Version:  2.0.40 2.2.26 2.4.37 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 4.2 4.3 4.4 4.5

Linux/include/linux/fs.h

  1 #ifndef _LINUX_FS_H
  2 #define _LINUX_FS_H
  3 
  4 #include <linux/linkage.h>
  5 #include <linux/wait.h>
  6 #include <linux/kdev_t.h>
  7 #include <linux/dcache.h>
  8 #include <linux/path.h>
  9 #include <linux/stat.h>
 10 #include <linux/cache.h>
 11 #include <linux/list.h>
 12 #include <linux/list_lru.h>
 13 #include <linux/llist.h>
 14 #include <linux/radix-tree.h>
 15 #include <linux/rbtree.h>
 16 #include <linux/init.h>
 17 #include <linux/pid.h>
 18 #include <linux/bug.h>
 19 #include <linux/mutex.h>
 20 #include <linux/rwsem.h>
 21 #include <linux/capability.h>
 22 #include <linux/semaphore.h>
 23 #include <linux/fiemap.h>
 24 #include <linux/rculist_bl.h>
 25 #include <linux/atomic.h>
 26 #include <linux/shrinker.h>
 27 #include <linux/migrate_mode.h>
 28 #include <linux/uidgid.h>
 29 #include <linux/lockdep.h>
 30 #include <linux/percpu-rwsem.h>
 31 #include <linux/blk_types.h>
 32 #include <linux/workqueue.h>
 33 #include <linux/percpu-rwsem.h>
 34 #include <linux/delayed_call.h>
 35 
 36 #include <asm/byteorder.h>
 37 #include <uapi/linux/fs.h>
 38 
 39 struct backing_dev_info;
 40 struct bdi_writeback;
 41 struct export_operations;
 42 struct hd_geometry;
 43 struct iovec;
 44 struct kiocb;
 45 struct kobject;
 46 struct pipe_inode_info;
 47 struct poll_table_struct;
 48 struct kstatfs;
 49 struct vm_area_struct;
 50 struct vfsmount;
 51 struct cred;
 52 struct swap_info_struct;
 53 struct seq_file;
 54 struct workqueue_struct;
 55 struct iov_iter;
 56 
 57 extern void __init inode_init(void);
 58 extern void __init inode_init_early(void);
 59 extern void __init files_init(void);
 60 extern void __init files_maxfiles_init(void);
 61 
 62 extern struct files_stat_struct files_stat;
 63 extern unsigned long get_max_files(void);
 64 extern int sysctl_nr_open;
 65 extern struct inodes_stat_t inodes_stat;
 66 extern int leases_enable, lease_break_time;
 67 extern int sysctl_protected_symlinks;
 68 extern int sysctl_protected_hardlinks;
 69 
 70 struct buffer_head;
 71 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
 72                         struct buffer_head *bh_result, int create);
 73 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
 74                         ssize_t bytes, void *private);
 75 typedef void (dax_iodone_t)(struct buffer_head *bh_map, int uptodate);
 76 
 77 #define MAY_EXEC                0x00000001
 78 #define MAY_WRITE               0x00000002
 79 #define MAY_READ                0x00000004
 80 #define MAY_APPEND              0x00000008
 81 #define MAY_ACCESS              0x00000010
 82 #define MAY_OPEN                0x00000020
 83 #define MAY_CHDIR               0x00000040
 84 /* called from RCU mode, don't block */
 85 #define MAY_NOT_BLOCK           0x00000080
 86 
 87 /*
 88  * flags in file.f_mode.  Note that FMODE_READ and FMODE_WRITE must correspond
 89  * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
 90  */
 91 
 92 /* file is open for reading */
 93 #define FMODE_READ              ((__force fmode_t)0x1)
 94 /* file is open for writing */
 95 #define FMODE_WRITE             ((__force fmode_t)0x2)
 96 /* file is seekable */
 97 #define FMODE_LSEEK             ((__force fmode_t)0x4)
 98 /* file can be accessed using pread */
 99 #define FMODE_PREAD             ((__force fmode_t)0x8)
100 /* file can be accessed using pwrite */
101 #define FMODE_PWRITE            ((__force fmode_t)0x10)
102 /* File is opened for execution with sys_execve / sys_uselib */
103 #define FMODE_EXEC              ((__force fmode_t)0x20)
104 /* File is opened with O_NDELAY (only set for block devices) */
105 #define FMODE_NDELAY            ((__force fmode_t)0x40)
106 /* File is opened with O_EXCL (only set for block devices) */
107 #define FMODE_EXCL              ((__force fmode_t)0x80)
108 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
109    (specialy hack for floppy.c) */
110 #define FMODE_WRITE_IOCTL       ((__force fmode_t)0x100)
111 /* 32bit hashes as llseek() offset (for directories) */
112 #define FMODE_32BITHASH         ((__force fmode_t)0x200)
113 /* 64bit hashes as llseek() offset (for directories) */
114 #define FMODE_64BITHASH         ((__force fmode_t)0x400)
115 
116 /*
117  * Don't update ctime and mtime.
118  *
119  * Currently a special hack for the XFS open_by_handle ioctl, but we'll
120  * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
121  */
122 #define FMODE_NOCMTIME          ((__force fmode_t)0x800)
123 
124 /* Expect random access pattern */
125 #define FMODE_RANDOM            ((__force fmode_t)0x1000)
126 
127 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
128 #define FMODE_UNSIGNED_OFFSET   ((__force fmode_t)0x2000)
129 
130 /* File is opened with O_PATH; almost nothing can be done with it */
131 #define FMODE_PATH              ((__force fmode_t)0x4000)
132 
133 /* File needs atomic accesses to f_pos */
134 #define FMODE_ATOMIC_POS        ((__force fmode_t)0x8000)
135 /* Write access to underlying fs */
136 #define FMODE_WRITER            ((__force fmode_t)0x10000)
137 /* Has read method(s) */
138 #define FMODE_CAN_READ          ((__force fmode_t)0x20000)
139 /* Has write method(s) */
140 #define FMODE_CAN_WRITE         ((__force fmode_t)0x40000)
141 
142 /* File was opened by fanotify and shouldn't generate fanotify events */
143 #define FMODE_NONOTIFY          ((__force fmode_t)0x4000000)
144 
145 /*
146  * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
147  * that indicates that they should check the contents of the iovec are
148  * valid, but not check the memory that the iovec elements
149  * points too.
150  */
151 #define CHECK_IOVEC_ONLY -1
152 
153 /*
154  * The below are the various read and write types that we support. Some of
155  * them include behavioral modifiers that send information down to the
156  * block layer and IO scheduler. Terminology:
157  *
158  *      The block layer uses device plugging to defer IO a little bit, in
159  *      the hope that we will see more IO very shortly. This increases
160  *      coalescing of adjacent IO and thus reduces the number of IOs we
161  *      have to send to the device. It also allows for better queuing,
162  *      if the IO isn't mergeable. If the caller is going to be waiting
163  *      for the IO, then he must ensure that the device is unplugged so
164  *      that the IO is dispatched to the driver.
165  *
166  *      All IO is handled async in Linux. This is fine for background
167  *      writes, but for reads or writes that someone waits for completion
168  *      on, we want to notify the block layer and IO scheduler so that they
169  *      know about it. That allows them to make better scheduling
170  *      decisions. So when the below references 'sync' and 'async', it
171  *      is referencing this priority hint.
172  *
173  * With that in mind, the available types are:
174  *
175  * READ                 A normal read operation. Device will be plugged.
176  * READ_SYNC            A synchronous read. Device is not plugged, caller can
177  *                      immediately wait on this read without caring about
178  *                      unplugging.
179  * READA                Used for read-ahead operations. Lower priority, and the
180  *                      block layer could (in theory) choose to ignore this
181  *                      request if it runs into resource problems.
182  * WRITE                A normal async write. Device will be plugged.
183  * WRITE_SYNC           Synchronous write. Identical to WRITE, but passes down
184  *                      the hint that someone will be waiting on this IO
185  *                      shortly. The write equivalent of READ_SYNC.
186  * WRITE_ODIRECT        Special case write for O_DIRECT only.
187  * WRITE_FLUSH          Like WRITE_SYNC but with preceding cache flush.
188  * WRITE_FUA            Like WRITE_SYNC but data is guaranteed to be on
189  *                      non-volatile media on completion.
190  * WRITE_FLUSH_FUA      Combination of WRITE_FLUSH and FUA. The IO is preceded
191  *                      by a cache flush and data is guaranteed to be on
192  *                      non-volatile media on completion.
193  *
194  */
195 #define RW_MASK                 REQ_WRITE
196 #define RWA_MASK                REQ_RAHEAD
197 
198 #define READ                    0
199 #define WRITE                   RW_MASK
200 #define READA                   RWA_MASK
201 
202 #define READ_SYNC               (READ | REQ_SYNC)
203 #define WRITE_SYNC              (WRITE | REQ_SYNC | REQ_NOIDLE)
204 #define WRITE_ODIRECT           (WRITE | REQ_SYNC)
205 #define WRITE_FLUSH             (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
206 #define WRITE_FUA               (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
207 #define WRITE_FLUSH_FUA         (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
208 
209 /*
210  * Attribute flags.  These should be or-ed together to figure out what
211  * has been changed!
212  */
213 #define ATTR_MODE       (1 << 0)
214 #define ATTR_UID        (1 << 1)
215 #define ATTR_GID        (1 << 2)
216 #define ATTR_SIZE       (1 << 3)
217 #define ATTR_ATIME      (1 << 4)
218 #define ATTR_MTIME      (1 << 5)
219 #define ATTR_CTIME      (1 << 6)
220 #define ATTR_ATIME_SET  (1 << 7)
221 #define ATTR_MTIME_SET  (1 << 8)
222 #define ATTR_FORCE      (1 << 9) /* Not a change, but a change it */
223 #define ATTR_ATTR_FLAG  (1 << 10)
224 #define ATTR_KILL_SUID  (1 << 11)
225 #define ATTR_KILL_SGID  (1 << 12)
226 #define ATTR_FILE       (1 << 13)
227 #define ATTR_KILL_PRIV  (1 << 14)
228 #define ATTR_OPEN       (1 << 15) /* Truncating from open(O_TRUNC) */
229 #define ATTR_TIMES_SET  (1 << 16)
230 
231 /*
232  * Whiteout is represented by a char device.  The following constants define the
233  * mode and device number to use.
234  */
235 #define WHITEOUT_MODE 0
236 #define WHITEOUT_DEV 0
237 
238 /*
239  * This is the Inode Attributes structure, used for notify_change().  It
240  * uses the above definitions as flags, to know which values have changed.
241  * Also, in this manner, a Filesystem can look at only the values it cares
242  * about.  Basically, these are the attributes that the VFS layer can
243  * request to change from the FS layer.
244  *
245  * Derek Atkins <warlord@MIT.EDU> 94-10-20
246  */
247 struct iattr {
248         unsigned int    ia_valid;
249         umode_t         ia_mode;
250         kuid_t          ia_uid;
251         kgid_t          ia_gid;
252         loff_t          ia_size;
253         struct timespec ia_atime;
254         struct timespec ia_mtime;
255         struct timespec ia_ctime;
256 
257         /*
258          * Not an attribute, but an auxiliary info for filesystems wanting to
259          * implement an ftruncate() like method.  NOTE: filesystem should
260          * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
261          */
262         struct file     *ia_file;
263 };
264 
265 /*
266  * Includes for diskquotas.
267  */
268 #include <linux/quota.h>
269 
270 /*
271  * Maximum number of layers of fs stack.  Needs to be limited to
272  * prevent kernel stack overflow
273  */
274 #define FILESYSTEM_MAX_STACK_DEPTH 2
275 
276 /** 
277  * enum positive_aop_returns - aop return codes with specific semantics
278  *
279  * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
280  *                          completed, that the page is still locked, and
281  *                          should be considered active.  The VM uses this hint
282  *                          to return the page to the active list -- it won't
283  *                          be a candidate for writeback again in the near
284  *                          future.  Other callers must be careful to unlock
285  *                          the page if they get this return.  Returned by
286  *                          writepage(); 
287  *
288  * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
289  *                      unlocked it and the page might have been truncated.
290  *                      The caller should back up to acquiring a new page and
291  *                      trying again.  The aop will be taking reasonable
292  *                      precautions not to livelock.  If the caller held a page
293  *                      reference, it should drop it before retrying.  Returned
294  *                      by readpage().
295  *
296  * address_space_operation functions return these large constants to indicate
297  * special semantics to the caller.  These are much larger than the bytes in a
298  * page to allow for functions that return the number of bytes operated on in a
299  * given page.
300  */
301 
302 enum positive_aop_returns {
303         AOP_WRITEPAGE_ACTIVATE  = 0x80000,
304         AOP_TRUNCATED_PAGE      = 0x80001,
305 };
306 
307 #define AOP_FLAG_UNINTERRUPTIBLE        0x0001 /* will not do a short write */
308 #define AOP_FLAG_CONT_EXPAND            0x0002 /* called from cont_expand */
309 #define AOP_FLAG_NOFS                   0x0004 /* used by filesystem to direct
310                                                 * helper code (eg buffer layer)
311                                                 * to clear GFP_FS from alloc */
312 
313 /*
314  * oh the beauties of C type declarations.
315  */
316 struct page;
317 struct address_space;
318 struct writeback_control;
319 
320 #define IOCB_EVENTFD            (1 << 0)
321 #define IOCB_APPEND             (1 << 1)
322 #define IOCB_DIRECT             (1 << 2)
323 
324 struct kiocb {
325         struct file             *ki_filp;
326         loff_t                  ki_pos;
327         void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
328         void                    *private;
329         int                     ki_flags;
330 };
331 
332 static inline bool is_sync_kiocb(struct kiocb *kiocb)
333 {
334         return kiocb->ki_complete == NULL;
335 }
336 
337 static inline int iocb_flags(struct file *file);
338 
339 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
340 {
341         *kiocb = (struct kiocb) {
342                 .ki_filp = filp,
343                 .ki_flags = iocb_flags(filp),
344         };
345 }
346 
347 /*
348  * "descriptor" for what we're up to with a read.
349  * This allows us to use the same read code yet
350  * have multiple different users of the data that
351  * we read from a file.
352  *
353  * The simplest case just copies the data to user
354  * mode.
355  */
356 typedef struct {
357         size_t written;
358         size_t count;
359         union {
360                 char __user *buf;
361                 void *data;
362         } arg;
363         int error;
364 } read_descriptor_t;
365 
366 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
367                 unsigned long, unsigned long);
368 
369 struct address_space_operations {
370         int (*writepage)(struct page *page, struct writeback_control *wbc);
371         int (*readpage)(struct file *, struct page *);
372 
373         /* Write back some dirty pages from this mapping. */
374         int (*writepages)(struct address_space *, struct writeback_control *);
375 
376         /* Set a page dirty.  Return true if this dirtied it */
377         int (*set_page_dirty)(struct page *page);
378 
379         int (*readpages)(struct file *filp, struct address_space *mapping,
380                         struct list_head *pages, unsigned nr_pages);
381 
382         int (*write_begin)(struct file *, struct address_space *mapping,
383                                 loff_t pos, unsigned len, unsigned flags,
384                                 struct page **pagep, void **fsdata);
385         int (*write_end)(struct file *, struct address_space *mapping,
386                                 loff_t pos, unsigned len, unsigned copied,
387                                 struct page *page, void *fsdata);
388 
389         /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
390         sector_t (*bmap)(struct address_space *, sector_t);
391         void (*invalidatepage) (struct page *, unsigned int, unsigned int);
392         int (*releasepage) (struct page *, gfp_t);
393         void (*freepage)(struct page *);
394         ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter, loff_t offset);
395         /*
396          * migrate the contents of a page to the specified target. If
397          * migrate_mode is MIGRATE_ASYNC, it must not block.
398          */
399         int (*migratepage) (struct address_space *,
400                         struct page *, struct page *, enum migrate_mode);
401         int (*launder_page) (struct page *);
402         int (*is_partially_uptodate) (struct page *, unsigned long,
403                                         unsigned long);
404         void (*is_dirty_writeback) (struct page *, bool *, bool *);
405         int (*error_remove_page)(struct address_space *, struct page *);
406 
407         /* swapfile support */
408         int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
409                                 sector_t *span);
410         void (*swap_deactivate)(struct file *file);
411 };
412 
413 extern const struct address_space_operations empty_aops;
414 
415 /*
416  * pagecache_write_begin/pagecache_write_end must be used by general code
417  * to write into the pagecache.
418  */
419 int pagecache_write_begin(struct file *, struct address_space *mapping,
420                                 loff_t pos, unsigned len, unsigned flags,
421                                 struct page **pagep, void **fsdata);
422 
423 int pagecache_write_end(struct file *, struct address_space *mapping,
424                                 loff_t pos, unsigned len, unsigned copied,
425                                 struct page *page, void *fsdata);
426 
427 struct address_space {
428         struct inode            *host;          /* owner: inode, block_device */
429         struct radix_tree_root  page_tree;      /* radix tree of all pages */
430         spinlock_t              tree_lock;      /* and lock protecting it */
431         atomic_t                i_mmap_writable;/* count VM_SHARED mappings */
432         struct rb_root          i_mmap;         /* tree of private and shared mappings */
433         struct rw_semaphore     i_mmap_rwsem;   /* protect tree, count, list */
434         /* Protected by tree_lock together with the radix tree */
435         unsigned long           nrpages;        /* number of total pages */
436         /* number of shadow or DAX exceptional entries */
437         unsigned long           nrexceptional;
438         pgoff_t                 writeback_index;/* writeback starts here */
439         const struct address_space_operations *a_ops;   /* methods */
440         unsigned long           flags;          /* error bits/gfp mask */
441         spinlock_t              private_lock;   /* for use by the address_space */
442         struct list_head        private_list;   /* ditto */
443         void                    *private_data;  /* ditto */
444 } __attribute__((aligned(sizeof(long))));
445         /*
446          * On most architectures that alignment is already the case; but
447          * must be enforced here for CRIS, to let the least significant bit
448          * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
449          */
450 struct request_queue;
451 
452 struct block_device {
453         dev_t                   bd_dev;  /* not a kdev_t - it's a search key */
454         int                     bd_openers;
455         struct inode *          bd_inode;       /* will die */
456         struct super_block *    bd_super;
457         struct mutex            bd_mutex;       /* open/close mutex */
458         struct list_head        bd_inodes;
459         void *                  bd_claiming;
460         void *                  bd_holder;
461         int                     bd_holders;
462         bool                    bd_write_holder;
463 #ifdef CONFIG_SYSFS
464         struct list_head        bd_holder_disks;
465 #endif
466         struct block_device *   bd_contains;
467         unsigned                bd_block_size;
468         struct hd_struct *      bd_part;
469         /* number of times partitions within this device have been opened. */
470         unsigned                bd_part_count;
471         int                     bd_invalidated;
472         struct gendisk *        bd_disk;
473         struct request_queue *  bd_queue;
474         struct list_head        bd_list;
475         /*
476          * Private data.  You must have bd_claim'ed the block_device
477          * to use this.  NOTE:  bd_claim allows an owner to claim
478          * the same device multiple times, the owner must take special
479          * care to not mess up bd_private for that case.
480          */
481         unsigned long           bd_private;
482 
483         /* The counter of freeze processes */
484         int                     bd_fsfreeze_count;
485         /* Mutex for freeze */
486         struct mutex            bd_fsfreeze_mutex;
487 };
488 
489 /*
490  * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
491  * radix trees
492  */
493 #define PAGECACHE_TAG_DIRTY     0
494 #define PAGECACHE_TAG_WRITEBACK 1
495 #define PAGECACHE_TAG_TOWRITE   2
496 
497 int mapping_tagged(struct address_space *mapping, int tag);
498 
499 static inline void i_mmap_lock_write(struct address_space *mapping)
500 {
501         down_write(&mapping->i_mmap_rwsem);
502 }
503 
504 static inline void i_mmap_unlock_write(struct address_space *mapping)
505 {
506         up_write(&mapping->i_mmap_rwsem);
507 }
508 
509 static inline void i_mmap_lock_read(struct address_space *mapping)
510 {
511         down_read(&mapping->i_mmap_rwsem);
512 }
513 
514 static inline void i_mmap_unlock_read(struct address_space *mapping)
515 {
516         up_read(&mapping->i_mmap_rwsem);
517 }
518 
519 /*
520  * Might pages of this file be mapped into userspace?
521  */
522 static inline int mapping_mapped(struct address_space *mapping)
523 {
524         return  !RB_EMPTY_ROOT(&mapping->i_mmap);
525 }
526 
527 /*
528  * Might pages of this file have been modified in userspace?
529  * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
530  * marks vma as VM_SHARED if it is shared, and the file was opened for
531  * writing i.e. vma may be mprotected writable even if now readonly.
532  *
533  * If i_mmap_writable is negative, no new writable mappings are allowed. You
534  * can only deny writable mappings, if none exists right now.
535  */
536 static inline int mapping_writably_mapped(struct address_space *mapping)
537 {
538         return atomic_read(&mapping->i_mmap_writable) > 0;
539 }
540 
541 static inline int mapping_map_writable(struct address_space *mapping)
542 {
543         return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
544                 0 : -EPERM;
545 }
546 
547 static inline void mapping_unmap_writable(struct address_space *mapping)
548 {
549         atomic_dec(&mapping->i_mmap_writable);
550 }
551 
552 static inline int mapping_deny_writable(struct address_space *mapping)
553 {
554         return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
555                 0 : -EBUSY;
556 }
557 
558 static inline void mapping_allow_writable(struct address_space *mapping)
559 {
560         atomic_inc(&mapping->i_mmap_writable);
561 }
562 
563 /*
564  * Use sequence counter to get consistent i_size on 32-bit processors.
565  */
566 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
567 #include <linux/seqlock.h>
568 #define __NEED_I_SIZE_ORDERED
569 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
570 #else
571 #define i_size_ordered_init(inode) do { } while (0)
572 #endif
573 
574 struct posix_acl;
575 #define ACL_NOT_CACHED ((void *)(-1))
576 
577 #define IOP_FASTPERM    0x0001
578 #define IOP_LOOKUP      0x0002
579 #define IOP_NOFOLLOW    0x0004
580 
581 /*
582  * Keep mostly read-only and often accessed (especially for
583  * the RCU path lookup and 'stat' data) fields at the beginning
584  * of the 'struct inode'
585  */
586 struct inode {
587         umode_t                 i_mode;
588         unsigned short          i_opflags;
589         kuid_t                  i_uid;
590         kgid_t                  i_gid;
591         unsigned int            i_flags;
592 
593 #ifdef CONFIG_FS_POSIX_ACL
594         struct posix_acl        *i_acl;
595         struct posix_acl        *i_default_acl;
596 #endif
597 
598         const struct inode_operations   *i_op;
599         struct super_block      *i_sb;
600         struct address_space    *i_mapping;
601 
602 #ifdef CONFIG_SECURITY
603         void                    *i_security;
604 #endif
605 
606         /* Stat data, not accessed from path walking */
607         unsigned long           i_ino;
608         /*
609          * Filesystems may only read i_nlink directly.  They shall use the
610          * following functions for modification:
611          *
612          *    (set|clear|inc|drop)_nlink
613          *    inode_(inc|dec)_link_count
614          */
615         union {
616                 const unsigned int i_nlink;
617                 unsigned int __i_nlink;
618         };
619         dev_t                   i_rdev;
620         loff_t                  i_size;
621         struct timespec         i_atime;
622         struct timespec         i_mtime;
623         struct timespec         i_ctime;
624         spinlock_t              i_lock; /* i_blocks, i_bytes, maybe i_size */
625         unsigned short          i_bytes;
626         unsigned int            i_blkbits;
627         blkcnt_t                i_blocks;
628 
629 #ifdef __NEED_I_SIZE_ORDERED
630         seqcount_t              i_size_seqcount;
631 #endif
632 
633         /* Misc */
634         unsigned long           i_state;
635         struct mutex            i_mutex;
636 
637         unsigned long           dirtied_when;   /* jiffies of first dirtying */
638         unsigned long           dirtied_time_when;
639 
640         struct hlist_node       i_hash;
641         struct list_head        i_io_list;      /* backing dev IO list */
642 #ifdef CONFIG_CGROUP_WRITEBACK
643         struct bdi_writeback    *i_wb;          /* the associated cgroup wb */
644 
645         /* foreign inode detection, see wbc_detach_inode() */
646         int                     i_wb_frn_winner;
647         u16                     i_wb_frn_avg_time;
648         u16                     i_wb_frn_history;
649 #endif
650         struct list_head        i_lru;          /* inode LRU list */
651         struct list_head        i_sb_list;
652         union {
653                 struct hlist_head       i_dentry;
654                 struct rcu_head         i_rcu;
655         };
656         u64                     i_version;
657         atomic_t                i_count;
658         atomic_t                i_dio_count;
659         atomic_t                i_writecount;
660 #ifdef CONFIG_IMA
661         atomic_t                i_readcount; /* struct files open RO */
662 #endif
663         const struct file_operations    *i_fop; /* former ->i_op->default_file_ops */
664         struct file_lock_context        *i_flctx;
665         struct address_space    i_data;
666         struct list_head        i_devices;
667         union {
668                 struct pipe_inode_info  *i_pipe;
669                 struct block_device     *i_bdev;
670                 struct cdev             *i_cdev;
671                 char                    *i_link;
672         };
673 
674         __u32                   i_generation;
675 
676 #ifdef CONFIG_FSNOTIFY
677         __u32                   i_fsnotify_mask; /* all events this inode cares about */
678         struct hlist_head       i_fsnotify_marks;
679 #endif
680 
681         void                    *i_private; /* fs or device private pointer */
682 };
683 
684 static inline int inode_unhashed(struct inode *inode)
685 {
686         return hlist_unhashed(&inode->i_hash);
687 }
688 
689 /*
690  * inode->i_mutex nesting subclasses for the lock validator:
691  *
692  * 0: the object of the current VFS operation
693  * 1: parent
694  * 2: child/target
695  * 3: xattr
696  * 4: second non-directory
697  * 5: second parent (when locking independent directories in rename)
698  *
699  * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
700  * non-directories at once.
701  *
702  * The locking order between these classes is
703  * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
704  */
705 enum inode_i_mutex_lock_class
706 {
707         I_MUTEX_NORMAL,
708         I_MUTEX_PARENT,
709         I_MUTEX_CHILD,
710         I_MUTEX_XATTR,
711         I_MUTEX_NONDIR2,
712         I_MUTEX_PARENT2,
713 };
714 
715 static inline void inode_lock(struct inode *inode)
716 {
717         mutex_lock(&inode->i_mutex);
718 }
719 
720 static inline void inode_unlock(struct inode *inode)
721 {
722         mutex_unlock(&inode->i_mutex);
723 }
724 
725 static inline int inode_trylock(struct inode *inode)
726 {
727         return mutex_trylock(&inode->i_mutex);
728 }
729 
730 static inline int inode_is_locked(struct inode *inode)
731 {
732         return mutex_is_locked(&inode->i_mutex);
733 }
734 
735 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
736 {
737         mutex_lock_nested(&inode->i_mutex, subclass);
738 }
739 
740 void lock_two_nondirectories(struct inode *, struct inode*);
741 void unlock_two_nondirectories(struct inode *, struct inode*);
742 
743 /*
744  * NOTE: in a 32bit arch with a preemptable kernel and
745  * an UP compile the i_size_read/write must be atomic
746  * with respect to the local cpu (unlike with preempt disabled),
747  * but they don't need to be atomic with respect to other cpus like in
748  * true SMP (so they need either to either locally disable irq around
749  * the read or for example on x86 they can be still implemented as a
750  * cmpxchg8b without the need of the lock prefix). For SMP compiles
751  * and 64bit archs it makes no difference if preempt is enabled or not.
752  */
753 static inline loff_t i_size_read(const struct inode *inode)
754 {
755 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
756         loff_t i_size;
757         unsigned int seq;
758 
759         do {
760                 seq = read_seqcount_begin(&inode->i_size_seqcount);
761                 i_size = inode->i_size;
762         } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
763         return i_size;
764 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
765         loff_t i_size;
766 
767         preempt_disable();
768         i_size = inode->i_size;
769         preempt_enable();
770         return i_size;
771 #else
772         return inode->i_size;
773 #endif
774 }
775 
776 /*
777  * NOTE: unlike i_size_read(), i_size_write() does need locking around it
778  * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
779  * can be lost, resulting in subsequent i_size_read() calls spinning forever.
780  */
781 static inline void i_size_write(struct inode *inode, loff_t i_size)
782 {
783 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
784         preempt_disable();
785         write_seqcount_begin(&inode->i_size_seqcount);
786         inode->i_size = i_size;
787         write_seqcount_end(&inode->i_size_seqcount);
788         preempt_enable();
789 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
790         preempt_disable();
791         inode->i_size = i_size;
792         preempt_enable();
793 #else
794         inode->i_size = i_size;
795 #endif
796 }
797 
798 /* Helper functions so that in most cases filesystems will
799  * not need to deal directly with kuid_t and kgid_t and can
800  * instead deal with the raw numeric values that are stored
801  * in the filesystem.
802  */
803 static inline uid_t i_uid_read(const struct inode *inode)
804 {
805         return from_kuid(&init_user_ns, inode->i_uid);
806 }
807 
808 static inline gid_t i_gid_read(const struct inode *inode)
809 {
810         return from_kgid(&init_user_ns, inode->i_gid);
811 }
812 
813 static inline void i_uid_write(struct inode *inode, uid_t uid)
814 {
815         inode->i_uid = make_kuid(&init_user_ns, uid);
816 }
817 
818 static inline void i_gid_write(struct inode *inode, gid_t gid)
819 {
820         inode->i_gid = make_kgid(&init_user_ns, gid);
821 }
822 
823 static inline unsigned iminor(const struct inode *inode)
824 {
825         return MINOR(inode->i_rdev);
826 }
827 
828 static inline unsigned imajor(const struct inode *inode)
829 {
830         return MAJOR(inode->i_rdev);
831 }
832 
833 extern struct block_device *I_BDEV(struct inode *inode);
834 
835 struct fown_struct {
836         rwlock_t lock;          /* protects pid, uid, euid fields */
837         struct pid *pid;        /* pid or -pgrp where SIGIO should be sent */
838         enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
839         kuid_t uid, euid;       /* uid/euid of process setting the owner */
840         int signum;             /* posix.1b rt signal to be delivered on IO */
841 };
842 
843 /*
844  * Track a single file's readahead state
845  */
846 struct file_ra_state {
847         pgoff_t start;                  /* where readahead started */
848         unsigned int size;              /* # of readahead pages */
849         unsigned int async_size;        /* do asynchronous readahead when
850                                            there are only # of pages ahead */
851 
852         unsigned int ra_pages;          /* Maximum readahead window */
853         unsigned int mmap_miss;         /* Cache miss stat for mmap accesses */
854         loff_t prev_pos;                /* Cache last read() position */
855 };
856 
857 /*
858  * Check if @index falls in the readahead windows.
859  */
860 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
861 {
862         return (index >= ra->start &&
863                 index <  ra->start + ra->size);
864 }
865 
866 struct file {
867         union {
868                 struct llist_node       fu_llist;
869                 struct rcu_head         fu_rcuhead;
870         } f_u;
871         struct path             f_path;
872         struct inode            *f_inode;       /* cached value */
873         const struct file_operations    *f_op;
874 
875         /*
876          * Protects f_ep_links, f_flags.
877          * Must not be taken from IRQ context.
878          */
879         spinlock_t              f_lock;
880         atomic_long_t           f_count;
881         unsigned int            f_flags;
882         fmode_t                 f_mode;
883         struct mutex            f_pos_lock;
884         loff_t                  f_pos;
885         struct fown_struct      f_owner;
886         const struct cred       *f_cred;
887         struct file_ra_state    f_ra;
888 
889         u64                     f_version;
890 #ifdef CONFIG_SECURITY
891         void                    *f_security;
892 #endif
893         /* needed for tty driver, and maybe others */
894         void                    *private_data;
895 
896 #ifdef CONFIG_EPOLL
897         /* Used by fs/eventpoll.c to link all the hooks to this file */
898         struct list_head        f_ep_links;
899         struct list_head        f_tfile_llink;
900 #endif /* #ifdef CONFIG_EPOLL */
901         struct address_space    *f_mapping;
902 } __attribute__((aligned(4)));  /* lest something weird decides that 2 is OK */
903 
904 struct file_handle {
905         __u32 handle_bytes;
906         int handle_type;
907         /* file identifier */
908         unsigned char f_handle[0];
909 };
910 
911 static inline struct file *get_file(struct file *f)
912 {
913         atomic_long_inc(&f->f_count);
914         return f;
915 }
916 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
917 #define fput_atomic(x)  atomic_long_add_unless(&(x)->f_count, -1, 1)
918 #define file_count(x)   atomic_long_read(&(x)->f_count)
919 
920 #define MAX_NON_LFS     ((1UL<<31) - 1)
921 
922 /* Page cache limit. The filesystems should put that into their s_maxbytes 
923    limits, otherwise bad things can happen in VM. */ 
924 #if BITS_PER_LONG==32
925 #define MAX_LFS_FILESIZE        (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1) 
926 #elif BITS_PER_LONG==64
927 #define MAX_LFS_FILESIZE        ((loff_t)0x7fffffffffffffffLL)
928 #endif
929 
930 #define FL_POSIX        1
931 #define FL_FLOCK        2
932 #define FL_DELEG        4       /* NFSv4 delegation */
933 #define FL_ACCESS       8       /* not trying to lock, just looking */
934 #define FL_EXISTS       16      /* when unlocking, test for existence */
935 #define FL_LEASE        32      /* lease held on this file */
936 #define FL_CLOSE        64      /* unlock on close */
937 #define FL_SLEEP        128     /* A blocking lock */
938 #define FL_DOWNGRADE_PENDING    256 /* Lease is being downgraded */
939 #define FL_UNLOCK_PENDING       512 /* Lease is being broken */
940 #define FL_OFDLCK       1024    /* lock is "owned" by struct file */
941 #define FL_LAYOUT       2048    /* outstanding pNFS layout */
942 
943 /*
944  * Special return value from posix_lock_file() and vfs_lock_file() for
945  * asynchronous locking.
946  */
947 #define FILE_LOCK_DEFERRED 1
948 
949 /* legacy typedef, should eventually be removed */
950 typedef void *fl_owner_t;
951 
952 struct file_lock;
953 
954 struct file_lock_operations {
955         void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
956         void (*fl_release_private)(struct file_lock *);
957 };
958 
959 struct lock_manager_operations {
960         int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
961         unsigned long (*lm_owner_key)(struct file_lock *);
962         fl_owner_t (*lm_get_owner)(fl_owner_t);
963         void (*lm_put_owner)(fl_owner_t);
964         void (*lm_notify)(struct file_lock *);  /* unblock callback */
965         int (*lm_grant)(struct file_lock *, int);
966         bool (*lm_break)(struct file_lock *);
967         int (*lm_change)(struct file_lock *, int, struct list_head *);
968         void (*lm_setup)(struct file_lock *, void **);
969 };
970 
971 struct lock_manager {
972         struct list_head list;
973         /*
974          * NFSv4 and up also want opens blocked during the grace period;
975          * NLM doesn't care:
976          */
977         bool block_opens;
978 };
979 
980 struct net;
981 void locks_start_grace(struct net *, struct lock_manager *);
982 void locks_end_grace(struct lock_manager *);
983 int locks_in_grace(struct net *);
984 int opens_in_grace(struct net *);
985 
986 /* that will die - we need it for nfs_lock_info */
987 #include <linux/nfs_fs_i.h>
988 
989 /*
990  * struct file_lock represents a generic "file lock". It's used to represent
991  * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
992  * note that the same struct is used to represent both a request for a lock and
993  * the lock itself, but the same object is never used for both.
994  *
995  * FIXME: should we create a separate "struct lock_request" to help distinguish
996  * these two uses?
997  *
998  * The varous i_flctx lists are ordered by:
999  *
1000  * 1) lock owner
1001  * 2) lock range start
1002  * 3) lock range end
1003  *
1004  * Obviously, the last two criteria only matter for POSIX locks.
1005  */
1006 struct file_lock {
1007         struct file_lock *fl_next;      /* singly linked list for this inode  */
1008         struct list_head fl_list;       /* link into file_lock_context */
1009         struct hlist_node fl_link;      /* node in global lists */
1010         struct list_head fl_block;      /* circular list of blocked processes */
1011         fl_owner_t fl_owner;
1012         unsigned int fl_flags;
1013         unsigned char fl_type;
1014         unsigned int fl_pid;
1015         int fl_link_cpu;                /* what cpu's list is this on? */
1016         struct pid *fl_nspid;
1017         wait_queue_head_t fl_wait;
1018         struct file *fl_file;
1019         loff_t fl_start;
1020         loff_t fl_end;
1021 
1022         struct fasync_struct *  fl_fasync; /* for lease break notifications */
1023         /* for lease breaks: */
1024         unsigned long fl_break_time;
1025         unsigned long fl_downgrade_time;
1026 
1027         const struct file_lock_operations *fl_ops;      /* Callbacks for filesystems */
1028         const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1029         union {
1030                 struct nfs_lock_info    nfs_fl;
1031                 struct nfs4_lock_info   nfs4_fl;
1032                 struct {
1033                         struct list_head link;  /* link in AFS vnode's pending_locks list */
1034                         int state;              /* state of grant or error if -ve */
1035                 } afs;
1036         } fl_u;
1037 };
1038 
1039 struct file_lock_context {
1040         spinlock_t              flc_lock;
1041         struct list_head        flc_flock;
1042         struct list_head        flc_posix;
1043         struct list_head        flc_lease;
1044 };
1045 
1046 /* The following constant reflects the upper bound of the file/locking space */
1047 #ifndef OFFSET_MAX
1048 #define INT_LIMIT(x)    (~((x)1 << (sizeof(x)*8 - 1)))
1049 #define OFFSET_MAX      INT_LIMIT(loff_t)
1050 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1051 #endif
1052 
1053 #include <linux/fcntl.h>
1054 
1055 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1056 
1057 #ifdef CONFIG_FILE_LOCKING
1058 extern int fcntl_getlk(struct file *, unsigned int, struct flock __user *);
1059 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1060                         struct flock __user *);
1061 
1062 #if BITS_PER_LONG == 32
1063 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 __user *);
1064 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1065                         struct flock64 __user *);
1066 #endif
1067 
1068 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1069 extern int fcntl_getlease(struct file *filp);
1070 
1071 /* fs/locks.c */
1072 void locks_free_lock_context(struct inode *inode);
1073 void locks_free_lock(struct file_lock *fl);
1074 extern void locks_init_lock(struct file_lock *);
1075 extern struct file_lock * locks_alloc_lock(void);
1076 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1077 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1078 extern void locks_remove_posix(struct file *, fl_owner_t);
1079 extern void locks_remove_file(struct file *);
1080 extern void locks_release_private(struct file_lock *);
1081 extern void posix_test_lock(struct file *, struct file_lock *);
1082 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1083 extern int posix_unblock_lock(struct file_lock *);
1084 extern int vfs_test_lock(struct file *, struct file_lock *);
1085 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1086 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1087 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1088 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1089 extern void lease_get_mtime(struct inode *, struct timespec *time);
1090 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1091 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1092 extern int lease_modify(struct file_lock *, int, struct list_head *);
1093 struct files_struct;
1094 extern void show_fd_locks(struct seq_file *f,
1095                          struct file *filp, struct files_struct *files);
1096 #else /* !CONFIG_FILE_LOCKING */
1097 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1098                               struct flock __user *user)
1099 {
1100         return -EINVAL;
1101 }
1102 
1103 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1104                               unsigned int cmd, struct flock __user *user)
1105 {
1106         return -EACCES;
1107 }
1108 
1109 #if BITS_PER_LONG == 32
1110 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1111                                 struct flock64 __user *user)
1112 {
1113         return -EINVAL;
1114 }
1115 
1116 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1117                                 unsigned int cmd, struct flock64 __user *user)
1118 {
1119         return -EACCES;
1120 }
1121 #endif
1122 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1123 {
1124         return -EINVAL;
1125 }
1126 
1127 static inline int fcntl_getlease(struct file *filp)
1128 {
1129         return F_UNLCK;
1130 }
1131 
1132 static inline void
1133 locks_free_lock_context(struct inode *inode)
1134 {
1135 }
1136 
1137 static inline void locks_init_lock(struct file_lock *fl)
1138 {
1139         return;
1140 }
1141 
1142 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1143 {
1144         return;
1145 }
1146 
1147 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1148 {
1149         return;
1150 }
1151 
1152 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1153 {
1154         return;
1155 }
1156 
1157 static inline void locks_remove_file(struct file *filp)
1158 {
1159         return;
1160 }
1161 
1162 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1163 {
1164         return;
1165 }
1166 
1167 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1168                                   struct file_lock *conflock)
1169 {
1170         return -ENOLCK;
1171 }
1172 
1173 static inline int posix_unblock_lock(struct file_lock *waiter)
1174 {
1175         return -ENOENT;
1176 }
1177 
1178 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1179 {
1180         return 0;
1181 }
1182 
1183 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1184                                 struct file_lock *fl, struct file_lock *conf)
1185 {
1186         return -ENOLCK;
1187 }
1188 
1189 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1190 {
1191         return 0;
1192 }
1193 
1194 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1195 {
1196         return -ENOLCK;
1197 }
1198 
1199 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1200 {
1201         return 0;
1202 }
1203 
1204 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1205 {
1206         return;
1207 }
1208 
1209 static inline int generic_setlease(struct file *filp, long arg,
1210                                     struct file_lock **flp, void **priv)
1211 {
1212         return -EINVAL;
1213 }
1214 
1215 static inline int vfs_setlease(struct file *filp, long arg,
1216                                struct file_lock **lease, void **priv)
1217 {
1218         return -EINVAL;
1219 }
1220 
1221 static inline int lease_modify(struct file_lock *fl, int arg,
1222                                struct list_head *dispose)
1223 {
1224         return -EINVAL;
1225 }
1226 
1227 struct files_struct;
1228 static inline void show_fd_locks(struct seq_file *f,
1229                         struct file *filp, struct files_struct *files) {}
1230 #endif /* !CONFIG_FILE_LOCKING */
1231 
1232 static inline struct inode *file_inode(const struct file *f)
1233 {
1234         return f->f_inode;
1235 }
1236 
1237 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1238 {
1239         return locks_lock_inode_wait(file_inode(filp), fl);
1240 }
1241 
1242 struct fasync_struct {
1243         spinlock_t              fa_lock;
1244         int                     magic;
1245         int                     fa_fd;
1246         struct fasync_struct    *fa_next; /* singly linked list */
1247         struct file             *fa_file;
1248         struct rcu_head         fa_rcu;
1249 };
1250 
1251 #define FASYNC_MAGIC 0x4601
1252 
1253 /* SMP safe fasync helpers: */
1254 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1255 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1256 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1257 extern struct fasync_struct *fasync_alloc(void);
1258 extern void fasync_free(struct fasync_struct *);
1259 
1260 /* can be called from interrupts */
1261 extern void kill_fasync(struct fasync_struct **, int, int);
1262 
1263 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1264 extern void f_setown(struct file *filp, unsigned long arg, int force);
1265 extern void f_delown(struct file *filp);
1266 extern pid_t f_getown(struct file *filp);
1267 extern int send_sigurg(struct fown_struct *fown);
1268 
1269 struct mm_struct;
1270 
1271 /*
1272  *      Umount options
1273  */
1274 
1275 #define MNT_FORCE       0x00000001      /* Attempt to forcibily umount */
1276 #define MNT_DETACH      0x00000002      /* Just detach from the tree */
1277 #define MNT_EXPIRE      0x00000004      /* Mark for expiry */
1278 #define UMOUNT_NOFOLLOW 0x00000008      /* Don't follow symlink on umount */
1279 #define UMOUNT_UNUSED   0x80000000      /* Flag guaranteed to be unused */
1280 
1281 /* sb->s_iflags */
1282 #define SB_I_CGROUPWB   0x00000001      /* cgroup-aware writeback enabled */
1283 #define SB_I_NOEXEC     0x00000002      /* Ignore executables on this fs */
1284 
1285 /* Possible states of 'frozen' field */
1286 enum {
1287         SB_UNFROZEN = 0,                /* FS is unfrozen */
1288         SB_FREEZE_WRITE = 1,            /* Writes, dir ops, ioctls frozen */
1289         SB_FREEZE_PAGEFAULT = 2,        /* Page faults stopped as well */
1290         SB_FREEZE_FS = 3,               /* For internal FS use (e.g. to stop
1291                                          * internal threads if needed) */
1292         SB_FREEZE_COMPLETE = 4,         /* ->freeze_fs finished successfully */
1293 };
1294 
1295 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1296 
1297 struct sb_writers {
1298         int                             frozen;         /* Is sb frozen? */
1299         wait_queue_head_t               wait_unfrozen;  /* for get_super_thawed() */
1300         struct percpu_rw_semaphore      rw_sem[SB_FREEZE_LEVELS];
1301 };
1302 
1303 struct super_block {
1304         struct list_head        s_list;         /* Keep this first */
1305         dev_t                   s_dev;          /* search index; _not_ kdev_t */
1306         unsigned char           s_blocksize_bits;
1307         unsigned long           s_blocksize;
1308         loff_t                  s_maxbytes;     /* Max file size */
1309         struct file_system_type *s_type;
1310         const struct super_operations   *s_op;
1311         const struct dquot_operations   *dq_op;
1312         const struct quotactl_ops       *s_qcop;
1313         const struct export_operations *s_export_op;
1314         unsigned long           s_flags;
1315         unsigned long           s_iflags;       /* internal SB_I_* flags */
1316         unsigned long           s_magic;
1317         struct dentry           *s_root;
1318         struct rw_semaphore     s_umount;
1319         int                     s_count;
1320         atomic_t                s_active;
1321 #ifdef CONFIG_SECURITY
1322         void                    *s_security;
1323 #endif
1324         const struct xattr_handler **s_xattr;
1325 
1326         struct hlist_bl_head    s_anon;         /* anonymous dentries for (nfs) exporting */
1327         struct list_head        s_mounts;       /* list of mounts; _not_ for fs use */
1328         struct block_device     *s_bdev;
1329         struct backing_dev_info *s_bdi;
1330         struct mtd_info         *s_mtd;
1331         struct hlist_node       s_instances;
1332         unsigned int            s_quota_types;  /* Bitmask of supported quota types */
1333         struct quota_info       s_dquot;        /* Diskquota specific options */
1334 
1335         struct sb_writers       s_writers;
1336 
1337         char s_id[32];                          /* Informational name */
1338         u8 s_uuid[16];                          /* UUID */
1339 
1340         void                    *s_fs_info;     /* Filesystem private info */
1341         unsigned int            s_max_links;
1342         fmode_t                 s_mode;
1343 
1344         /* Granularity of c/m/atime in ns.
1345            Cannot be worse than a second */
1346         u32                s_time_gran;
1347 
1348         /*
1349          * The next field is for VFS *only*. No filesystems have any business
1350          * even looking at it. You had been warned.
1351          */
1352         struct mutex s_vfs_rename_mutex;        /* Kludge */
1353 
1354         /*
1355          * Filesystem subtype.  If non-empty the filesystem type field
1356          * in /proc/mounts will be "type.subtype"
1357          */
1358         char *s_subtype;
1359 
1360         /*
1361          * Saved mount options for lazy filesystems using
1362          * generic_show_options()
1363          */
1364         char __rcu *s_options;
1365         const struct dentry_operations *s_d_op; /* default d_op for dentries */
1366 
1367         /*
1368          * Saved pool identifier for cleancache (-1 means none)
1369          */
1370         int cleancache_poolid;
1371 
1372         struct shrinker s_shrink;       /* per-sb shrinker handle */
1373 
1374         /* Number of inodes with nlink == 0 but still referenced */
1375         atomic_long_t s_remove_count;
1376 
1377         /* Being remounted read-only */
1378         int s_readonly_remount;
1379 
1380         /* AIO completions deferred from interrupt context */
1381         struct workqueue_struct *s_dio_done_wq;
1382         struct hlist_head s_pins;
1383 
1384         /*
1385          * Keep the lru lists last in the structure so they always sit on their
1386          * own individual cachelines.
1387          */
1388         struct list_lru         s_dentry_lru ____cacheline_aligned_in_smp;
1389         struct list_lru         s_inode_lru ____cacheline_aligned_in_smp;
1390         struct rcu_head         rcu;
1391         struct work_struct      destroy_work;
1392 
1393         struct mutex            s_sync_lock;    /* sync serialisation lock */
1394 
1395         /*
1396          * Indicates how deep in a filesystem stack this SB is
1397          */
1398         int s_stack_depth;
1399 
1400         /* s_inode_list_lock protects s_inodes */
1401         spinlock_t              s_inode_list_lock ____cacheline_aligned_in_smp;
1402         struct list_head        s_inodes;       /* all inodes */
1403 };
1404 
1405 extern struct timespec current_fs_time(struct super_block *sb);
1406 
1407 /*
1408  * Snapshotting support.
1409  */
1410 
1411 void __sb_end_write(struct super_block *sb, int level);
1412 int __sb_start_write(struct super_block *sb, int level, bool wait);
1413 
1414 #define __sb_writers_acquired(sb, lev)  \
1415         percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1416 #define __sb_writers_release(sb, lev)   \
1417         percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1418 
1419 /**
1420  * sb_end_write - drop write access to a superblock
1421  * @sb: the super we wrote to
1422  *
1423  * Decrement number of writers to the filesystem. Wake up possible waiters
1424  * wanting to freeze the filesystem.
1425  */
1426 static inline void sb_end_write(struct super_block *sb)
1427 {
1428         __sb_end_write(sb, SB_FREEZE_WRITE);
1429 }
1430 
1431 /**
1432  * sb_end_pagefault - drop write access to a superblock from a page fault
1433  * @sb: the super we wrote to
1434  *
1435  * Decrement number of processes handling write page fault to the filesystem.
1436  * Wake up possible waiters wanting to freeze the filesystem.
1437  */
1438 static inline void sb_end_pagefault(struct super_block *sb)
1439 {
1440         __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1441 }
1442 
1443 /**
1444  * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1445  * @sb: the super we wrote to
1446  *
1447  * Decrement fs-internal number of writers to the filesystem.  Wake up possible
1448  * waiters wanting to freeze the filesystem.
1449  */
1450 static inline void sb_end_intwrite(struct super_block *sb)
1451 {
1452         __sb_end_write(sb, SB_FREEZE_FS);
1453 }
1454 
1455 /**
1456  * sb_start_write - get write access to a superblock
1457  * @sb: the super we write to
1458  *
1459  * When a process wants to write data or metadata to a file system (i.e. dirty
1460  * a page or an inode), it should embed the operation in a sb_start_write() -
1461  * sb_end_write() pair to get exclusion against file system freezing. This
1462  * function increments number of writers preventing freezing. If the file
1463  * system is already frozen, the function waits until the file system is
1464  * thawed.
1465  *
1466  * Since freeze protection behaves as a lock, users have to preserve
1467  * ordering of freeze protection and other filesystem locks. Generally,
1468  * freeze protection should be the outermost lock. In particular, we have:
1469  *
1470  * sb_start_write
1471  *   -> i_mutex                 (write path, truncate, directory ops, ...)
1472  *   -> s_umount                (freeze_super, thaw_super)
1473  */
1474 static inline void sb_start_write(struct super_block *sb)
1475 {
1476         __sb_start_write(sb, SB_FREEZE_WRITE, true);
1477 }
1478 
1479 static inline int sb_start_write_trylock(struct super_block *sb)
1480 {
1481         return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1482 }
1483 
1484 /**
1485  * sb_start_pagefault - get write access to a superblock from a page fault
1486  * @sb: the super we write to
1487  *
1488  * When a process starts handling write page fault, it should embed the
1489  * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1490  * exclusion against file system freezing. This is needed since the page fault
1491  * is going to dirty a page. This function increments number of running page
1492  * faults preventing freezing. If the file system is already frozen, the
1493  * function waits until the file system is thawed.
1494  *
1495  * Since page fault freeze protection behaves as a lock, users have to preserve
1496  * ordering of freeze protection and other filesystem locks. It is advised to
1497  * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1498  * handling code implies lock dependency:
1499  *
1500  * mmap_sem
1501  *   -> sb_start_pagefault
1502  */
1503 static inline void sb_start_pagefault(struct super_block *sb)
1504 {
1505         __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1506 }
1507 
1508 /*
1509  * sb_start_intwrite - get write access to a superblock for internal fs purposes
1510  * @sb: the super we write to
1511  *
1512  * This is the third level of protection against filesystem freezing. It is
1513  * free for use by a filesystem. The only requirement is that it must rank
1514  * below sb_start_pagefault.
1515  *
1516  * For example filesystem can call sb_start_intwrite() when starting a
1517  * transaction which somewhat eases handling of freezing for internal sources
1518  * of filesystem changes (internal fs threads, discarding preallocation on file
1519  * close, etc.).
1520  */
1521 static inline void sb_start_intwrite(struct super_block *sb)
1522 {
1523         __sb_start_write(sb, SB_FREEZE_FS, true);
1524 }
1525 
1526 
1527 extern bool inode_owner_or_capable(const struct inode *inode);
1528 
1529 /*
1530  * VFS helper functions..
1531  */
1532 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1533 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1534 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1535 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1536 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1537 extern int vfs_rmdir(struct inode *, struct dentry *);
1538 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1539 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1540 extern int vfs_whiteout(struct inode *, struct dentry *);
1541 
1542 /*
1543  * VFS dentry helper functions.
1544  */
1545 extern void dentry_unhash(struct dentry *dentry);
1546 
1547 /*
1548  * VFS file helper functions.
1549  */
1550 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1551                         umode_t mode);
1552 /*
1553  * VFS FS_IOC_FIEMAP helper definitions.
1554  */
1555 struct fiemap_extent_info {
1556         unsigned int fi_flags;          /* Flags as passed from user */
1557         unsigned int fi_extents_mapped; /* Number of mapped extents */
1558         unsigned int fi_extents_max;    /* Size of fiemap_extent array */
1559         struct fiemap_extent __user *fi_extents_start; /* Start of
1560                                                         fiemap_extent array */
1561 };
1562 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1563                             u64 phys, u64 len, u32 flags);
1564 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1565 
1566 /*
1567  * File types
1568  *
1569  * NOTE! These match bits 12..15 of stat.st_mode
1570  * (ie "(i_mode >> 12) & 15").
1571  */
1572 #define DT_UNKNOWN      0
1573 #define DT_FIFO         1
1574 #define DT_CHR          2
1575 #define DT_DIR          4
1576 #define DT_BLK          6
1577 #define DT_REG          8
1578 #define DT_LNK          10
1579 #define DT_SOCK         12
1580 #define DT_WHT          14
1581 
1582 /*
1583  * This is the "filldir" function type, used by readdir() to let
1584  * the kernel specify what kind of dirent layout it wants to have.
1585  * This allows the kernel to read directories into kernel space or
1586  * to have different dirent layouts depending on the binary type.
1587  */
1588 struct dir_context;
1589 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1590                          unsigned);
1591 
1592 struct dir_context {
1593         const filldir_t actor;
1594         loff_t pos;
1595 };
1596 
1597 struct block_device_operations;
1598 
1599 /* These macros are for out of kernel modules to test that
1600  * the kernel supports the unlocked_ioctl and compat_ioctl
1601  * fields in struct file_operations. */
1602 #define HAVE_COMPAT_IOCTL 1
1603 #define HAVE_UNLOCKED_IOCTL 1
1604 
1605 /*
1606  * These flags let !MMU mmap() govern direct device mapping vs immediate
1607  * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1608  *
1609  * NOMMU_MAP_COPY:      Copy can be mapped (MAP_PRIVATE)
1610  * NOMMU_MAP_DIRECT:    Can be mapped directly (MAP_SHARED)
1611  * NOMMU_MAP_READ:      Can be mapped for reading
1612  * NOMMU_MAP_WRITE:     Can be mapped for writing
1613  * NOMMU_MAP_EXEC:      Can be mapped for execution
1614  */
1615 #define NOMMU_MAP_COPY          0x00000001
1616 #define NOMMU_MAP_DIRECT        0x00000008
1617 #define NOMMU_MAP_READ          VM_MAYREAD
1618 #define NOMMU_MAP_WRITE         VM_MAYWRITE
1619 #define NOMMU_MAP_EXEC          VM_MAYEXEC
1620 
1621 #define NOMMU_VMFLAGS \
1622         (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1623 
1624 
1625 struct iov_iter;
1626 
1627 struct file_operations {
1628         struct module *owner;
1629         loff_t (*llseek) (struct file *, loff_t, int);
1630         ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1631         ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1632         ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1633         ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1634         int (*iterate) (struct file *, struct dir_context *);
1635         unsigned int (*poll) (struct file *, struct poll_table_struct *);
1636         long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1637         long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1638         int (*mmap) (struct file *, struct vm_area_struct *);
1639         int (*open) (struct inode *, struct file *);
1640         int (*flush) (struct file *, fl_owner_t id);
1641         int (*release) (struct inode *, struct file *);
1642         int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1643         int (*aio_fsync) (struct kiocb *, int datasync);
1644         int (*fasync) (int, struct file *, int);
1645         int (*lock) (struct file *, int, struct file_lock *);
1646         ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1647         unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1648         int (*check_flags)(int);
1649         int (*flock) (struct file *, int, struct file_lock *);
1650         ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1651         ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1652         int (*setlease)(struct file *, long, struct file_lock **, void **);
1653         long (*fallocate)(struct file *file, int mode, loff_t offset,
1654                           loff_t len);
1655         void (*show_fdinfo)(struct seq_file *m, struct file *f);
1656 #ifndef CONFIG_MMU
1657         unsigned (*mmap_capabilities)(struct file *);
1658 #endif
1659         ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1660                         loff_t, size_t, unsigned int);
1661         int (*clone_file_range)(struct file *, loff_t, struct file *, loff_t,
1662                         u64);
1663         ssize_t (*dedupe_file_range)(struct file *, u64, u64, struct file *,
1664                         u64);
1665 };
1666 
1667 struct inode_operations {
1668         struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1669         const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
1670         int (*permission) (struct inode *, int);
1671         struct posix_acl * (*get_acl)(struct inode *, int);
1672 
1673         int (*readlink) (struct dentry *, char __user *,int);
1674 
1675         int (*create) (struct inode *,struct dentry *, umode_t, bool);
1676         int (*link) (struct dentry *,struct inode *,struct dentry *);
1677         int (*unlink) (struct inode *,struct dentry *);
1678         int (*symlink) (struct inode *,struct dentry *,const char *);
1679         int (*mkdir) (struct inode *,struct dentry *,umode_t);
1680         int (*rmdir) (struct inode *,struct dentry *);
1681         int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1682         int (*rename) (struct inode *, struct dentry *,
1683                         struct inode *, struct dentry *);
1684         int (*rename2) (struct inode *, struct dentry *,
1685                         struct inode *, struct dentry *, unsigned int);
1686         int (*setattr) (struct dentry *, struct iattr *);
1687         int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1688         int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1689         ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1690         ssize_t (*listxattr) (struct dentry *, char *, size_t);
1691         int (*removexattr) (struct dentry *, const char *);
1692         int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1693                       u64 len);
1694         int (*update_time)(struct inode *, struct timespec *, int);
1695         int (*atomic_open)(struct inode *, struct dentry *,
1696                            struct file *, unsigned open_flag,
1697                            umode_t create_mode, int *opened);
1698         int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1699         int (*set_acl)(struct inode *, struct posix_acl *, int);
1700 } ____cacheline_aligned;
1701 
1702 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1703                               unsigned long nr_segs, unsigned long fast_segs,
1704                               struct iovec *fast_pointer,
1705                               struct iovec **ret_pointer);
1706 
1707 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1708 extern ssize_t __vfs_write(struct file *, const char __user *, size_t, loff_t *);
1709 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1710 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1711 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1712                 unsigned long, loff_t *);
1713 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1714                 unsigned long, loff_t *);
1715 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1716                                    loff_t, size_t, unsigned int);
1717 extern int vfs_clone_file_range(struct file *file_in, loff_t pos_in,
1718                 struct file *file_out, loff_t pos_out, u64 len);
1719 extern int vfs_dedupe_file_range(struct file *file,
1720                                  struct file_dedupe_range *same);
1721 
1722 struct super_operations {
1723         struct inode *(*alloc_inode)(struct super_block *sb);
1724         void (*destroy_inode)(struct inode *);
1725 
1726         void (*dirty_inode) (struct inode *, int flags);
1727         int (*write_inode) (struct inode *, struct writeback_control *wbc);
1728         int (*drop_inode) (struct inode *);
1729         void (*evict_inode) (struct inode *);
1730         void (*put_super) (struct super_block *);
1731         int (*sync_fs)(struct super_block *sb, int wait);
1732         int (*freeze_super) (struct super_block *);
1733         int (*freeze_fs) (struct super_block *);
1734         int (*thaw_super) (struct super_block *);
1735         int (*unfreeze_fs) (struct super_block *);
1736         int (*statfs) (struct dentry *, struct kstatfs *);
1737         int (*remount_fs) (struct super_block *, int *, char *);
1738         void (*umount_begin) (struct super_block *);
1739 
1740         int (*show_options)(struct seq_file *, struct dentry *);
1741         int (*show_devname)(struct seq_file *, struct dentry *);
1742         int (*show_path)(struct seq_file *, struct dentry *);
1743         int (*show_stats)(struct seq_file *, struct dentry *);
1744 #ifdef CONFIG_QUOTA
1745         ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1746         ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1747         struct dquot **(*get_dquots)(struct inode *);
1748 #endif
1749         int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1750         long (*nr_cached_objects)(struct super_block *,
1751                                   struct shrink_control *);
1752         long (*free_cached_objects)(struct super_block *,
1753                                     struct shrink_control *);
1754 };
1755 
1756 /*
1757  * Inode flags - they have no relation to superblock flags now
1758  */
1759 #define S_SYNC          1       /* Writes are synced at once */
1760 #define S_NOATIME       2       /* Do not update access times */
1761 #define S_APPEND        4       /* Append-only file */
1762 #define S_IMMUTABLE     8       /* Immutable file */
1763 #define S_DEAD          16      /* removed, but still open directory */
1764 #define S_NOQUOTA       32      /* Inode is not counted to quota */
1765 #define S_DIRSYNC       64      /* Directory modifications are synchronous */
1766 #define S_NOCMTIME      128     /* Do not update file c/mtime */
1767 #define S_SWAPFILE      256     /* Do not truncate: swapon got its bmaps */
1768 #define S_PRIVATE       512     /* Inode is fs-internal */
1769 #define S_IMA           1024    /* Inode has an associated IMA struct */
1770 #define S_AUTOMOUNT     2048    /* Automount/referral quasi-directory */
1771 #define S_NOSEC         4096    /* no suid or xattr security attributes */
1772 #ifdef CONFIG_FS_DAX
1773 #define S_DAX           8192    /* Direct Access, avoiding the page cache */
1774 #else
1775 #define S_DAX           0       /* Make all the DAX code disappear */
1776 #endif
1777 
1778 /*
1779  * Note that nosuid etc flags are inode-specific: setting some file-system
1780  * flags just means all the inodes inherit those flags by default. It might be
1781  * possible to override it selectively if you really wanted to with some
1782  * ioctl() that is not currently implemented.
1783  *
1784  * Exception: MS_RDONLY is always applied to the entire file system.
1785  *
1786  * Unfortunately, it is possible to change a filesystems flags with it mounted
1787  * with files in use.  This means that all of the inodes will not have their
1788  * i_flags updated.  Hence, i_flags no longer inherit the superblock mount
1789  * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1790  */
1791 #define __IS_FLG(inode, flg)    ((inode)->i_sb->s_flags & (flg))
1792 
1793 #define IS_RDONLY(inode)        ((inode)->i_sb->s_flags & MS_RDONLY)
1794 #define IS_SYNC(inode)          (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1795                                         ((inode)->i_flags & S_SYNC))
1796 #define IS_DIRSYNC(inode)       (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1797                                         ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1798 #define IS_MANDLOCK(inode)      __IS_FLG(inode, MS_MANDLOCK)
1799 #define IS_NOATIME(inode)       __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1800 #define IS_I_VERSION(inode)     __IS_FLG(inode, MS_I_VERSION)
1801 
1802 #define IS_NOQUOTA(inode)       ((inode)->i_flags & S_NOQUOTA)
1803 #define IS_APPEND(inode)        ((inode)->i_flags & S_APPEND)
1804 #define IS_IMMUTABLE(inode)     ((inode)->i_flags & S_IMMUTABLE)
1805 #define IS_POSIXACL(inode)      __IS_FLG(inode, MS_POSIXACL)
1806 
1807 #define IS_DEADDIR(inode)       ((inode)->i_flags & S_DEAD)
1808 #define IS_NOCMTIME(inode)      ((inode)->i_flags & S_NOCMTIME)
1809 #define IS_SWAPFILE(inode)      ((inode)->i_flags & S_SWAPFILE)
1810 #define IS_PRIVATE(inode)       ((inode)->i_flags & S_PRIVATE)
1811 #define IS_IMA(inode)           ((inode)->i_flags & S_IMA)
1812 #define IS_AUTOMOUNT(inode)     ((inode)->i_flags & S_AUTOMOUNT)
1813 #define IS_NOSEC(inode)         ((inode)->i_flags & S_NOSEC)
1814 #define IS_DAX(inode)           ((inode)->i_flags & S_DAX)
1815 
1816 #define IS_WHITEOUT(inode)      (S_ISCHR(inode->i_mode) && \
1817                                  (inode)->i_rdev == WHITEOUT_DEV)
1818 
1819 /*
1820  * Inode state bits.  Protected by inode->i_lock
1821  *
1822  * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1823  * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1824  *
1825  * Four bits define the lifetime of an inode.  Initially, inodes are I_NEW,
1826  * until that flag is cleared.  I_WILL_FREE, I_FREEING and I_CLEAR are set at
1827  * various stages of removing an inode.
1828  *
1829  * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1830  *
1831  * I_DIRTY_SYNC         Inode is dirty, but doesn't have to be written on
1832  *                      fdatasync().  i_atime is the usual cause.
1833  * I_DIRTY_DATASYNC     Data-related inode changes pending. We keep track of
1834  *                      these changes separately from I_DIRTY_SYNC so that we
1835  *                      don't have to write inode on fdatasync() when only
1836  *                      mtime has changed in it.
1837  * I_DIRTY_PAGES        Inode has dirty pages.  Inode itself may be clean.
1838  * I_NEW                Serves as both a mutex and completion notification.
1839  *                      New inodes set I_NEW.  If two processes both create
1840  *                      the same inode, one of them will release its inode and
1841  *                      wait for I_NEW to be released before returning.
1842  *                      Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1843  *                      also cause waiting on I_NEW, without I_NEW actually
1844  *                      being set.  find_inode() uses this to prevent returning
1845  *                      nearly-dead inodes.
1846  * I_WILL_FREE          Must be set when calling write_inode_now() if i_count
1847  *                      is zero.  I_FREEING must be set when I_WILL_FREE is
1848  *                      cleared.
1849  * I_FREEING            Set when inode is about to be freed but still has dirty
1850  *                      pages or buffers attached or the inode itself is still
1851  *                      dirty.
1852  * I_CLEAR              Added by clear_inode().  In this state the inode is
1853  *                      clean and can be destroyed.  Inode keeps I_FREEING.
1854  *
1855  *                      Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1856  *                      prohibited for many purposes.  iget() must wait for
1857  *                      the inode to be completely released, then create it
1858  *                      anew.  Other functions will just ignore such inodes,
1859  *                      if appropriate.  I_NEW is used for waiting.
1860  *
1861  * I_SYNC               Writeback of inode is running. The bit is set during
1862  *                      data writeback, and cleared with a wakeup on the bit
1863  *                      address once it is done. The bit is also used to pin
1864  *                      the inode in memory for flusher thread.
1865  *
1866  * I_REFERENCED         Marks the inode as recently references on the LRU list.
1867  *
1868  * I_DIO_WAKEUP         Never set.  Only used as a key for wait_on_bit().
1869  *
1870  * I_WB_SWITCH          Cgroup bdi_writeback switching in progress.  Used to
1871  *                      synchronize competing switching instances and to tell
1872  *                      wb stat updates to grab mapping->tree_lock.  See
1873  *                      inode_switch_wb_work_fn() for details.
1874  *
1875  * Q: What is the difference between I_WILL_FREE and I_FREEING?
1876  */
1877 #define I_DIRTY_SYNC            (1 << 0)
1878 #define I_DIRTY_DATASYNC        (1 << 1)
1879 #define I_DIRTY_PAGES           (1 << 2)
1880 #define __I_NEW                 3
1881 #define I_NEW                   (1 << __I_NEW)
1882 #define I_WILL_FREE             (1 << 4)
1883 #define I_FREEING               (1 << 5)
1884 #define I_CLEAR                 (1 << 6)
1885 #define __I_SYNC                7
1886 #define I_SYNC                  (1 << __I_SYNC)
1887 #define I_REFERENCED            (1 << 8)
1888 #define __I_DIO_WAKEUP          9
1889 #define I_DIO_WAKEUP            (1 << __I_DIO_WAKEUP)
1890 #define I_LINKABLE              (1 << 10)
1891 #define I_DIRTY_TIME            (1 << 11)
1892 #define __I_DIRTY_TIME_EXPIRED  12
1893 #define I_DIRTY_TIME_EXPIRED    (1 << __I_DIRTY_TIME_EXPIRED)
1894 #define I_WB_SWITCH             (1 << 13)
1895 
1896 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1897 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1898 
1899 extern void __mark_inode_dirty(struct inode *, int);
1900 static inline void mark_inode_dirty(struct inode *inode)
1901 {
1902         __mark_inode_dirty(inode, I_DIRTY);
1903 }
1904 
1905 static inline void mark_inode_dirty_sync(struct inode *inode)
1906 {
1907         __mark_inode_dirty(inode, I_DIRTY_SYNC);
1908 }
1909 
1910 extern void inc_nlink(struct inode *inode);
1911 extern void drop_nlink(struct inode *inode);
1912 extern void clear_nlink(struct inode *inode);
1913 extern void set_nlink(struct inode *inode, unsigned int nlink);
1914 
1915 static inline void inode_inc_link_count(struct inode *inode)
1916 {
1917         inc_nlink(inode);
1918         mark_inode_dirty(inode);
1919 }
1920 
1921 static inline void inode_dec_link_count(struct inode *inode)
1922 {
1923         drop_nlink(inode);
1924         mark_inode_dirty(inode);
1925 }
1926 
1927 /**
1928  * inode_inc_iversion - increments i_version
1929  * @inode: inode that need to be updated
1930  *
1931  * Every time the inode is modified, the i_version field will be incremented.
1932  * The filesystem has to be mounted with i_version flag
1933  */
1934 
1935 static inline void inode_inc_iversion(struct inode *inode)
1936 {
1937        spin_lock(&inode->i_lock);
1938        inode->i_version++;
1939        spin_unlock(&inode->i_lock);
1940 }
1941 
1942 enum file_time_flags {
1943         S_ATIME = 1,
1944         S_MTIME = 2,
1945         S_CTIME = 4,
1946         S_VERSION = 8,
1947 };
1948 
1949 extern bool atime_needs_update(const struct path *, struct inode *);
1950 extern void touch_atime(const struct path *);
1951 static inline void file_accessed(struct file *file)
1952 {
1953         if (!(file->f_flags & O_NOATIME))
1954                 touch_atime(&file->f_path);
1955 }
1956 
1957 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1958 int sync_inode_metadata(struct inode *inode, int wait);
1959 
1960 struct file_system_type {
1961         const char *name;
1962         int fs_flags;
1963 #define FS_REQUIRES_DEV         1 
1964 #define FS_BINARY_MOUNTDATA     2
1965 #define FS_HAS_SUBTYPE          4
1966 #define FS_USERNS_MOUNT         8       /* Can be mounted by userns root */
1967 #define FS_USERNS_DEV_MOUNT     16 /* A userns mount does not imply MNT_NODEV */
1968 #define FS_USERNS_VISIBLE       32      /* FS must already be visible */
1969 #define FS_RENAME_DOES_D_MOVE   32768   /* FS will handle d_move() during rename() internally. */
1970         struct dentry *(*mount) (struct file_system_type *, int,
1971                        const char *, void *);
1972         void (*kill_sb) (struct super_block *);
1973         struct module *owner;
1974         struct file_system_type * next;
1975         struct hlist_head fs_supers;
1976 
1977         struct lock_class_key s_lock_key;
1978         struct lock_class_key s_umount_key;
1979         struct lock_class_key s_vfs_rename_key;
1980         struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
1981 
1982         struct lock_class_key i_lock_key;
1983         struct lock_class_key i_mutex_key;
1984         struct lock_class_key i_mutex_dir_key;
1985 };
1986 
1987 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
1988 
1989 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1990         void *data, int (*fill_super)(struct super_block *, void *, int));
1991 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1992         int flags, const char *dev_name, void *data,
1993         int (*fill_super)(struct super_block *, void *, int));
1994 extern struct dentry *mount_single(struct file_system_type *fs_type,
1995         int flags, void *data,
1996         int (*fill_super)(struct super_block *, void *, int));
1997 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1998         int flags, void *data,
1999         int (*fill_super)(struct super_block *, void *, int));
2000 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2001 void generic_shutdown_super(struct super_block *sb);
2002 void kill_block_super(struct super_block *sb);
2003 void kill_anon_super(struct super_block *sb);
2004 void kill_litter_super(struct super_block *sb);
2005 void deactivate_super(struct super_block *sb);
2006 void deactivate_locked_super(struct super_block *sb);
2007 int set_anon_super(struct super_block *s, void *data);
2008 int get_anon_bdev(dev_t *);
2009 void free_anon_bdev(dev_t);
2010 struct super_block *sget(struct file_system_type *type,
2011                         int (*test)(struct super_block *,void *),
2012                         int (*set)(struct super_block *,void *),
2013                         int flags, void *data);
2014 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
2015         const struct super_operations *ops,
2016         const struct dentry_operations *dops,
2017         unsigned long);
2018 
2019 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2020 #define fops_get(fops) \
2021         (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2022 #define fops_put(fops) \
2023         do { if (fops) module_put((fops)->owner); } while(0)
2024 /*
2025  * This one is to be used *ONLY* from ->open() instances.
2026  * fops must be non-NULL, pinned down *and* module dependencies
2027  * should be sufficient to pin the caller down as well.
2028  */
2029 #define replace_fops(f, fops) \
2030         do {    \
2031                 struct file *__file = (f); \
2032                 fops_put(__file->f_op); \
2033                 BUG_ON(!(__file->f_op = (fops))); \
2034         } while(0)
2035 
2036 extern int register_filesystem(struct file_system_type *);
2037 extern int unregister_filesystem(struct file_system_type *);
2038 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
2039 #define kern_mount(type) kern_mount_data(type, NULL)
2040 extern void kern_unmount(struct vfsmount *mnt);
2041 extern int may_umount_tree(struct vfsmount *);
2042 extern int may_umount(struct vfsmount *);
2043 extern long do_mount(const char *, const char __user *,
2044                      const char *, unsigned long, void *);
2045 extern struct vfsmount *collect_mounts(struct path *);
2046 extern void drop_collected_mounts(struct vfsmount *);
2047 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2048                           struct vfsmount *);
2049 extern int vfs_statfs(struct path *, struct kstatfs *);
2050 extern int user_statfs(const char __user *, struct kstatfs *);
2051 extern int fd_statfs(int, struct kstatfs *);
2052 extern int vfs_ustat(dev_t, struct kstatfs *);
2053 extern int freeze_super(struct super_block *super);
2054 extern int thaw_super(struct super_block *super);
2055 extern bool our_mnt(struct vfsmount *mnt);
2056 
2057 extern int current_umask(void);
2058 
2059 extern void ihold(struct inode * inode);
2060 extern void iput(struct inode *);
2061 extern int generic_update_time(struct inode *, struct timespec *, int);
2062 
2063 /* /sys/fs */
2064 extern struct kobject *fs_kobj;
2065 
2066 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
2067 
2068 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2069 extern int locks_mandatory_locked(struct file *);
2070 extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
2071 
2072 /*
2073  * Candidates for mandatory locking have the setgid bit set
2074  * but no group execute bit -  an otherwise meaningless combination.
2075  */
2076 
2077 static inline int __mandatory_lock(struct inode *ino)
2078 {
2079         return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2080 }
2081 
2082 /*
2083  * ... and these candidates should be on MS_MANDLOCK mounted fs,
2084  * otherwise these will be advisory locks
2085  */
2086 
2087 static inline int mandatory_lock(struct inode *ino)
2088 {
2089         return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2090 }
2091 
2092 static inline int locks_verify_locked(struct file *file)
2093 {
2094         if (mandatory_lock(file_inode(file)))
2095                 return locks_mandatory_locked(file);
2096         return 0;
2097 }
2098 
2099 static inline int locks_verify_truncate(struct inode *inode,
2100                                     struct file *f,
2101                                     loff_t size)
2102 {
2103         if (!inode->i_flctx || !mandatory_lock(inode))
2104                 return 0;
2105 
2106         if (size < inode->i_size) {
2107                 return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2108                                 F_WRLCK);
2109         } else {
2110                 return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2111                                 F_WRLCK);
2112         }
2113 }
2114 
2115 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2116 
2117 static inline int locks_mandatory_locked(struct file *file)
2118 {
2119         return 0;
2120 }
2121 
2122 static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2123                                        loff_t start, loff_t end, unsigned char type)
2124 {
2125         return 0;
2126 }
2127 
2128 static inline int __mandatory_lock(struct inode *inode)
2129 {
2130         return 0;
2131 }
2132 
2133 static inline int mandatory_lock(struct inode *inode)
2134 {
2135         return 0;
2136 }
2137 
2138 static inline int locks_verify_locked(struct file *file)
2139 {
2140         return 0;
2141 }
2142 
2143 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2144                                         size_t size)
2145 {
2146         return 0;
2147 }
2148 
2149 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2150 
2151 
2152 #ifdef CONFIG_FILE_LOCKING
2153 static inline int break_lease(struct inode *inode, unsigned int mode)
2154 {
2155         /*
2156          * Since this check is lockless, we must ensure that any refcounts
2157          * taken are done before checking i_flctx->flc_lease. Otherwise, we
2158          * could end up racing with tasks trying to set a new lease on this
2159          * file.
2160          */
2161         smp_mb();
2162         if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2163                 return __break_lease(inode, mode, FL_LEASE);
2164         return 0;
2165 }
2166 
2167 static inline int break_deleg(struct inode *inode, unsigned int mode)
2168 {
2169         /*
2170          * Since this check is lockless, we must ensure that any refcounts
2171          * taken are done before checking i_flctx->flc_lease. Otherwise, we
2172          * could end up racing with tasks trying to set a new lease on this
2173          * file.
2174          */
2175         smp_mb();
2176         if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2177                 return __break_lease(inode, mode, FL_DELEG);
2178         return 0;
2179 }
2180 
2181 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2182 {
2183         int ret;
2184 
2185         ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2186         if (ret == -EWOULDBLOCK && delegated_inode) {
2187                 *delegated_inode = inode;
2188                 ihold(inode);
2189         }
2190         return ret;
2191 }
2192 
2193 static inline int break_deleg_wait(struct inode **delegated_inode)
2194 {
2195         int ret;
2196 
2197         ret = break_deleg(*delegated_inode, O_WRONLY);
2198         iput(*delegated_inode);
2199         *delegated_inode = NULL;
2200         return ret;
2201 }
2202 
2203 static inline int break_layout(struct inode *inode, bool wait)
2204 {
2205         smp_mb();
2206         if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2207                 return __break_lease(inode,
2208                                 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2209                                 FL_LAYOUT);
2210         return 0;
2211 }
2212 
2213 #else /* !CONFIG_FILE_LOCKING */
2214 static inline int break_lease(struct inode *inode, unsigned int mode)
2215 {
2216         return 0;
2217 }
2218 
2219 static inline int break_deleg(struct inode *inode, unsigned int mode)
2220 {
2221         return 0;
2222 }
2223 
2224 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2225 {
2226         return 0;
2227 }
2228 
2229 static inline int break_deleg_wait(struct inode **delegated_inode)
2230 {
2231         BUG();
2232         return 0;
2233 }
2234 
2235 static inline int break_layout(struct inode *inode, bool wait)
2236 {
2237         return 0;
2238 }
2239 
2240 #endif /* CONFIG_FILE_LOCKING */
2241 
2242 /* fs/open.c */
2243 struct audit_names;
2244 struct filename {
2245         const char              *name;  /* pointer to actual string */
2246         const __user char       *uptr;  /* original userland pointer */
2247         struct audit_names      *aname;
2248         int                     refcnt;
2249         const char              iname[];
2250 };
2251 
2252 extern long vfs_truncate(struct path *, loff_t);
2253 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2254                        struct file *filp);
2255 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2256                         loff_t len);
2257 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2258                         umode_t mode);
2259 extern struct file *file_open_name(struct filename *, int, umode_t);
2260 extern struct file *filp_open(const char *, int, umode_t);
2261 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2262                                    const char *, int);
2263 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2264 extern int filp_close(struct file *, fl_owner_t id);
2265 
2266 extern struct filename *getname_flags(const char __user *, int, int *);
2267 extern struct filename *getname(const char __user *);
2268 extern struct filename *getname_kernel(const char *);
2269 extern void putname(struct filename *name);
2270 
2271 enum {
2272         FILE_CREATED = 1,
2273         FILE_OPENED = 2
2274 };
2275 extern int finish_open(struct file *file, struct dentry *dentry,
2276                         int (*open)(struct inode *, struct file *),
2277                         int *opened);
2278 extern int finish_no_open(struct file *file, struct dentry *dentry);
2279 
2280 /* fs/ioctl.c */
2281 
2282 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2283 
2284 /* fs/dcache.c */
2285 extern void __init vfs_caches_init_early(void);
2286 extern void __init vfs_caches_init(void);
2287 
2288 extern struct kmem_cache *names_cachep;
2289 
2290 #define __getname()             kmem_cache_alloc(names_cachep, GFP_KERNEL)
2291 #define __putname(name)         kmem_cache_free(names_cachep, (void *)(name))
2292 
2293 #ifdef CONFIG_BLOCK
2294 extern int register_blkdev(unsigned int, const char *);
2295 extern void unregister_blkdev(unsigned int, const char *);
2296 extern struct block_device *bdget(dev_t);
2297 extern struct block_device *bdgrab(struct block_device *bdev);
2298 extern void bd_set_size(struct block_device *, loff_t size);
2299 extern void bd_forget(struct inode *inode);
2300 extern void bdput(struct block_device *);
2301 extern void invalidate_bdev(struct block_device *);
2302 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2303 extern int sync_blockdev(struct block_device *bdev);
2304 extern void kill_bdev(struct block_device *);
2305 extern struct super_block *freeze_bdev(struct block_device *);
2306 extern void emergency_thaw_all(void);
2307 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2308 extern int fsync_bdev(struct block_device *);
2309 #ifdef CONFIG_FS_DAX
2310 extern bool blkdev_dax_capable(struct block_device *bdev);
2311 #else
2312 static inline bool blkdev_dax_capable(struct block_device *bdev)
2313 {
2314         return false;
2315 }
2316 #endif
2317 
2318 extern struct super_block *blockdev_superblock;
2319 
2320 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2321 {
2322         return sb == blockdev_superblock;
2323 }
2324 #else
2325 static inline void bd_forget(struct inode *inode) {}
2326 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2327 static inline void kill_bdev(struct block_device *bdev) {}
2328 static inline void invalidate_bdev(struct block_device *bdev) {}
2329 
2330 static inline struct super_block *freeze_bdev(struct block_device *sb)
2331 {
2332         return NULL;
2333 }
2334 
2335 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2336 {
2337         return 0;
2338 }
2339 
2340 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2341 {
2342 }
2343 
2344 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2345 {
2346         return false;
2347 }
2348 #endif
2349 extern int sync_filesystem(struct super_block *);
2350 extern const struct file_operations def_blk_fops;
2351 extern const struct file_operations def_chr_fops;
2352 #ifdef CONFIG_BLOCK
2353 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2354 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2355 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2356 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2357 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2358                                                void *holder);
2359 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2360                                               void *holder);
2361 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2362 extern int __blkdev_reread_part(struct block_device *bdev);
2363 extern int blkdev_reread_part(struct block_device *bdev);
2364 
2365 #ifdef CONFIG_SYSFS
2366 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2367 extern void bd_unlink_disk_holder(struct block_device *bdev,
2368                                   struct gendisk *disk);
2369 #else
2370 static inline int bd_link_disk_holder(struct block_device *bdev,
2371                                       struct gendisk *disk)
2372 {
2373         return 0;
2374 }
2375 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2376                                          struct gendisk *disk)
2377 {
2378 }
2379 #endif
2380 #endif
2381 
2382 /* fs/char_dev.c */
2383 #define CHRDEV_MAJOR_HASH_SIZE  255
2384 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2385 extern int register_chrdev_region(dev_t, unsigned, const char *);
2386 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2387                              unsigned int count, const char *name,
2388                              const struct file_operations *fops);
2389 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2390                                 unsigned int count, const char *name);
2391 extern void unregister_chrdev_region(dev_t, unsigned);
2392 extern void chrdev_show(struct seq_file *,off_t);
2393 
2394 static inline int register_chrdev(unsigned int major, const char *name,
2395                                   const struct file_operations *fops)
2396 {
2397         return __register_chrdev(major, 0, 256, name, fops);
2398 }
2399 
2400 static inline void unregister_chrdev(unsigned int major, const char *name)
2401 {
2402         __unregister_chrdev(major, 0, 256, name);
2403 }
2404 
2405 /* fs/block_dev.c */
2406 #define BDEVNAME_SIZE   32      /* Largest string for a blockdev identifier */
2407 #define BDEVT_SIZE      10      /* Largest string for MAJ:MIN for blkdev */
2408 
2409 #ifdef CONFIG_BLOCK
2410 #define BLKDEV_MAJOR_HASH_SIZE  255
2411 extern const char *__bdevname(dev_t, char *buffer);
2412 extern const char *bdevname(struct block_device *bdev, char *buffer);
2413 extern struct block_device *lookup_bdev(const char *);
2414 extern void blkdev_show(struct seq_file *,off_t);
2415 
2416 #else
2417 #define BLKDEV_MAJOR_HASH_SIZE  0
2418 #endif
2419 
2420 extern void init_special_inode(struct inode *, umode_t, dev_t);
2421 
2422 /* Invalid inode operations -- fs/bad_inode.c */
2423 extern void make_bad_inode(struct inode *);
2424 extern bool is_bad_inode(struct inode *);
2425 
2426 #ifdef CONFIG_BLOCK
2427 /*
2428  * return READ, READA, or WRITE
2429  */
2430 #define bio_rw(bio)             ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2431 
2432 /*
2433  * return data direction, READ or WRITE
2434  */
2435 #define bio_data_dir(bio)       ((bio)->bi_rw & 1)
2436 
2437 extern void check_disk_size_change(struct gendisk *disk,
2438                                    struct block_device *bdev);
2439 extern int revalidate_disk(struct gendisk *);
2440 extern int check_disk_change(struct block_device *);
2441 extern int __invalidate_device(struct block_device *, bool);
2442 extern int invalidate_partition(struct gendisk *, int);
2443 #endif
2444 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2445                                         pgoff_t start, pgoff_t end);
2446 
2447 static inline void invalidate_remote_inode(struct inode *inode)
2448 {
2449         if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2450             S_ISLNK(inode->i_mode))
2451                 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2452 }
2453 extern int invalidate_inode_pages2(struct address_space *mapping);
2454 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2455                                          pgoff_t start, pgoff_t end);
2456 extern int write_inode_now(struct inode *, int);
2457 extern int filemap_fdatawrite(struct address_space *);
2458 extern int filemap_flush(struct address_space *);
2459 extern int filemap_fdatawait(struct address_space *);
2460 extern void filemap_fdatawait_keep_errors(struct address_space *);
2461 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2462                                    loff_t lend);
2463 extern int filemap_write_and_wait(struct address_space *mapping);
2464 extern int filemap_write_and_wait_range(struct address_space *mapping,
2465                                         loff_t lstart, loff_t lend);
2466 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2467                                 loff_t start, loff_t end, int sync_mode);
2468 extern int filemap_fdatawrite_range(struct address_space *mapping,
2469                                 loff_t start, loff_t end);
2470 
2471 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2472                            int datasync);
2473 extern int vfs_fsync(struct file *file, int datasync);
2474 static inline int generic_write_sync(struct file *file, loff_t pos, loff_t count)
2475 {
2476         if (!(file->f_flags & O_DSYNC) && !IS_SYNC(file->f_mapping->host))
2477                 return 0;
2478         return vfs_fsync_range(file, pos, pos + count - 1,
2479                                (file->f_flags & __O_SYNC) ? 0 : 1);
2480 }
2481 extern void emergency_sync(void);
2482 extern void emergency_remount(void);
2483 #ifdef CONFIG_BLOCK
2484 extern sector_t bmap(struct inode *, sector_t);
2485 #endif
2486 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2487 extern int inode_permission(struct inode *, int);
2488 extern int __inode_permission(struct inode *, int);
2489 extern int generic_permission(struct inode *, int);
2490 extern int __check_sticky(struct inode *dir, struct inode *inode);
2491 
2492 static inline bool execute_ok(struct inode *inode)
2493 {
2494         return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2495 }
2496 
2497 static inline void file_start_write(struct file *file)
2498 {
2499         if (!S_ISREG(file_inode(file)->i_mode))
2500                 return;
2501         __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2502 }
2503 
2504 static inline bool file_start_write_trylock(struct file *file)
2505 {
2506         if (!S_ISREG(file_inode(file)->i_mode))
2507                 return true;
2508         return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2509 }
2510 
2511 static inline void file_end_write(struct file *file)
2512 {
2513         if (!S_ISREG(file_inode(file)->i_mode))
2514                 return;
2515         __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2516 }
2517 
2518 /*
2519  * get_write_access() gets write permission for a file.
2520  * put_write_access() releases this write permission.
2521  * This is used for regular files.
2522  * We cannot support write (and maybe mmap read-write shared) accesses and
2523  * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2524  * can have the following values:
2525  * 0: no writers, no VM_DENYWRITE mappings
2526  * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2527  * > 0: (i_writecount) users are writing to the file.
2528  *
2529  * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2530  * except for the cases where we don't hold i_writecount yet. Then we need to
2531  * use {get,deny}_write_access() - these functions check the sign and refuse
2532  * to do the change if sign is wrong.
2533  */
2534 static inline int get_write_access(struct inode *inode)
2535 {
2536         return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2537 }
2538 static inline int deny_write_access(struct file *file)
2539 {
2540         struct inode *inode = file_inode(file);
2541         return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2542 }
2543 static inline void put_write_access(struct inode * inode)
2544 {
2545         atomic_dec(&inode->i_writecount);
2546 }
2547 static inline void allow_write_access(struct file *file)
2548 {
2549         if (file)
2550                 atomic_inc(&file_inode(file)->i_writecount);
2551 }
2552 static inline bool inode_is_open_for_write(const struct inode *inode)
2553 {
2554         return atomic_read(&inode->i_writecount) > 0;
2555 }
2556 
2557 #ifdef CONFIG_IMA
2558 static inline void i_readcount_dec(struct inode *inode)
2559 {
2560         BUG_ON(!atomic_read(&inode->i_readcount));
2561         atomic_dec(&inode->i_readcount);
2562 }
2563 static inline void i_readcount_inc(struct inode *inode)
2564 {
2565         atomic_inc(&inode->i_readcount);
2566 }
2567 #else
2568 static inline void i_readcount_dec(struct inode *inode)
2569 {
2570         return;
2571 }
2572 static inline void i_readcount_inc(struct inode *inode)
2573 {
2574         return;
2575 }
2576 #endif
2577 extern int do_pipe_flags(int *, int);
2578 
2579 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2580 extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
2581 extern ssize_t __kernel_write(struct file *, const char *, size_t, loff_t *);
2582 extern struct file * open_exec(const char *);
2583  
2584 /* fs/dcache.c -- generic fs support functions */
2585 extern bool is_subdir(struct dentry *, struct dentry *);
2586 extern bool path_is_under(struct path *, struct path *);
2587 
2588 extern char *file_path(struct file *, char *, int);
2589 
2590 #include <linux/err.h>
2591 
2592 /* needed for stackable file system support */
2593 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2594 
2595 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2596 
2597 extern int inode_init_always(struct super_block *, struct inode *);
2598 extern void inode_init_once(struct inode *);
2599 extern void address_space_init_once(struct address_space *mapping);
2600 extern struct inode * igrab(struct inode *);
2601 extern ino_t iunique(struct super_block *, ino_t);
2602 extern int inode_needs_sync(struct inode *inode);
2603 extern int generic_delete_inode(struct inode *inode);
2604 static inline int generic_drop_inode(struct inode *inode)
2605 {
2606         return !inode->i_nlink || inode_unhashed(inode);
2607 }
2608 
2609 extern struct inode *ilookup5_nowait(struct super_block *sb,
2610                 unsigned long hashval, int (*test)(struct inode *, void *),
2611                 void *data);
2612 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2613                 int (*test)(struct inode *, void *), void *data);
2614 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2615 
2616 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2617 extern struct inode * iget_locked(struct super_block *, unsigned long);
2618 extern struct inode *find_inode_nowait(struct super_block *,
2619                                        unsigned long,
2620                                        int (*match)(struct inode *,
2621                                                     unsigned long, void *),
2622                                        void *data);
2623 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2624 extern int insert_inode_locked(struct inode *);
2625 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2626 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2627 #else
2628 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2629 #endif
2630 extern void unlock_new_inode(struct inode *);
2631 extern unsigned int get_next_ino(void);
2632 
2633 extern void __iget(struct inode * inode);
2634 extern void iget_failed(struct inode *);
2635 extern void clear_inode(struct inode *);
2636 extern void __destroy_inode(struct inode *);
2637 extern struct inode *new_inode_pseudo(struct super_block *sb);
2638 extern struct inode *new_inode(struct super_block *sb);
2639 extern void free_inode_nonrcu(struct inode *inode);
2640 extern int should_remove_suid(struct dentry *);
2641 extern int file_remove_privs(struct file *);
2642 extern int dentry_needs_remove_privs(struct dentry *dentry);
2643 static inline int file_needs_remove_privs(struct file *file)
2644 {
2645         return dentry_needs_remove_privs(file->f_path.dentry);
2646 }
2647 
2648 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2649 static inline void insert_inode_hash(struct inode *inode)
2650 {
2651         __insert_inode_hash(inode, inode->i_ino);
2652 }
2653 
2654 extern void __remove_inode_hash(struct inode *);
2655 static inline void remove_inode_hash(struct inode *inode)
2656 {
2657         if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
2658                 __remove_inode_hash(inode);
2659 }
2660 
2661 extern void inode_sb_list_add(struct inode *inode);
2662 
2663 #ifdef CONFIG_BLOCK
2664 extern blk_qc_t submit_bio(int, struct bio *);
2665 extern int bdev_read_only(struct block_device *);
2666 #endif
2667 extern int set_blocksize(struct block_device *, int);
2668 extern int sb_set_blocksize(struct super_block *, int);
2669 extern int sb_min_blocksize(struct super_block *, int);
2670 
2671 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2672 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2673 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2674 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2675 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2676 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2677 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *, loff_t);
2678 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2679 
2680 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos);
2681 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos);
2682 
2683 /* fs/block_dev.c */
2684 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
2685 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2686 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2687                         int datasync);
2688 extern void block_sync_page(struct page *page);
2689 
2690 /* fs/splice.c */
2691 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2692                 struct pipe_inode_info *, size_t, unsigned int);
2693 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2694                 struct pipe_inode_info *, size_t, unsigned int);
2695 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2696                 struct file *, loff_t *, size_t, unsigned int);
2697 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2698                 struct file *out, loff_t *, size_t len, unsigned int flags);
2699 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2700                 loff_t *opos, size_t len, unsigned int flags);
2701 
2702 
2703 extern void
2704 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2705 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2706 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2707 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
2708 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2709 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2710                 int whence, loff_t maxsize, loff_t eof);
2711 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2712                 int whence, loff_t size);
2713 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
2714 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
2715 extern int generic_file_open(struct inode * inode, struct file * filp);
2716 extern int nonseekable_open(struct inode * inode, struct file * filp);
2717 
2718 #ifdef CONFIG_BLOCK
2719 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2720                             loff_t file_offset);
2721 
2722 enum {
2723         /* need locking between buffered and direct access */
2724         DIO_LOCKING     = 0x01,
2725 
2726         /* filesystem does not support filling holes */
2727         DIO_SKIP_HOLES  = 0x02,
2728 
2729         /* filesystem can handle aio writes beyond i_size */
2730         DIO_ASYNC_EXTEND = 0x04,
2731 
2732         /* inode/fs/bdev does not need truncate protection */
2733         DIO_SKIP_DIO_COUNT = 0x08,
2734 };
2735 
2736 void dio_end_io(struct bio *bio, int error);
2737 
2738 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
2739                              struct block_device *bdev, struct iov_iter *iter,
2740                              loff_t offset, get_block_t get_block,
2741                              dio_iodone_t end_io, dio_submit_t submit_io,
2742                              int flags);
2743 
2744 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
2745                                          struct inode *inode,
2746                                          struct iov_iter *iter, loff_t offset,
2747                                          get_block_t get_block)
2748 {
2749         return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
2750                                     offset, get_block, NULL, NULL,
2751                                     DIO_LOCKING | DIO_SKIP_HOLES);
2752 }
2753 #endif
2754 
2755 void inode_dio_wait(struct inode *inode);
2756 
2757 /*
2758  * inode_dio_begin - signal start of a direct I/O requests
2759  * @inode: inode the direct I/O happens on
2760  *
2761  * This is called once we've finished processing a direct I/O request,
2762  * and is used to wake up callers waiting for direct I/O to be quiesced.
2763  */
2764 static inline void inode_dio_begin(struct inode *inode)
2765 {
2766         atomic_inc(&inode->i_dio_count);
2767 }
2768 
2769 /*
2770  * inode_dio_end - signal finish of a direct I/O requests
2771  * @inode: inode the direct I/O happens on
2772  *
2773  * This is called once we've finished processing a direct I/O request,
2774  * and is used to wake up callers waiting for direct I/O to be quiesced.
2775  */
2776 static inline void inode_dio_end(struct inode *inode)
2777 {
2778         if (atomic_dec_and_test(&inode->i_dio_count))
2779                 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
2780 }
2781 
2782 extern void inode_set_flags(struct inode *inode, unsigned int flags,
2783                             unsigned int mask);
2784 
2785 extern const struct file_operations generic_ro_fops;
2786 
2787 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2788 
2789 extern int readlink_copy(char __user *, int, const char *);
2790 extern int page_readlink(struct dentry *, char __user *, int);
2791 extern const char *page_get_link(struct dentry *, struct inode *,
2792                                  struct delayed_call *);
2793 extern void page_put_link(void *);
2794 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2795                 int nofs);
2796 extern int page_symlink(struct inode *inode, const char *symname, int len);
2797 extern const struct inode_operations page_symlink_inode_operations;
2798 extern void kfree_link(void *);
2799 extern int generic_readlink(struct dentry *, char __user *, int);
2800 extern void generic_fillattr(struct inode *, struct kstat *);
2801 int vfs_getattr_nosec(struct path *path, struct kstat *stat);
2802 extern int vfs_getattr(struct path *, struct kstat *);
2803 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2804 void inode_add_bytes(struct inode *inode, loff_t bytes);
2805 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
2806 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2807 loff_t inode_get_bytes(struct inode *inode);
2808 void inode_set_bytes(struct inode *inode, loff_t bytes);
2809 const char *simple_get_link(struct dentry *, struct inode *,
2810                             struct delayed_call *);
2811 extern const struct inode_operations simple_symlink_inode_operations;
2812 
2813 extern int iterate_dir(struct file *, struct dir_context *);
2814 
2815 extern int vfs_stat(const char __user *, struct kstat *);
2816 extern int vfs_lstat(const char __user *, struct kstat *);
2817 extern int vfs_fstat(unsigned int, struct kstat *);
2818 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2819 
2820 extern int __generic_block_fiemap(struct inode *inode,
2821                                   struct fiemap_extent_info *fieinfo,
2822                                   loff_t start, loff_t len,
2823                                   get_block_t *get_block);
2824 extern int generic_block_fiemap(struct inode *inode,
2825                                 struct fiemap_extent_info *fieinfo, u64 start,
2826                                 u64 len, get_block_t *get_block);
2827 
2828 extern void get_filesystem(struct file_system_type *fs);
2829 extern void put_filesystem(struct file_system_type *fs);
2830 extern struct file_system_type *get_fs_type(const char *name);
2831 extern struct super_block *get_super(struct block_device *);
2832 extern struct super_block *get_super_thawed(struct block_device *);
2833 extern struct super_block *get_active_super(struct block_device *bdev);
2834 extern void drop_super(struct super_block *sb);
2835 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2836 extern void iterate_supers_type(struct file_system_type *,
2837                                 void (*)(struct super_block *, void *), void *);
2838 
2839 extern int dcache_dir_open(struct inode *, struct file *);
2840 extern int dcache_dir_close(struct inode *, struct file *);
2841 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2842 extern int dcache_readdir(struct file *, struct dir_context *);
2843 extern int simple_setattr(struct dentry *, struct iattr *);
2844 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2845 extern int simple_statfs(struct dentry *, struct kstatfs *);
2846 extern int simple_open(struct inode *inode, struct file *file);
2847 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2848 extern int simple_unlink(struct inode *, struct dentry *);
2849 extern int simple_rmdir(struct inode *, struct dentry *);
2850 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2851 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2852 extern int simple_empty(struct dentry *);
2853 extern int simple_readpage(struct file *file, struct page *page);
2854 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2855                         loff_t pos, unsigned len, unsigned flags,
2856                         struct page **pagep, void **fsdata);
2857 extern int simple_write_end(struct file *file, struct address_space *mapping,
2858                         loff_t pos, unsigned len, unsigned copied,
2859                         struct page *page, void *fsdata);
2860 extern int always_delete_dentry(const struct dentry *);
2861 extern struct inode *alloc_anon_inode(struct super_block *);
2862 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
2863 extern const struct dentry_operations simple_dentry_operations;
2864 
2865 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
2866 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2867 extern const struct file_operations simple_dir_operations;
2868 extern const struct inode_operations simple_dir_inode_operations;
2869 extern void make_empty_dir_inode(struct inode *inode);
2870 extern bool is_empty_dir_inode(struct inode *inode);
2871 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2872 struct dentry *d_alloc_name(struct dentry *, const char *);
2873 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2874 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2875 extern void simple_release_fs(struct vfsmount **mount, int *count);
2876 
2877 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2878                         loff_t *ppos, const void *from, size_t available);
2879 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2880                 const void __user *from, size_t count);
2881 
2882 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
2883 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2884 
2885 extern int generic_check_addressable(unsigned, u64);
2886 
2887 #ifdef CONFIG_MIGRATION
2888 extern int buffer_migrate_page(struct address_space *,
2889                                 struct page *, struct page *,
2890                                 enum migrate_mode);
2891 #else
2892 #define buffer_migrate_page NULL
2893 #endif
2894 
2895 extern int inode_change_ok(const struct inode *, struct iattr *);
2896 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2897 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2898 
2899 extern int file_update_time(struct file *file);
2900 
2901 extern int generic_show_options(struct seq_file *m, struct dentry *root);
2902 extern void save_mount_options(struct super_block *sb, char *options);
2903 extern void replace_mount_options(struct super_block *sb, char *options);
2904 
2905 static inline bool io_is_direct(struct file *filp)
2906 {
2907         return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host);
2908 }
2909 
2910 static inline int iocb_flags(struct file *file)
2911 {
2912         int res = 0;
2913         if (file->f_flags & O_APPEND)
2914                 res |= IOCB_APPEND;
2915         if (io_is_direct(file))
2916                 res |= IOCB_DIRECT;
2917         return res;
2918 }
2919 
2920 static inline ino_t parent_ino(struct dentry *dentry)
2921 {
2922         ino_t res;
2923 
2924         /*
2925          * Don't strictly need d_lock here? If the parent ino could change
2926          * then surely we'd have a deeper race in the caller?
2927          */
2928         spin_lock(&dentry->d_lock);
2929         res = dentry->d_parent->d_inode->i_ino;
2930         spin_unlock(&dentry->d_lock);
2931         return res;
2932 }
2933 
2934 /* Transaction based IO helpers */
2935 
2936 /*
2937  * An argresp is stored in an allocated page and holds the
2938  * size of the argument or response, along with its content
2939  */
2940 struct simple_transaction_argresp {
2941         ssize_t size;
2942         char data[0];
2943 };
2944 
2945 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2946 
2947 char *simple_transaction_get(struct file *file, const char __user *buf,
2948                                 size_t size);
2949 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2950                                 size_t size, loff_t *pos);
2951 int simple_transaction_release(struct inode *inode, struct file *file);
2952 
2953 void simple_transaction_set(struct file *file, size_t n);
2954 
2955 /*
2956  * simple attribute files
2957  *
2958  * These attributes behave similar to those in sysfs:
2959  *
2960  * Writing to an attribute immediately sets a value, an open file can be
2961  * written to multiple times.
2962  *
2963  * Reading from an attribute creates a buffer from the value that might get
2964  * read with multiple read calls. When the attribute has been read
2965  * completely, no further read calls are possible until the file is opened
2966  * again.
2967  *
2968  * All attributes contain a text representation of a numeric value
2969  * that are accessed with the get() and set() functions.
2970  */
2971 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt)            \
2972 static int __fops ## _open(struct inode *inode, struct file *file)      \
2973 {                                                                       \
2974         __simple_attr_check_format(__fmt, 0ull);                        \
2975         return simple_attr_open(inode, file, __get, __set, __fmt);      \
2976 }                                                                       \
2977 static const struct file_operations __fops = {                          \
2978         .owner   = THIS_MODULE,                                         \
2979         .open    = __fops ## _open,                                     \
2980         .release = simple_attr_release,                                 \
2981         .read    = simple_attr_read,                                    \
2982         .write   = simple_attr_write,                                   \
2983         .llseek  = generic_file_llseek,                                 \
2984 }
2985 
2986 static inline __printf(1, 2)
2987 void __simple_attr_check_format(const char *fmt, ...)
2988 {
2989         /* don't do anything, just let the compiler check the arguments; */
2990 }
2991 
2992 int simple_attr_open(struct inode *inode, struct file *file,
2993                      int (*get)(void *, u64 *), int (*set)(void *, u64),
2994                      const char *fmt);
2995 int simple_attr_release(struct inode *inode, struct file *file);
2996 ssize_t simple_attr_read(struct file *file, char __user *buf,
2997                          size_t len, loff_t *ppos);
2998 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2999                           size_t len, loff_t *ppos);
3000 
3001 struct ctl_table;
3002 int proc_nr_files(struct ctl_table *table, int write,
3003                   void __user *buffer, size_t *lenp, loff_t *ppos);
3004 int proc_nr_dentry(struct ctl_table *table, int write,
3005                   void __user *buffer, size_t *lenp, loff_t *ppos);
3006 int proc_nr_inodes(struct ctl_table *table, int write,
3007                    void __user *buffer, size_t *lenp, loff_t *ppos);
3008 int __init get_filesystem_list(char *buf);
3009 
3010 #define __FMODE_EXEC            ((__force int) FMODE_EXEC)
3011 #define __FMODE_NONOTIFY        ((__force int) FMODE_NONOTIFY)
3012 
3013 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3014 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3015                                             (flag & __FMODE_NONOTIFY)))
3016 
3017 static inline bool is_sxid(umode_t mode)
3018 {
3019         return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3020 }
3021 
3022 static inline int check_sticky(struct inode *dir, struct inode *inode)
3023 {
3024         if (!(dir->i_mode & S_ISVTX))
3025                 return 0;
3026 
3027         return __check_sticky(dir, inode);
3028 }
3029 
3030 static inline void inode_has_no_xattr(struct inode *inode)
3031 {
3032         if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
3033                 inode->i_flags |= S_NOSEC;
3034 }
3035 
3036 static inline bool is_root_inode(struct inode *inode)
3037 {
3038         return inode == inode->i_sb->s_root->d_inode;
3039 }
3040 
3041 static inline bool dir_emit(struct dir_context *ctx,
3042                             const char *name, int namelen,
3043                             u64 ino, unsigned type)
3044 {
3045         return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3046 }
3047 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3048 {
3049         return ctx->actor(ctx, ".", 1, ctx->pos,
3050                           file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3051 }
3052 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3053 {
3054         return ctx->actor(ctx, "..", 2, ctx->pos,
3055                           parent_ino(file->f_path.dentry), DT_DIR) == 0;
3056 }
3057 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3058 {
3059         if (ctx->pos == 0) {
3060                 if (!dir_emit_dot(file, ctx))
3061                         return false;
3062                 ctx->pos = 1;
3063         }
3064         if (ctx->pos == 1) {
3065                 if (!dir_emit_dotdot(file, ctx))
3066                         return false;
3067                 ctx->pos = 2;
3068         }
3069         return true;
3070 }
3071 static inline bool dir_relax(struct inode *inode)
3072 {
3073         inode_unlock(inode);
3074         inode_lock(inode);
3075         return !IS_DEADDIR(inode);
3076 }
3077 
3078 extern bool path_noexec(const struct path *path);
3079 extern void inode_nohighmem(struct inode *inode);
3080 
3081 #endif /* _LINUX_FS_H */
3082 

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