struct dir_context *);
2726 
2727 extern int vfs_stat(const char __user *, struct kstat *);
2728 extern int vfs_lstat(const char __user *, struct kstat *);
2729 extern int vfs_fstat(unsigned int, struct kstat *);
2730 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2731 
2732 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2733                     unsigned long arg);
2734 extern int __generic_block_fiemap(struct inode *inode,
2735                                   struct fiemap_extent_info *fieinfo,
2736                                   loff_t start, loff_t len,
2737                                   get_block_t *get_block);
2738 extern int generic_block_fiemap(struct inode *inode,
2739                                 struct fiemap_extent_info *fieinfo, u64 start,
2740                                 u64 len, get_block_t *get_block);
2741 
2742 extern void get_filesystem(struct file_system_type *fs);
2743 extern void put_filesystem(struct file_system_type *fs);
2744 extern struct file_system_type *get_fs_type(const char *name);
2745 extern struct super_block *get_super(struct block_device *);
2746 extern struct super_block *get_super_thawed(struct block_device *);
2747 extern struct super_block *get_active_super(struct block_device *bdev);
2748 extern void drop_super(struct super_block *sb);
2749 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2750 extern void iterate_supers_type(struct file_system_type *,
2751                                 void (*)(struct super_block *, void *), void *);
2752 
2753 extern int dcache_dir_open(struct inode *, struct file *);
2754 extern int dcache_dir_close(struct inode *, struct file *);
2755 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2756 extern int dcache_readdir(struct file *, struct dir_context *);
2757 extern int simple_setattr(struct dentry *, struct iattr *);
2758 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2759 extern int simple_statfs(struct dentry *, struct kstatfs *);
2760 extern int simple_open(struct inode *inode, struct file *file);
2761 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2762 extern int simple_unlink(struct inode *, struct dentry *);
2763 extern int simple_rmdir(struct inode *, struct dentry *);
2764 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2765 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2766 extern int simple_empty(struct dentry *);
2767 extern int simple_readpage(struct file *file, struct page *page);
2768 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2769                         loff_t pos, unsigned len, unsigned flags,
2770                         struct page **pagep, void **fsdata);
2771 extern int simple_write_end(struct file *file, struct address_space *mapping,
2772                         loff_t pos, unsigned len, unsigned copied,
2773                         struct page *page, void *fsdata);
2774 extern int always_delete_dentry(const struct dentry *);
2775 extern struct inode *alloc_anon_inode(struct super_block *);
2776 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
2777 extern const struct dentry_operations simple_dentry_operations;
2778 
2779 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
2780 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2781 extern const struct file_operations simple_dir_operations;
2782 extern const struct inode_operations simple_dir_inode_operations;
2783 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2784 struct dentry *d_alloc_name(struct dentry *, const char *);
2785 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2786 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2787 extern void simple_release_fs(struct vfsmount **mount, int *count);
2788 
2789 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2790                         loff_t *ppos, const void *from, size_t available);
2791 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2792                 const void __user *from, size_t count);
2793 
2794 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
2795 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2796 
2797 extern int generic_check_addressable(unsigned, u64);
2798 
2799 #ifdef CONFIG_MIGRATION
2800 extern int buffer_migrate_page(struct address_space *,
2801                                 struct page *, struct page *,
2802                                 enum migrate_mode);
2803 #else
2804 #define buffer_migrate_page NULL
2805 #endif
2806 
2807 extern int inode_change_ok(const struct inode *, struct iattr *);
2808 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2809 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2810 
2811 extern int file_update_time(struct file *file);
2812 
2813 extern int generic_show_options(struct seq_file *m, struct dentry *root);
2814 extern void save_mount_options(struct super_block *sb, char *options);
2815 extern void replace_mount_options(struct super_block *sb, char *options);
2816 
2817 static inline bool io_is_direct(struct file *filp)
2818 {
2819         return (filp->f_flags & O_DIRECT) || IS_DAX(file_inode(filp));
2820 }
2821 
2822 static inline int iocb_flags(struct file *file)
2823 {
2824         int res = 0;
2825         if (file->f_flags & O_APPEND)
2826                 res |= IOCB_APPEND;
2827         if (io_is_direct(file))
2828                 res |= IOCB_DIRECT;
2829         return res;
2830 }
2831 
2832 static inline ino_t parent_ino(struct dentry *dentry)
2833 {
2834         ino_t res;
2835 
2836         /*
2837          * Don't strictly need d_lock here? If the parent ino could change
2838          * then surely we'd have a deeper race in the caller?
2839          */
2840         spin_lock(&dentry->d_lock);
2841         res = dentry->d_parent->d_inode->i_ino;
2842         spin_unlock(&dentry->d_lock);
2843         return res;
2844 }
2845 
2846 /* Transaction based IO helpers */
2847 
2848 /*
2849  * An argresp is stored in an allocated page and holds the
2850  * size of the argument or response, along with its content
2851  */
2852 struct simple_transaction_argresp {
2853         ssize_t size;
2854         char data[0];
2855 };
2856 
2857 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2858 
2859 char *simple_transaction_get(struct file *file, const char __user *buf,
2860                                 size_t size);
2861 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2862                                 size_t size, loff_t *pos);
2863 int simple_transaction_release(struct inode *inode, struct file *file);
2864 
2865 void simple_transaction_set(struct file *file, size_t n);
2866 
2867 /*
2868  * simple attribute files
2869  *
2870  * These attributes behave similar to those in sysfs:
2871  *
2872  * Writing to an attribute immediately sets a value, an open file can be
2873  * written to multiple times.
2874  *
2875  * Reading from an attribute creates a buffer from the value that might get
2876  * read with multiple read calls. When the attribute has been read
2877  * completely, no further read calls are possible until the file is opened
2878  * again.
2879  *
2880  * All attributes contain a text representation of a numeric value
2881  * that are accessed with the get() and set() functions.
2882  */
2883 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt)            \
2884 static int __fops ## _open(struct inode *inode, struct file *file)      \
2885 {                                                                       \
2886         __simple_attr_check_format(__fmt, 0ull);                        \
2887         return simple_attr_open(inode, file, __get, __set, __fmt);      \
2888 }                                                                       \
2889 static const struct file_operations __fops = {                          \
2890         .owner   = THIS_MODULE,                                         \
2891         .open    = __fops ## _open,                                     \
2892         .release = simple_attr_release,                                 \
2893         .read    = simple_attr_read,                                    \
2894         .write   = simple_attr_write,                                   \
2895         .llseek  = generic_file_llseek,                                 \
2896 }
2897 
2898 static inline __printf(1, 2)
2899 void __simple_attr_check_format(const char *fmt, ...)
2900 {
2901         /* don't do anything, just let the compiler check the arguments; */
2902 }
2903 
2904 int simple_attr_open(struct inode *inode, struct file *file,
2905                      int (*get)(void *, u64 *), int (*set)(void *, u64),
2906                      const char *fmt);
2907 int simple_attr_release(struct inode *inode, struct file *file);
2908 ssize_t simple_attr_read(struct file *file, char __user *buf,
2909                          size_t len, loff_t *ppos);
2910 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2911                           size_t len, loff_t *ppos);
2912 
2913 struct ctl_table;
2914 int proc_nr_files(struct ctl_table *table, int write,
2915                   void __user *buffer, size_t *lenp, loff_t *ppos);
2916 int proc_nr_dentry(struct ctl_table *table, int write,
2917                   void __user *buffer, size_t *lenp, loff_t *ppos);
2918 int proc_nr_inodes(struct ctl_table *table, int write,
2919                    void __user *buffer, size_t *lenp, loff_t *ppos);
2920 int __init get_filesystem_list(char *buf);
2921 
2922 #define __FMODE_EXEC            ((__force int) FMODE_EXEC)
2923 #define __FMODE_NONOTIFY        ((__force int) FMODE_NONOTIFY)
2924 
2925 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2926 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2927                                             (flag & __FMODE_NONOTIFY)))
2928 
2929 static inline int is_sxid(umode_t mode)
2930 {
2931         return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
2932 }
2933 
2934 static inline int check_sticky(struct inode *dir, struct inode *inode)
2935 {
2936         if (!(dir->i_mode & S_ISVTX))
2937                 return 0;
2938 
2939         return __check_sticky(dir, inode);
2940 }
2941 
2942 static inline void inode_has_no_xattr(struct inode *inode)
2943 {
2944         if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
2945                 inode->i_flags |= S_NOSEC;
2946 }
2947 
2948 static inline bool is_root_inode(struct inode *inode)
2949 {
2950         return inode == inode->i_sb->s_root->d_inode;
2951 }
2952 
2953 static inline bool dir_emit(struct dir_context *ctx,
2954                             const char *name, int namelen,
2955                             u64 ino, unsigned type)
2956 {
2957         return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
2958 }
2959 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
2960 {
2961         return ctx->actor(ctx, ".", 1, ctx->pos,
2962                           file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
2963 }
2964 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
2965 {
2966         return ctx->actor(ctx, "..", 2, ctx->pos,
2967                           parent_ino(file->f_path.dentry), DT_DIR) == 0;
2968 }
2969 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
2970 {
2971         if (ctx->pos == 0) {
2972                 if (!dir_emit_dot(file, ctx))
2973                         return false;
2974                 ctx->pos = 1;
2975         }
2976         if (ctx->pos == 1) {
2977                 if (!dir_emit_dotdot(file, ctx))
2978                         return false;
2979                 ctx->pos = 2;
2980         }
2981         return true;
2982 }
2983 static inline bool dir_relax(struct inode *inode)
2984 {
2985         mutex_unlock(&inode->i_mutex);
2986         mutex_lock(&inode->i_mutex);
2987         return !IS_DEADDIR(inode);
2988 }
2989 
2990 #endif /* _LINUX_FS_H */
2991 

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