Version:  2.0.40 2.2.26 2.4.37 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 4.6 4.7 4.8 4.9

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 struct fscrypt_info;
 57 struct fscrypt_operations;
 58 
 59 extern void __init inode_init(void);
 60 extern void __init inode_init_early(void);
 61 extern void __init files_init(void);
 62 extern void __init files_maxfiles_init(void);
 63 
 64 extern struct files_stat_struct files_stat;
 65 extern unsigned long get_max_files(void);
 66 extern unsigned int sysctl_nr_open;
 67 extern struct inodes_stat_t inodes_stat;
 68 extern int leases_enable, lease_break_time;
 69 extern int sysctl_protected_symlinks;
 70 extern int sysctl_protected_hardlinks;
 71 
 72 struct buffer_head;
 73 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
 74                         struct buffer_head *bh_result, int create);
 75 typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
 76                         ssize_t bytes, void *private);
 77 
 78 #define MAY_EXEC                0x00000001
 79 #define MAY_WRITE               0x00000002
 80 #define MAY_READ                0x00000004
 81 #define MAY_APPEND              0x00000008
 82 #define MAY_ACCESS              0x00000010
 83 #define MAY_OPEN                0x00000020
 84 #define MAY_CHDIR               0x00000040
 85 /* called from RCU mode, don't block */
 86 #define MAY_NOT_BLOCK           0x00000080
 87 
 88 /*
 89  * flags in file.f_mode.  Note that FMODE_READ and FMODE_WRITE must correspond
 90  * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
 91  */
 92 
 93 /* file is open for reading */
 94 #define FMODE_READ              ((__force fmode_t)0x1)
 95 /* file is open for writing */
 96 #define FMODE_WRITE             ((__force fmode_t)0x2)
 97 /* file is seekable */
 98 #define FMODE_LSEEK             ((__force fmode_t)0x4)
 99 /* file can be accessed using pread */
100 #define FMODE_PREAD             ((__force fmode_t)0x8)
101 /* file can be accessed using pwrite */
102 #define FMODE_PWRITE            ((__force fmode_t)0x10)
103 /* File is opened for execution with sys_execve / sys_uselib */
104 #define FMODE_EXEC              ((__force fmode_t)0x20)
105 /* File is opened with O_NDELAY (only set for block devices) */
106 #define FMODE_NDELAY            ((__force fmode_t)0x40)
107 /* File is opened with O_EXCL (only set for block devices) */
108 #define FMODE_EXCL              ((__force fmode_t)0x80)
109 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
110    (specialy hack for floppy.c) */
111 #define FMODE_WRITE_IOCTL       ((__force fmode_t)0x100)
112 /* 32bit hashes as llseek() offset (for directories) */
113 #define FMODE_32BITHASH         ((__force fmode_t)0x200)
114 /* 64bit hashes as llseek() offset (for directories) */
115 #define FMODE_64BITHASH         ((__force fmode_t)0x400)
116 
117 /*
118  * Don't update ctime and mtime.
119  *
120  * Currently a special hack for the XFS open_by_handle ioctl, but we'll
121  * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
122  */
123 #define FMODE_NOCMTIME          ((__force fmode_t)0x800)
124 
125 /* Expect random access pattern */
126 #define FMODE_RANDOM            ((__force fmode_t)0x1000)
127 
128 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
129 #define FMODE_UNSIGNED_OFFSET   ((__force fmode_t)0x2000)
130 
131 /* File is opened with O_PATH; almost nothing can be done with it */
132 #define FMODE_PATH              ((__force fmode_t)0x4000)
133 
134 /* File needs atomic accesses to f_pos */
135 #define FMODE_ATOMIC_POS        ((__force fmode_t)0x8000)
136 /* Write access to underlying fs */
137 #define FMODE_WRITER            ((__force fmode_t)0x10000)
138 /* Has read method(s) */
139 #define FMODE_CAN_READ          ((__force fmode_t)0x20000)
140 /* Has write method(s) */
141 #define FMODE_CAN_WRITE         ((__force fmode_t)0x40000)
142 
143 /* File was opened by fanotify and shouldn't generate fanotify events */
144 #define FMODE_NONOTIFY          ((__force fmode_t)0x4000000)
145 
146 /*
147  * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
148  * that indicates that they should check the contents of the iovec are
149  * valid, but not check the memory that the iovec elements
150  * points too.
151  */
152 #define CHECK_IOVEC_ONLY -1
153 
154 /*
155  * The below are the various read and write flags that we support. Some of
156  * them include behavioral modifiers that send information down to the
157  * block layer and IO scheduler. They should be used along with a req_op.
158  * Terminology:
159  *
160  *      The block layer uses device plugging to defer IO a little bit, in
161  *      the hope that we will see more IO very shortly. This increases
162  *      coalescing of adjacent IO and thus reduces the number of IOs we
163  *      have to send to the device. It also allows for better queuing,
164  *      if the IO isn't mergeable. If the caller is going to be waiting
165  *      for the IO, then he must ensure that the device is unplugged so
166  *      that the IO is dispatched to the driver.
167  *
168  *      All IO is handled async in Linux. This is fine for background
169  *      writes, but for reads or writes that someone waits for completion
170  *      on, we want to notify the block layer and IO scheduler so that they
171  *      know about it. That allows them to make better scheduling
172  *      decisions. So when the below references 'sync' and 'async', it
173  *      is referencing this priority hint.
174  *
175  * With that in mind, the available types are:
176  *
177  * READ                 A normal read operation. Device will be plugged.
178  * READ_SYNC            A synchronous read. Device is not plugged, caller can
179  *                      immediately wait on this read without caring about
180  *                      unplugging.
181  * WRITE                A normal async write. Device will be plugged.
182  * WRITE_SYNC           Synchronous write. Identical to WRITE, but passes down
183  *                      the hint that someone will be waiting on this IO
184  *                      shortly. The write equivalent of READ_SYNC.
185  * WRITE_ODIRECT        Special case write for O_DIRECT only.
186  * WRITE_FLUSH          Like WRITE_SYNC but with preceding cache flush.
187  * WRITE_FUA            Like WRITE_SYNC but data is guaranteed to be on
188  *                      non-volatile media on completion.
189  * WRITE_FLUSH_FUA      Combination of WRITE_FLUSH and FUA. The IO is preceded
190  *                      by a cache flush and data is guaranteed to be on
191  *                      non-volatile media on completion.
192  *
193  */
194 #define RW_MASK                 REQ_OP_WRITE
195 
196 #define READ                    REQ_OP_READ
197 #define WRITE                   REQ_OP_WRITE
198 
199 #define READ_SYNC               REQ_SYNC
200 #define WRITE_SYNC              (REQ_SYNC | REQ_NOIDLE)
201 #define WRITE_ODIRECT           REQ_SYNC
202 #define WRITE_FLUSH             (REQ_SYNC | REQ_NOIDLE | REQ_PREFLUSH)
203 #define WRITE_FUA               (REQ_SYNC | REQ_NOIDLE | REQ_FUA)
204 #define WRITE_FLUSH_FUA         (REQ_SYNC | REQ_NOIDLE | REQ_PREFLUSH | REQ_FUA)
205 
206 /*
207  * Attribute flags.  These should be or-ed together to figure out what
208  * has been changed!
209  */
210 #define ATTR_MODE       (1 << 0)
211 #define ATTR_UID        (1 << 1)
212 #define ATTR_GID        (1 << 2)
213 #define ATTR_SIZE       (1 << 3)
214 #define ATTR_ATIME      (1 << 4)
215 #define ATTR_MTIME      (1 << 5)
216 #define ATTR_CTIME      (1 << 6)
217 #define ATTR_ATIME_SET  (1 << 7)
218 #define ATTR_MTIME_SET  (1 << 8)
219 #define ATTR_FORCE      (1 << 9) /* Not a change, but a change it */
220 #define ATTR_ATTR_FLAG  (1 << 10)
221 #define ATTR_KILL_SUID  (1 << 11)
222 #define ATTR_KILL_SGID  (1 << 12)
223 #define ATTR_FILE       (1 << 13)
224 #define ATTR_KILL_PRIV  (1 << 14)
225 #define ATTR_OPEN       (1 << 15) /* Truncating from open(O_TRUNC) */
226 #define ATTR_TIMES_SET  (1 << 16)
227 #define ATTR_TOUCH      (1 << 17)
228 
229 /*
230  * Whiteout is represented by a char device.  The following constants define the
231  * mode and device number to use.
232  */
233 #define WHITEOUT_MODE 0
234 #define WHITEOUT_DEV 0
235 
236 /*
237  * This is the Inode Attributes structure, used for notify_change().  It
238  * uses the above definitions as flags, to know which values have changed.
239  * Also, in this manner, a Filesystem can look at only the values it cares
240  * about.  Basically, these are the attributes that the VFS layer can
241  * request to change from the FS layer.
242  *
243  * Derek Atkins <warlord@MIT.EDU> 94-10-20
244  */
245 struct iattr {
246         unsigned int    ia_valid;
247         umode_t         ia_mode;
248         kuid_t          ia_uid;
249         kgid_t          ia_gid;
250         loff_t          ia_size;
251         struct timespec ia_atime;
252         struct timespec ia_mtime;
253         struct timespec ia_ctime;
254 
255         /*
256          * Not an attribute, but an auxiliary info for filesystems wanting to
257          * implement an ftruncate() like method.  NOTE: filesystem should
258          * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
259          */
260         struct file     *ia_file;
261 };
262 
263 /*
264  * Includes for diskquotas.
265  */
266 #include <linux/quota.h>
267 
268 /*
269  * Maximum number of layers of fs stack.  Needs to be limited to
270  * prevent kernel stack overflow
271  */
272 #define FILESYSTEM_MAX_STACK_DEPTH 2
273 
274 /** 
275  * enum positive_aop_returns - aop return codes with specific semantics
276  *
277  * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
278  *                          completed, that the page is still locked, and
279  *                          should be considered active.  The VM uses this hint
280  *                          to return the page to the active list -- it won't
281  *                          be a candidate for writeback again in the near
282  *                          future.  Other callers must be careful to unlock
283  *                          the page if they get this return.  Returned by
284  *                          writepage(); 
285  *
286  * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
287  *                      unlocked it and the page might have been truncated.
288  *                      The caller should back up to acquiring a new page and
289  *                      trying again.  The aop will be taking reasonable
290  *                      precautions not to livelock.  If the caller held a page
291  *                      reference, it should drop it before retrying.  Returned
292  *                      by readpage().
293  *
294  * address_space_operation functions return these large constants to indicate
295  * special semantics to the caller.  These are much larger than the bytes in a
296  * page to allow for functions that return the number of bytes operated on in a
297  * given page.
298  */
299 
300 enum positive_aop_returns {
301         AOP_WRITEPAGE_ACTIVATE  = 0x80000,
302         AOP_TRUNCATED_PAGE      = 0x80001,
303 };
304 
305 #define AOP_FLAG_UNINTERRUPTIBLE        0x0001 /* will not do a short write */
306 #define AOP_FLAG_CONT_EXPAND            0x0002 /* called from cont_expand */
307 #define AOP_FLAG_NOFS                   0x0004 /* used by filesystem to direct
308                                                 * helper code (eg buffer layer)
309                                                 * to clear GFP_FS from alloc */
310 
311 /*
312  * oh the beauties of C type declarations.
313  */
314 struct page;
315 struct address_space;
316 struct writeback_control;
317 
318 #define IOCB_EVENTFD            (1 << 0)
319 #define IOCB_APPEND             (1 << 1)
320 #define IOCB_DIRECT             (1 << 2)
321 #define IOCB_HIPRI              (1 << 3)
322 #define IOCB_DSYNC              (1 << 4)
323 #define IOCB_SYNC               (1 << 5)
324 #define IOCB_WRITE              (1 << 6)
325 
326 struct kiocb {
327         struct file             *ki_filp;
328         loff_t                  ki_pos;
329         void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
330         void                    *private;
331         int                     ki_flags;
332 };
333 
334 static inline bool is_sync_kiocb(struct kiocb *kiocb)
335 {
336         return kiocb->ki_complete == NULL;
337 }
338 
339 static inline int iocb_flags(struct file *file);
340 
341 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
342 {
343         *kiocb = (struct kiocb) {
344                 .ki_filp = filp,
345                 .ki_flags = iocb_flags(filp),
346         };
347 }
348 
349 /*
350  * "descriptor" for what we're up to with a read.
351  * This allows us to use the same read code yet
352  * have multiple different users of the data that
353  * we read from a file.
354  *
355  * The simplest case just copies the data to user
356  * mode.
357  */
358 typedef struct {
359         size_t written;
360         size_t count;
361         union {
362                 char __user *buf;
363                 void *data;
364         } arg;
365         int error;
366 } read_descriptor_t;
367 
368 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
369                 unsigned long, unsigned long);
370 
371 struct address_space_operations {
372         int (*writepage)(struct page *page, struct writeback_control *wbc);
373         int (*readpage)(struct file *, struct page *);
374 
375         /* Write back some dirty pages from this mapping. */
376         int (*writepages)(struct address_space *, struct writeback_control *);
377 
378         /* Set a page dirty.  Return true if this dirtied it */
379         int (*set_page_dirty)(struct page *page);
380 
381         int (*readpages)(struct file *filp, struct address_space *mapping,
382                         struct list_head *pages, unsigned nr_pages);
383 
384         int (*write_begin)(struct file *, struct address_space *mapping,
385                                 loff_t pos, unsigned len, unsigned flags,
386                                 struct page **pagep, void **fsdata);
387         int (*write_end)(struct file *, struct address_space *mapping,
388                                 loff_t pos, unsigned len, unsigned copied,
389                                 struct page *page, void *fsdata);
390 
391         /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
392         sector_t (*bmap)(struct address_space *, sector_t);
393         void (*invalidatepage) (struct page *, unsigned int, unsigned int);
394         int (*releasepage) (struct page *, gfp_t);
395         void (*freepage)(struct page *);
396         ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
397         /*
398          * migrate the contents of a page to the specified target. If
399          * migrate_mode is MIGRATE_ASYNC, it must not block.
400          */
401         int (*migratepage) (struct address_space *,
402                         struct page *, struct page *, enum migrate_mode);
403         bool (*isolate_page)(struct page *, isolate_mode_t);
404         void (*putback_page)(struct page *);
405         int (*launder_page) (struct page *);
406         int (*is_partially_uptodate) (struct page *, unsigned long,
407                                         unsigned long);
408         void (*is_dirty_writeback) (struct page *, bool *, bool *);
409         int (*error_remove_page)(struct address_space *, struct page *);
410 
411         /* swapfile support */
412         int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
413                                 sector_t *span);
414         void (*swap_deactivate)(struct file *file);
415 };
416 
417 extern const struct address_space_operations empty_aops;
418 
419 /*
420  * pagecache_write_begin/pagecache_write_end must be used by general code
421  * to write into the pagecache.
422  */
423 int pagecache_write_begin(struct file *, struct address_space *mapping,
424                                 loff_t pos, unsigned len, unsigned flags,
425                                 struct page **pagep, void **fsdata);
426 
427 int pagecache_write_end(struct file *, struct address_space *mapping,
428                                 loff_t pos, unsigned len, unsigned copied,
429                                 struct page *page, void *fsdata);
430 
431 struct address_space {
432         struct inode            *host;          /* owner: inode, block_device */
433         struct radix_tree_root  page_tree;      /* radix tree of all pages */
434         spinlock_t              tree_lock;      /* and lock protecting it */
435         atomic_t                i_mmap_writable;/* count VM_SHARED mappings */
436         struct rb_root          i_mmap;         /* tree of private and shared mappings */
437         struct rw_semaphore     i_mmap_rwsem;   /* protect tree, count, list */
438         /* Protected by tree_lock together with the radix tree */
439         unsigned long           nrpages;        /* number of total pages */
440         /* number of shadow or DAX exceptional entries */
441         unsigned long           nrexceptional;
442         pgoff_t                 writeback_index;/* writeback starts here */
443         const struct address_space_operations *a_ops;   /* methods */
444         unsigned long           flags;          /* error bits */
445         spinlock_t              private_lock;   /* for use by the address_space */
446         gfp_t                   gfp_mask;       /* implicit gfp mask for allocations */
447         struct list_head        private_list;   /* ditto */
448         void                    *private_data;  /* ditto */
449 } __attribute__((aligned(sizeof(long))));
450         /*
451          * On most architectures that alignment is already the case; but
452          * must be enforced here for CRIS, to let the least significant bit
453          * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
454          */
455 struct request_queue;
456 
457 struct block_device {
458         dev_t                   bd_dev;  /* not a kdev_t - it's a search key */
459         int                     bd_openers;
460         struct inode *          bd_inode;       /* will die */
461         struct super_block *    bd_super;
462         struct mutex            bd_mutex;       /* open/close mutex */
463         void *                  bd_claiming;
464         void *                  bd_holder;
465         int                     bd_holders;
466         bool                    bd_write_holder;
467 #ifdef CONFIG_SYSFS
468         struct list_head        bd_holder_disks;
469 #endif
470         struct block_device *   bd_contains;
471         unsigned                bd_block_size;
472         struct hd_struct *      bd_part;
473         /* number of times partitions within this device have been opened. */
474         unsigned                bd_part_count;
475         int                     bd_invalidated;
476         struct gendisk *        bd_disk;
477         struct request_queue *  bd_queue;
478         struct list_head        bd_list;
479         /*
480          * Private data.  You must have bd_claim'ed the block_device
481          * to use this.  NOTE:  bd_claim allows an owner to claim
482          * the same device multiple times, the owner must take special
483          * care to not mess up bd_private for that case.
484          */
485         unsigned long           bd_private;
486 
487         /* The counter of freeze processes */
488         int                     bd_fsfreeze_count;
489         /* Mutex for freeze */
490         struct mutex            bd_fsfreeze_mutex;
491 };
492 
493 /*
494  * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
495  * radix trees
496  */
497 #define PAGECACHE_TAG_DIRTY     0
498 #define PAGECACHE_TAG_WRITEBACK 1
499 #define PAGECACHE_TAG_TOWRITE   2
500 
501 int mapping_tagged(struct address_space *mapping, int tag);
502 
503 static inline void i_mmap_lock_write(struct address_space *mapping)
504 {
505         down_write(&mapping->i_mmap_rwsem);
506 }
507 
508 static inline void i_mmap_unlock_write(struct address_space *mapping)
509 {
510         up_write(&mapping->i_mmap_rwsem);
511 }
512 
513 static inline void i_mmap_lock_read(struct address_space *mapping)
514 {
515         down_read(&mapping->i_mmap_rwsem);
516 }
517 
518 static inline void i_mmap_unlock_read(struct address_space *mapping)
519 {
520         up_read(&mapping->i_mmap_rwsem);
521 }
522 
523 /*
524  * Might pages of this file be mapped into userspace?
525  */
526 static inline int mapping_mapped(struct address_space *mapping)
527 {
528         return  !RB_EMPTY_ROOT(&mapping->i_mmap);
529 }
530 
531 /*
532  * Might pages of this file have been modified in userspace?
533  * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
534  * marks vma as VM_SHARED if it is shared, and the file was opened for
535  * writing i.e. vma may be mprotected writable even if now readonly.
536  *
537  * If i_mmap_writable is negative, no new writable mappings are allowed. You
538  * can only deny writable mappings, if none exists right now.
539  */
540 static inline int mapping_writably_mapped(struct address_space *mapping)
541 {
542         return atomic_read(&mapping->i_mmap_writable) > 0;
543 }
544 
545 static inline int mapping_map_writable(struct address_space *mapping)
546 {
547         return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
548                 0 : -EPERM;
549 }
550 
551 static inline void mapping_unmap_writable(struct address_space *mapping)
552 {
553         atomic_dec(&mapping->i_mmap_writable);
554 }
555 
556 static inline int mapping_deny_writable(struct address_space *mapping)
557 {
558         return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
559                 0 : -EBUSY;
560 }
561 
562 static inline void mapping_allow_writable(struct address_space *mapping)
563 {
564         atomic_inc(&mapping->i_mmap_writable);
565 }
566 
567 /*
568  * Use sequence counter to get consistent i_size on 32-bit processors.
569  */
570 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
571 #include <linux/seqlock.h>
572 #define __NEED_I_SIZE_ORDERED
573 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
574 #else
575 #define i_size_ordered_init(inode) do { } while (0)
576 #endif
577 
578 struct posix_acl;
579 #define ACL_NOT_CACHED ((void *)(-1))
580 #define ACL_DONT_CACHE ((void *)(-3))
581 
582 static inline struct posix_acl *
583 uncached_acl_sentinel(struct task_struct *task)
584 {
585         return (void *)task + 1;
586 }
587 
588 static inline bool
589 is_uncached_acl(struct posix_acl *acl)
590 {
591         return (long)acl & 1;
592 }
593 
594 #define IOP_FASTPERM    0x0001
595 #define IOP_LOOKUP      0x0002
596 #define IOP_NOFOLLOW    0x0004
597 #define IOP_XATTR       0x0008
598 
599 /*
600  * Keep mostly read-only and often accessed (especially for
601  * the RCU path lookup and 'stat' data) fields at the beginning
602  * of the 'struct inode'
603  */
604 struct inode {
605         umode_t                 i_mode;
606         unsigned short          i_opflags;
607         kuid_t                  i_uid;
608         kgid_t                  i_gid;
609         unsigned int            i_flags;
610 
611 #ifdef CONFIG_FS_POSIX_ACL
612         struct posix_acl        *i_acl;
613         struct posix_acl        *i_default_acl;
614 #endif
615 
616         const struct inode_operations   *i_op;
617         struct super_block      *i_sb;
618         struct address_space    *i_mapping;
619 
620 #ifdef CONFIG_SECURITY
621         void                    *i_security;
622 #endif
623 
624         /* Stat data, not accessed from path walking */
625         unsigned long           i_ino;
626         /*
627          * Filesystems may only read i_nlink directly.  They shall use the
628          * following functions for modification:
629          *
630          *    (set|clear|inc|drop)_nlink
631          *    inode_(inc|dec)_link_count
632          */
633         union {
634                 const unsigned int i_nlink;
635                 unsigned int __i_nlink;
636         };
637         dev_t                   i_rdev;
638         loff_t                  i_size;
639         struct timespec         i_atime;
640         struct timespec         i_mtime;
641         struct timespec         i_ctime;
642         spinlock_t              i_lock; /* i_blocks, i_bytes, maybe i_size */
643         unsigned short          i_bytes;
644         unsigned int            i_blkbits;
645         blkcnt_t                i_blocks;
646 
647 #ifdef __NEED_I_SIZE_ORDERED
648         seqcount_t              i_size_seqcount;
649 #endif
650 
651         /* Misc */
652         unsigned long           i_state;
653         struct rw_semaphore     i_rwsem;
654 
655         unsigned long           dirtied_when;   /* jiffies of first dirtying */
656         unsigned long           dirtied_time_when;
657 
658         struct hlist_node       i_hash;
659         struct list_head        i_io_list;      /* backing dev IO list */
660 #ifdef CONFIG_CGROUP_WRITEBACK
661         struct bdi_writeback    *i_wb;          /* the associated cgroup wb */
662 
663         /* foreign inode detection, see wbc_detach_inode() */
664         int                     i_wb_frn_winner;
665         u16                     i_wb_frn_avg_time;
666         u16                     i_wb_frn_history;
667 #endif
668         struct list_head        i_lru;          /* inode LRU list */
669         struct list_head        i_sb_list;
670         struct list_head        i_wb_list;      /* backing dev writeback list */
671         union {
672                 struct hlist_head       i_dentry;
673                 struct rcu_head         i_rcu;
674         };
675         u64                     i_version;
676         atomic_t                i_count;
677         atomic_t                i_dio_count;
678         atomic_t                i_writecount;
679 #ifdef CONFIG_IMA
680         atomic_t                i_readcount; /* struct files open RO */
681 #endif
682         const struct file_operations    *i_fop; /* former ->i_op->default_file_ops */
683         struct file_lock_context        *i_flctx;
684         struct address_space    i_data;
685         struct list_head        i_devices;
686         union {
687                 struct pipe_inode_info  *i_pipe;
688                 struct block_device     *i_bdev;
689                 struct cdev             *i_cdev;
690                 char                    *i_link;
691                 unsigned                i_dir_seq;
692         };
693 
694         __u32                   i_generation;
695 
696 #ifdef CONFIG_FSNOTIFY
697         __u32                   i_fsnotify_mask; /* all events this inode cares about */
698         struct hlist_head       i_fsnotify_marks;
699 #endif
700 
701 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
702         struct fscrypt_info     *i_crypt_info;
703 #endif
704 
705         void                    *i_private; /* fs or device private pointer */
706 };
707 
708 static inline int inode_unhashed(struct inode *inode)
709 {
710         return hlist_unhashed(&inode->i_hash);
711 }
712 
713 /*
714  * inode->i_mutex nesting subclasses for the lock validator:
715  *
716  * 0: the object of the current VFS operation
717  * 1: parent
718  * 2: child/target
719  * 3: xattr
720  * 4: second non-directory
721  * 5: second parent (when locking independent directories in rename)
722  *
723  * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
724  * non-directories at once.
725  *
726  * The locking order between these classes is
727  * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
728  */
729 enum inode_i_mutex_lock_class
730 {
731         I_MUTEX_NORMAL,
732         I_MUTEX_PARENT,
733         I_MUTEX_CHILD,
734         I_MUTEX_XATTR,
735         I_MUTEX_NONDIR2,
736         I_MUTEX_PARENT2,
737 };
738 
739 static inline void inode_lock(struct inode *inode)
740 {
741         down_write(&inode->i_rwsem);
742 }
743 
744 static inline void inode_unlock(struct inode *inode)
745 {
746         up_write(&inode->i_rwsem);
747 }
748 
749 static inline void inode_lock_shared(struct inode *inode)
750 {
751         down_read(&inode->i_rwsem);
752 }
753 
754 static inline void inode_unlock_shared(struct inode *inode)
755 {
756         up_read(&inode->i_rwsem);
757 }
758 
759 static inline int inode_trylock(struct inode *inode)
760 {
761         return down_write_trylock(&inode->i_rwsem);
762 }
763 
764 static inline int inode_trylock_shared(struct inode *inode)
765 {
766         return down_read_trylock(&inode->i_rwsem);
767 }
768 
769 static inline int inode_is_locked(struct inode *inode)
770 {
771         return rwsem_is_locked(&inode->i_rwsem);
772 }
773 
774 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
775 {
776         down_write_nested(&inode->i_rwsem, subclass);
777 }
778 
779 void lock_two_nondirectories(struct inode *, struct inode*);
780 void unlock_two_nondirectories(struct inode *, struct inode*);
781 
782 /*
783  * NOTE: in a 32bit arch with a preemptable kernel and
784  * an UP compile the i_size_read/write must be atomic
785  * with respect to the local cpu (unlike with preempt disabled),
786  * but they don't need to be atomic with respect to other cpus like in
787  * true SMP (so they need either to either locally disable irq around
788  * the read or for example on x86 they can be still implemented as a
789  * cmpxchg8b without the need of the lock prefix). For SMP compiles
790  * and 64bit archs it makes no difference if preempt is enabled or not.
791  */
792 static inline loff_t i_size_read(const struct inode *inode)
793 {
794 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
795         loff_t i_size;
796         unsigned int seq;
797 
798         do {
799                 seq = read_seqcount_begin(&inode->i_size_seqcount);
800                 i_size = inode->i_size;
801         } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
802         return i_size;
803 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
804         loff_t i_size;
805 
806         preempt_disable();
807         i_size = inode->i_size;
808         preempt_enable();
809         return i_size;
810 #else
811         return inode->i_size;
812 #endif
813 }
814 
815 /*
816  * NOTE: unlike i_size_read(), i_size_write() does need locking around it
817  * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
818  * can be lost, resulting in subsequent i_size_read() calls spinning forever.
819  */
820 static inline void i_size_write(struct inode *inode, loff_t i_size)
821 {
822 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
823         preempt_disable();
824         write_seqcount_begin(&inode->i_size_seqcount);
825         inode->i_size = i_size;
826         write_seqcount_end(&inode->i_size_seqcount);
827         preempt_enable();
828 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
829         preempt_disable();
830         inode->i_size = i_size;
831         preempt_enable();
832 #else
833         inode->i_size = i_size;
834 #endif
835 }
836 
837 static inline unsigned iminor(const struct inode *inode)
838 {
839         return MINOR(inode->i_rdev);
840 }
841 
842 static inline unsigned imajor(const struct inode *inode)
843 {
844         return MAJOR(inode->i_rdev);
845 }
846 
847 extern struct block_device *I_BDEV(struct inode *inode);
848 
849 struct fown_struct {
850         rwlock_t lock;          /* protects pid, uid, euid fields */
851         struct pid *pid;        /* pid or -pgrp where SIGIO should be sent */
852         enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
853         kuid_t uid, euid;       /* uid/euid of process setting the owner */
854         int signum;             /* posix.1b rt signal to be delivered on IO */
855 };
856 
857 /*
858  * Track a single file's readahead state
859  */
860 struct file_ra_state {
861         pgoff_t start;                  /* where readahead started */
862         unsigned int size;              /* # of readahead pages */
863         unsigned int async_size;        /* do asynchronous readahead when
864                                            there are only # of pages ahead */
865 
866         unsigned int ra_pages;          /* Maximum readahead window */
867         unsigned int mmap_miss;         /* Cache miss stat for mmap accesses */
868         loff_t prev_pos;                /* Cache last read() position */
869 };
870 
871 /*
872  * Check if @index falls in the readahead windows.
873  */
874 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
875 {
876         return (index >= ra->start &&
877                 index <  ra->start + ra->size);
878 }
879 
880 struct file {
881         union {
882                 struct llist_node       fu_llist;
883                 struct rcu_head         fu_rcuhead;
884         } f_u;
885         struct path             f_path;
886         struct inode            *f_inode;       /* cached value */
887         const struct file_operations    *f_op;
888 
889         /*
890          * Protects f_ep_links, f_flags.
891          * Must not be taken from IRQ context.
892          */
893         spinlock_t              f_lock;
894         atomic_long_t           f_count;
895         unsigned int            f_flags;
896         fmode_t                 f_mode;
897         struct mutex            f_pos_lock;
898         loff_t                  f_pos;
899         struct fown_struct      f_owner;
900         const struct cred       *f_cred;
901         struct file_ra_state    f_ra;
902 
903         u64                     f_version;
904 #ifdef CONFIG_SECURITY
905         void                    *f_security;
906 #endif
907         /* needed for tty driver, and maybe others */
908         void                    *private_data;
909 
910 #ifdef CONFIG_EPOLL
911         /* Used by fs/eventpoll.c to link all the hooks to this file */
912         struct list_head        f_ep_links;
913         struct list_head        f_tfile_llink;
914 #endif /* #ifdef CONFIG_EPOLL */
915         struct address_space    *f_mapping;
916 } __attribute__((aligned(4)));  /* lest something weird decides that 2 is OK */
917 
918 struct file_handle {
919         __u32 handle_bytes;
920         int handle_type;
921         /* file identifier */
922         unsigned char f_handle[0];
923 };
924 
925 static inline struct file *get_file(struct file *f)
926 {
927         atomic_long_inc(&f->f_count);
928         return f;
929 }
930 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
931 #define fput_atomic(x)  atomic_long_add_unless(&(x)->f_count, -1, 1)
932 #define file_count(x)   atomic_long_read(&(x)->f_count)
933 
934 #define MAX_NON_LFS     ((1UL<<31) - 1)
935 
936 /* Page cache limit. The filesystems should put that into their s_maxbytes 
937    limits, otherwise bad things can happen in VM. */ 
938 #if BITS_PER_LONG==32
939 #define MAX_LFS_FILESIZE        (((loff_t)PAGE_SIZE << (BITS_PER_LONG-1))-1)
940 #elif BITS_PER_LONG==64
941 #define MAX_LFS_FILESIZE        ((loff_t)0x7fffffffffffffffLL)
942 #endif
943 
944 #define FL_POSIX        1
945 #define FL_FLOCK        2
946 #define FL_DELEG        4       /* NFSv4 delegation */
947 #define FL_ACCESS       8       /* not trying to lock, just looking */
948 #define FL_EXISTS       16      /* when unlocking, test for existence */
949 #define FL_LEASE        32      /* lease held on this file */
950 #define FL_CLOSE        64      /* unlock on close */
951 #define FL_SLEEP        128     /* A blocking lock */
952 #define FL_DOWNGRADE_PENDING    256 /* Lease is being downgraded */
953 #define FL_UNLOCK_PENDING       512 /* Lease is being broken */
954 #define FL_OFDLCK       1024    /* lock is "owned" by struct file */
955 #define FL_LAYOUT       2048    /* outstanding pNFS layout */
956 
957 /*
958  * Special return value from posix_lock_file() and vfs_lock_file() for
959  * asynchronous locking.
960  */
961 #define FILE_LOCK_DEFERRED 1
962 
963 /* legacy typedef, should eventually be removed */
964 typedef void *fl_owner_t;
965 
966 struct file_lock;
967 
968 struct file_lock_operations {
969         void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
970         void (*fl_release_private)(struct file_lock *);
971 };
972 
973 struct lock_manager_operations {
974         int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
975         unsigned long (*lm_owner_key)(struct file_lock *);
976         fl_owner_t (*lm_get_owner)(fl_owner_t);
977         void (*lm_put_owner)(fl_owner_t);
978         void (*lm_notify)(struct file_lock *);  /* unblock callback */
979         int (*lm_grant)(struct file_lock *, int);
980         bool (*lm_break)(struct file_lock *);
981         int (*lm_change)(struct file_lock *, int, struct list_head *);
982         void (*lm_setup)(struct file_lock *, void **);
983 };
984 
985 struct lock_manager {
986         struct list_head list;
987         /*
988          * NFSv4 and up also want opens blocked during the grace period;
989          * NLM doesn't care:
990          */
991         bool block_opens;
992 };
993 
994 struct net;
995 void locks_start_grace(struct net *, struct lock_manager *);
996 void locks_end_grace(struct lock_manager *);
997 int locks_in_grace(struct net *);
998 int opens_in_grace(struct net *);
999 
1000 /* that will die - we need it for nfs_lock_info */
1001 #include <linux/nfs_fs_i.h>
1002 
1003 /*
1004  * struct file_lock represents a generic "file lock". It's used to represent
1005  * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1006  * note that the same struct is used to represent both a request for a lock and
1007  * the lock itself, but the same object is never used for both.
1008  *
1009  * FIXME: should we create a separate "struct lock_request" to help distinguish
1010  * these two uses?
1011  *
1012  * The varous i_flctx lists are ordered by:
1013  *
1014  * 1) lock owner
1015  * 2) lock range start
1016  * 3) lock range end
1017  *
1018  * Obviously, the last two criteria only matter for POSIX locks.
1019  */
1020 struct file_lock {
1021         struct file_lock *fl_next;      /* singly linked list for this inode  */
1022         struct list_head fl_list;       /* link into file_lock_context */
1023         struct hlist_node fl_link;      /* node in global lists */
1024         struct list_head fl_block;      /* circular list of blocked processes */
1025         fl_owner_t fl_owner;
1026         unsigned int fl_flags;
1027         unsigned char fl_type;
1028         unsigned int fl_pid;
1029         int fl_link_cpu;                /* what cpu's list is this on? */
1030         struct pid *fl_nspid;
1031         wait_queue_head_t fl_wait;
1032         struct file *fl_file;
1033         loff_t fl_start;
1034         loff_t fl_end;
1035 
1036         struct fasync_struct *  fl_fasync; /* for lease break notifications */
1037         /* for lease breaks: */
1038         unsigned long fl_break_time;
1039         unsigned long fl_downgrade_time;
1040 
1041         const struct file_lock_operations *fl_ops;      /* Callbacks for filesystems */
1042         const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1043         union {
1044                 struct nfs_lock_info    nfs_fl;
1045                 struct nfs4_lock_info   nfs4_fl;
1046                 struct {
1047                         struct list_head link;  /* link in AFS vnode's pending_locks list */
1048                         int state;              /* state of grant or error if -ve */
1049                 } afs;
1050         } fl_u;
1051 };
1052 
1053 struct file_lock_context {
1054         spinlock_t              flc_lock;
1055         struct list_head        flc_flock;
1056         struct list_head        flc_posix;
1057         struct list_head        flc_lease;
1058 };
1059 
1060 /* The following constant reflects the upper bound of the file/locking space */
1061 #ifndef OFFSET_MAX
1062 #define INT_LIMIT(x)    (~((x)1 << (sizeof(x)*8 - 1)))
1063 #define OFFSET_MAX      INT_LIMIT(loff_t)
1064 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1065 #endif
1066 
1067 #include <linux/fcntl.h>
1068 
1069 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1070 
1071 /*
1072  * Return the inode to use for locking
1073  *
1074  * For overlayfs this should be the overlay inode, not the real inode returned
1075  * by file_inode().  For any other fs file_inode(filp) and locks_inode(filp) are
1076  * equal.
1077  */
1078 static inline struct inode *locks_inode(const struct file *f)
1079 {
1080         return f->f_path.dentry->d_inode;
1081 }
1082 
1083 #ifdef CONFIG_FILE_LOCKING
1084 extern int fcntl_getlk(struct file *, unsigned int, struct flock __user *);
1085 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1086                         struct flock __user *);
1087 
1088 #if BITS_PER_LONG == 32
1089 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 __user *);
1090 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1091                         struct flock64 __user *);
1092 #endif
1093 
1094 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1095 extern int fcntl_getlease(struct file *filp);
1096 
1097 /* fs/locks.c */
1098 void locks_free_lock_context(struct inode *inode);
1099 void locks_free_lock(struct file_lock *fl);
1100 extern void locks_init_lock(struct file_lock *);
1101 extern struct file_lock * locks_alloc_lock(void);
1102 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1103 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1104 extern void locks_remove_posix(struct file *, fl_owner_t);
1105 extern void locks_remove_file(struct file *);
1106 extern void locks_release_private(struct file_lock *);
1107 extern void posix_test_lock(struct file *, struct file_lock *);
1108 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1109 extern int posix_unblock_lock(struct file_lock *);
1110 extern int vfs_test_lock(struct file *, struct file_lock *);
1111 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1112 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1113 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1114 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1115 extern void lease_get_mtime(struct inode *, struct timespec *time);
1116 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1117 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1118 extern int lease_modify(struct file_lock *, int, struct list_head *);
1119 struct files_struct;
1120 extern void show_fd_locks(struct seq_file *f,
1121                          struct file *filp, struct files_struct *files);
1122 #else /* !CONFIG_FILE_LOCKING */
1123 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1124                               struct flock __user *user)
1125 {
1126         return -EINVAL;
1127 }
1128 
1129 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1130                               unsigned int cmd, struct flock __user *user)
1131 {
1132         return -EACCES;
1133 }
1134 
1135 #if BITS_PER_LONG == 32
1136 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1137                                 struct flock64 __user *user)
1138 {
1139         return -EINVAL;
1140 }
1141 
1142 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1143                                 unsigned int cmd, struct flock64 __user *user)
1144 {
1145         return -EACCES;
1146 }
1147 #endif
1148 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1149 {
1150         return -EINVAL;
1151 }
1152 
1153 static inline int fcntl_getlease(struct file *filp)
1154 {
1155         return F_UNLCK;
1156 }
1157 
1158 static inline void
1159 locks_free_lock_context(struct inode *inode)
1160 {
1161 }
1162 
1163 static inline void locks_init_lock(struct file_lock *fl)
1164 {
1165         return;
1166 }
1167 
1168 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1169 {
1170         return;
1171 }
1172 
1173 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1174 {
1175         return;
1176 }
1177 
1178 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1179 {
1180         return;
1181 }
1182 
1183 static inline void locks_remove_file(struct file *filp)
1184 {
1185         return;
1186 }
1187 
1188 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1189 {
1190         return;
1191 }
1192 
1193 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1194                                   struct file_lock *conflock)
1195 {
1196         return -ENOLCK;
1197 }
1198 
1199 static inline int posix_unblock_lock(struct file_lock *waiter)
1200 {
1201         return -ENOENT;
1202 }
1203 
1204 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1205 {
1206         return 0;
1207 }
1208 
1209 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1210                                 struct file_lock *fl, struct file_lock *conf)
1211 {
1212         return -ENOLCK;
1213 }
1214 
1215 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1216 {
1217         return 0;
1218 }
1219 
1220 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1221 {
1222         return -ENOLCK;
1223 }
1224 
1225 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1226 {
1227         return 0;
1228 }
1229 
1230 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1231 {
1232         return;
1233 }
1234 
1235 static inline int generic_setlease(struct file *filp, long arg,
1236                                     struct file_lock **flp, void **priv)
1237 {
1238         return -EINVAL;
1239 }
1240 
1241 static inline int vfs_setlease(struct file *filp, long arg,
1242                                struct file_lock **lease, void **priv)
1243 {
1244         return -EINVAL;
1245 }
1246 
1247 static inline int lease_modify(struct file_lock *fl, int arg,
1248                                struct list_head *dispose)
1249 {
1250         return -EINVAL;
1251 }
1252 
1253 struct files_struct;
1254 static inline void show_fd_locks(struct seq_file *f,
1255                         struct file *filp, struct files_struct *files) {}
1256 #endif /* !CONFIG_FILE_LOCKING */
1257 
1258 static inline struct inode *file_inode(const struct file *f)
1259 {
1260         return f->f_inode;
1261 }
1262 
1263 static inline struct dentry *file_dentry(const struct file *file)
1264 {
1265         return d_real(file->f_path.dentry, file_inode(file), 0);
1266 }
1267 
1268 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1269 {
1270         return locks_lock_inode_wait(locks_inode(filp), fl);
1271 }
1272 
1273 struct fasync_struct {
1274         spinlock_t              fa_lock;
1275         int                     magic;
1276         int                     fa_fd;
1277         struct fasync_struct    *fa_next; /* singly linked list */
1278         struct file             *fa_file;
1279         struct rcu_head         fa_rcu;
1280 };
1281 
1282 #define FASYNC_MAGIC 0x4601
1283 
1284 /* SMP safe fasync helpers: */
1285 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1286 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1287 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1288 extern struct fasync_struct *fasync_alloc(void);
1289 extern void fasync_free(struct fasync_struct *);
1290 
1291 /* can be called from interrupts */
1292 extern void kill_fasync(struct fasync_struct **, int, int);
1293 
1294 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1295 extern void f_setown(struct file *filp, unsigned long arg, int force);
1296 extern void f_delown(struct file *filp);
1297 extern pid_t f_getown(struct file *filp);
1298 extern int send_sigurg(struct fown_struct *fown);
1299 
1300 struct mm_struct;
1301 
1302 /*
1303  *      Umount options
1304  */
1305 
1306 #define MNT_FORCE       0x00000001      /* Attempt to forcibily umount */
1307 #define MNT_DETACH      0x00000002      /* Just detach from the tree */
1308 #define MNT_EXPIRE      0x00000004      /* Mark for expiry */
1309 #define UMOUNT_NOFOLLOW 0x00000008      /* Don't follow symlink on umount */
1310 #define UMOUNT_UNUSED   0x80000000      /* Flag guaranteed to be unused */
1311 
1312 /* sb->s_iflags */
1313 #define SB_I_CGROUPWB   0x00000001      /* cgroup-aware writeback enabled */
1314 #define SB_I_NOEXEC     0x00000002      /* Ignore executables on this fs */
1315 #define SB_I_NODEV      0x00000004      /* Ignore devices on this fs */
1316 
1317 /* sb->s_iflags to limit user namespace mounts */
1318 #define SB_I_USERNS_VISIBLE             0x00000010 /* fstype already mounted */
1319 
1320 /* Possible states of 'frozen' field */
1321 enum {
1322         SB_UNFROZEN = 0,                /* FS is unfrozen */
1323         SB_FREEZE_WRITE = 1,            /* Writes, dir ops, ioctls frozen */
1324         SB_FREEZE_PAGEFAULT = 2,        /* Page faults stopped as well */
1325         SB_FREEZE_FS = 3,               /* For internal FS use (e.g. to stop
1326                                          * internal threads if needed) */
1327         SB_FREEZE_COMPLETE = 4,         /* ->freeze_fs finished successfully */
1328 };
1329 
1330 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1331 
1332 struct sb_writers {
1333         int                             frozen;         /* Is sb frozen? */
1334         wait_queue_head_t               wait_unfrozen;  /* for get_super_thawed() */
1335         struct percpu_rw_semaphore      rw_sem[SB_FREEZE_LEVELS];
1336 };
1337 
1338 struct super_block {
1339         struct list_head        s_list;         /* Keep this first */
1340         dev_t                   s_dev;          /* search index; _not_ kdev_t */
1341         unsigned char           s_blocksize_bits;
1342         unsigned long           s_blocksize;
1343         loff_t                  s_maxbytes;     /* Max file size */
1344         struct file_system_type *s_type;
1345         const struct super_operations   *s_op;
1346         const struct dquot_operations   *dq_op;
1347         const struct quotactl_ops       *s_qcop;
1348         const struct export_operations *s_export_op;
1349         unsigned long           s_flags;
1350         unsigned long           s_iflags;       /* internal SB_I_* flags */
1351         unsigned long           s_magic;
1352         struct dentry           *s_root;
1353         struct rw_semaphore     s_umount;
1354         int                     s_count;
1355         atomic_t                s_active;
1356 #ifdef CONFIG_SECURITY
1357         void                    *s_security;
1358 #endif
1359         const struct xattr_handler **s_xattr;
1360 
1361         const struct fscrypt_operations *s_cop;
1362 
1363         struct hlist_bl_head    s_anon;         /* anonymous dentries for (nfs) exporting */
1364         struct list_head        s_mounts;       /* list of mounts; _not_ for fs use */
1365         struct block_device     *s_bdev;
1366         struct backing_dev_info *s_bdi;
1367         struct mtd_info         *s_mtd;
1368         struct hlist_node       s_instances;
1369         unsigned int            s_quota_types;  /* Bitmask of supported quota types */
1370         struct quota_info       s_dquot;        /* Diskquota specific options */
1371 
1372         struct sb_writers       s_writers;
1373 
1374         char s_id[32];                          /* Informational name */
1375         u8 s_uuid[16];                          /* UUID */
1376 
1377         void                    *s_fs_info;     /* Filesystem private info */
1378         unsigned int            s_max_links;
1379         fmode_t                 s_mode;
1380 
1381         /* Granularity of c/m/atime in ns.
1382            Cannot be worse than a second */
1383         u32                s_time_gran;
1384 
1385         /*
1386          * The next field is for VFS *only*. No filesystems have any business
1387          * even looking at it. You had been warned.
1388          */
1389         struct mutex s_vfs_rename_mutex;        /* Kludge */
1390 
1391         /*
1392          * Filesystem subtype.  If non-empty the filesystem type field
1393          * in /proc/mounts will be "type.subtype"
1394          */
1395         char *s_subtype;
1396 
1397         /*
1398          * Saved mount options for lazy filesystems using
1399          * generic_show_options()
1400          */
1401         char __rcu *s_options;
1402         const struct dentry_operations *s_d_op; /* default d_op for dentries */
1403 
1404         /*
1405          * Saved pool identifier for cleancache (-1 means none)
1406          */
1407         int cleancache_poolid;
1408 
1409         struct shrinker s_shrink;       /* per-sb shrinker handle */
1410 
1411         /* Number of inodes with nlink == 0 but still referenced */
1412         atomic_long_t s_remove_count;
1413 
1414         /* Being remounted read-only */
1415         int s_readonly_remount;
1416 
1417         /* AIO completions deferred from interrupt context */
1418         struct workqueue_struct *s_dio_done_wq;
1419         struct hlist_head s_pins;
1420 
1421         /*
1422          * Owning user namespace and default context in which to
1423          * interpret filesystem uids, gids, quotas, device nodes,
1424          * xattrs and security labels.
1425          */
1426         struct user_namespace *s_user_ns;
1427 
1428         /*
1429          * Keep the lru lists last in the structure so they always sit on their
1430          * own individual cachelines.
1431          */
1432         struct list_lru         s_dentry_lru ____cacheline_aligned_in_smp;
1433         struct list_lru         s_inode_lru ____cacheline_aligned_in_smp;
1434         struct rcu_head         rcu;
1435         struct work_struct      destroy_work;
1436 
1437         struct mutex            s_sync_lock;    /* sync serialisation lock */
1438 
1439         /*
1440          * Indicates how deep in a filesystem stack this SB is
1441          */
1442         int s_stack_depth;
1443 
1444         /* s_inode_list_lock protects s_inodes */
1445         spinlock_t              s_inode_list_lock ____cacheline_aligned_in_smp;
1446         struct list_head        s_inodes;       /* all inodes */
1447 
1448         spinlock_t              s_inode_wblist_lock;
1449         struct list_head        s_inodes_wb;    /* writeback inodes */
1450 };
1451 
1452 /* Helper functions so that in most cases filesystems will
1453  * not need to deal directly with kuid_t and kgid_t and can
1454  * instead deal with the raw numeric values that are stored
1455  * in the filesystem.
1456  */
1457 static inline uid_t i_uid_read(const struct inode *inode)
1458 {
1459         return from_kuid(inode->i_sb->s_user_ns, inode->i_uid);
1460 }
1461 
1462 static inline gid_t i_gid_read(const struct inode *inode)
1463 {
1464         return from_kgid(inode->i_sb->s_user_ns, inode->i_gid);
1465 }
1466 
1467 static inline void i_uid_write(struct inode *inode, uid_t uid)
1468 {
1469         inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid);
1470 }
1471 
1472 static inline void i_gid_write(struct inode *inode, gid_t gid)
1473 {
1474         inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid);
1475 }
1476 
1477 extern struct timespec current_fs_time(struct super_block *sb);
1478 extern struct timespec current_time(struct inode *inode);
1479 
1480 /*
1481  * Snapshotting support.
1482  */
1483 
1484 void __sb_end_write(struct super_block *sb, int level);
1485 int __sb_start_write(struct super_block *sb, int level, bool wait);
1486 
1487 #define __sb_writers_acquired(sb, lev)  \
1488         percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1489 #define __sb_writers_release(sb, lev)   \
1490         percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1491 
1492 /**
1493  * sb_end_write - drop write access to a superblock
1494  * @sb: the super we wrote to
1495  *
1496  * Decrement number of writers to the filesystem. Wake up possible waiters
1497  * wanting to freeze the filesystem.
1498  */
1499 static inline void sb_end_write(struct super_block *sb)
1500 {
1501         __sb_end_write(sb, SB_FREEZE_WRITE);
1502 }
1503 
1504 /**
1505  * sb_end_pagefault - drop write access to a superblock from a page fault
1506  * @sb: the super we wrote to
1507  *
1508  * Decrement number of processes handling write page fault to the filesystem.
1509  * Wake up possible waiters wanting to freeze the filesystem.
1510  */
1511 static inline void sb_end_pagefault(struct super_block *sb)
1512 {
1513         __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1514 }
1515 
1516 /**
1517  * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1518  * @sb: the super we wrote to
1519  *
1520  * Decrement fs-internal number of writers to the filesystem.  Wake up possible
1521  * waiters wanting to freeze the filesystem.
1522  */
1523 static inline void sb_end_intwrite(struct super_block *sb)
1524 {
1525         __sb_end_write(sb, SB_FREEZE_FS);
1526 }
1527 
1528 /**
1529  * sb_start_write - get write access to a superblock
1530  * @sb: the super we write to
1531  *
1532  * When a process wants to write data or metadata to a file system (i.e. dirty
1533  * a page or an inode), it should embed the operation in a sb_start_write() -
1534  * sb_end_write() pair to get exclusion against file system freezing. This
1535  * function increments number of writers preventing freezing. If the file
1536  * system is already frozen, the function waits until the file system is
1537  * thawed.
1538  *
1539  * Since freeze protection behaves as a lock, users have to preserve
1540  * ordering of freeze protection and other filesystem locks. Generally,
1541  * freeze protection should be the outermost lock. In particular, we have:
1542  *
1543  * sb_start_write
1544  *   -> i_mutex                 (write path, truncate, directory ops, ...)
1545  *   -> s_umount                (freeze_super, thaw_super)
1546  */
1547 static inline void sb_start_write(struct super_block *sb)
1548 {
1549         __sb_start_write(sb, SB_FREEZE_WRITE, true);
1550 }
1551 
1552 static inline int sb_start_write_trylock(struct super_block *sb)
1553 {
1554         return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1555 }
1556 
1557 /**
1558  * sb_start_pagefault - get write access to a superblock from a page fault
1559  * @sb: the super we write to
1560  *
1561  * When a process starts handling write page fault, it should embed the
1562  * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1563  * exclusion against file system freezing. This is needed since the page fault
1564  * is going to dirty a page. This function increments number of running page
1565  * faults preventing freezing. If the file system is already frozen, the
1566  * function waits until the file system is thawed.
1567  *
1568  * Since page fault freeze protection behaves as a lock, users have to preserve
1569  * ordering of freeze protection and other filesystem locks. It is advised to
1570  * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1571  * handling code implies lock dependency:
1572  *
1573  * mmap_sem
1574  *   -> sb_start_pagefault
1575  */
1576 static inline void sb_start_pagefault(struct super_block *sb)
1577 {
1578         __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1579 }
1580 
1581 /*
1582  * sb_start_intwrite - get write access to a superblock for internal fs purposes
1583  * @sb: the super we write to
1584  *
1585  * This is the third level of protection against filesystem freezing. It is
1586  * free for use by a filesystem. The only requirement is that it must rank
1587  * below sb_start_pagefault.
1588  *
1589  * For example filesystem can call sb_start_intwrite() when starting a
1590  * transaction which somewhat eases handling of freezing for internal sources
1591  * of filesystem changes (internal fs threads, discarding preallocation on file
1592  * close, etc.).
1593  */
1594 static inline void sb_start_intwrite(struct super_block *sb)
1595 {
1596         __sb_start_write(sb, SB_FREEZE_FS, true);
1597 }
1598 
1599 
1600 extern bool inode_owner_or_capable(const struct inode *inode);
1601 
1602 /*
1603  * VFS helper functions..
1604  */
1605 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1606 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1607 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1608 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1609 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1610 extern int vfs_rmdir(struct inode *, struct dentry *);
1611 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1612 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1613 extern int vfs_whiteout(struct inode *, struct dentry *);
1614 
1615 /*
1616  * VFS file helper functions.
1617  */
1618 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1619                         umode_t mode);
1620 extern bool may_open_dev(const struct path *path);
1621 /*
1622  * VFS FS_IOC_FIEMAP helper definitions.
1623  */
1624 struct fiemap_extent_info {
1625         unsigned int fi_flags;          /* Flags as passed from user */
1626         unsigned int fi_extents_mapped; /* Number of mapped extents */
1627         unsigned int fi_extents_max;    /* Size of fiemap_extent array */
1628         struct fiemap_extent __user *fi_extents_start; /* Start of
1629                                                         fiemap_extent array */
1630 };
1631 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1632                             u64 phys, u64 len, u32 flags);
1633 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1634 
1635 /*
1636  * File types
1637  *
1638  * NOTE! These match bits 12..15 of stat.st_mode
1639  * (ie "(i_mode >> 12) & 15").
1640  */
1641 #define DT_UNKNOWN      0
1642 #define DT_FIFO         1
1643 #define DT_CHR          2
1644 #define DT_DIR          4
1645 #define DT_BLK          6
1646 #define DT_REG          8
1647 #define DT_LNK          10
1648 #define DT_SOCK         12
1649 #define DT_WHT          14
1650 
1651 /*
1652  * This is the "filldir" function type, used by readdir() to let
1653  * the kernel specify what kind of dirent layout it wants to have.
1654  * This allows the kernel to read directories into kernel space or
1655  * to have different dirent layouts depending on the binary type.
1656  */
1657 struct dir_context;
1658 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1659                          unsigned);
1660 
1661 struct dir_context {
1662         const filldir_t actor;
1663         loff_t pos;
1664 };
1665 
1666 struct block_device_operations;
1667 
1668 /* These macros are for out of kernel modules to test that
1669  * the kernel supports the unlocked_ioctl and compat_ioctl
1670  * fields in struct file_operations. */
1671 #define HAVE_COMPAT_IOCTL 1
1672 #define HAVE_UNLOCKED_IOCTL 1
1673 
1674 /*
1675  * These flags let !MMU mmap() govern direct device mapping vs immediate
1676  * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1677  *
1678  * NOMMU_MAP_COPY:      Copy can be mapped (MAP_PRIVATE)
1679  * NOMMU_MAP_DIRECT:    Can be mapped directly (MAP_SHARED)
1680  * NOMMU_MAP_READ:      Can be mapped for reading
1681  * NOMMU_MAP_WRITE:     Can be mapped for writing
1682  * NOMMU_MAP_EXEC:      Can be mapped for execution
1683  */
1684 #define NOMMU_MAP_COPY          0x00000001
1685 #define NOMMU_MAP_DIRECT        0x00000008
1686 #define NOMMU_MAP_READ          VM_MAYREAD
1687 #define NOMMU_MAP_WRITE         VM_MAYWRITE
1688 #define NOMMU_MAP_EXEC          VM_MAYEXEC
1689 
1690 #define NOMMU_VMFLAGS \
1691         (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1692 
1693 
1694 struct iov_iter;
1695 
1696 struct file_operations {
1697         struct module *owner;
1698         loff_t (*llseek) (struct file *, loff_t, int);
1699         ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1700         ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1701         ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1702         ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1703         int (*iterate) (struct file *, struct dir_context *);
1704         int (*iterate_shared) (struct file *, struct dir_context *);
1705         unsigned int (*poll) (struct file *, struct poll_table_struct *);
1706         long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1707         long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1708         int (*mmap) (struct file *, struct vm_area_struct *);
1709         int (*open) (struct inode *, struct file *);
1710         int (*flush) (struct file *, fl_owner_t id);
1711         int (*release) (struct inode *, struct file *);
1712         int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1713         int (*fasync) (int, struct file *, int);
1714         int (*lock) (struct file *, int, struct file_lock *);
1715         ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1716         unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1717         int (*check_flags)(int);
1718         int (*flock) (struct file *, int, struct file_lock *);
1719         ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1720         ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1721         int (*setlease)(struct file *, long, struct file_lock **, void **);
1722         long (*fallocate)(struct file *file, int mode, loff_t offset,
1723                           loff_t len);
1724         void (*show_fdinfo)(struct seq_file *m, struct file *f);
1725 #ifndef CONFIG_MMU
1726         unsigned (*mmap_capabilities)(struct file *);
1727 #endif
1728         ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1729                         loff_t, size_t, unsigned int);
1730         int (*clone_file_range)(struct file *, loff_t, struct file *, loff_t,
1731                         u64);
1732         ssize_t (*dedupe_file_range)(struct file *, u64, u64, struct file *,
1733                         u64);
1734 };
1735 
1736 struct inode_operations {
1737         struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1738         const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
1739         int (*permission) (struct inode *, int);
1740         struct posix_acl * (*get_acl)(struct inode *, int);
1741 
1742         int (*readlink) (struct dentry *, char __user *,int);
1743 
1744         int (*create) (struct inode *,struct dentry *, umode_t, bool);
1745         int (*link) (struct dentry *,struct inode *,struct dentry *);
1746         int (*unlink) (struct inode *,struct dentry *);
1747         int (*symlink) (struct inode *,struct dentry *,const char *);
1748         int (*mkdir) (struct inode *,struct dentry *,umode_t);
1749         int (*rmdir) (struct inode *,struct dentry *);
1750         int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1751         int (*rename) (struct inode *, struct dentry *,
1752                         struct inode *, struct dentry *, unsigned int);
1753         int (*setattr) (struct dentry *, struct iattr *);
1754         int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1755         ssize_t (*listxattr) (struct dentry *, char *, size_t);
1756         int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1757                       u64 len);
1758         int (*update_time)(struct inode *, struct timespec *, int);
1759         int (*atomic_open)(struct inode *, struct dentry *,
1760                            struct file *, unsigned open_flag,
1761                            umode_t create_mode, int *opened);
1762         int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1763         int (*set_acl)(struct inode *, struct posix_acl *, int);
1764 } ____cacheline_aligned;
1765 
1766 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1767                               unsigned long nr_segs, unsigned long fast_segs,
1768                               struct iovec *fast_pointer,
1769                               struct iovec **ret_pointer);
1770 
1771 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1772 extern ssize_t __vfs_write(struct file *, const char __user *, size_t, loff_t *);
1773 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1774 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1775 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1776                 unsigned long, loff_t *, int);
1777 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1778                 unsigned long, loff_t *, int);
1779 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1780                                    loff_t, size_t, unsigned int);
1781 extern int vfs_clone_file_range(struct file *file_in, loff_t pos_in,
1782                 struct file *file_out, loff_t pos_out, u64 len);
1783 extern int vfs_dedupe_file_range(struct file *file,
1784                                  struct file_dedupe_range *same);
1785 
1786 struct super_operations {
1787         struct inode *(*alloc_inode)(struct super_block *sb);
1788         void (*destroy_inode)(struct inode *);
1789 
1790         void (*dirty_inode) (struct inode *, int flags);
1791         int (*write_inode) (struct inode *, struct writeback_control *wbc);
1792         int (*drop_inode) (struct inode *);
1793         void (*evict_inode) (struct inode *);
1794         void (*put_super) (struct super_block *);
1795         int (*sync_fs)(struct super_block *sb, int wait);
1796         int (*freeze_super) (struct super_block *);
1797         int (*freeze_fs) (struct super_block *);
1798         int (*thaw_super) (struct super_block *);
1799         int (*unfreeze_fs) (struct super_block *);
1800         int (*statfs) (struct dentry *, struct kstatfs *);
1801         int (*remount_fs) (struct super_block *, int *, char *);
1802         void (*umount_begin) (struct super_block *);
1803 
1804         int (*show_options)(struct seq_file *, struct dentry *);
1805         int (*show_devname)(struct seq_file *, struct dentry *);
1806         int (*show_path)(struct seq_file *, struct dentry *);
1807         int (*show_stats)(struct seq_file *, struct dentry *);
1808 #ifdef CONFIG_QUOTA
1809         ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1810         ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1811         struct dquot **(*get_dquots)(struct inode *);
1812 #endif
1813         int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1814         long (*nr_cached_objects)(struct super_block *,
1815                                   struct shrink_control *);
1816         long (*free_cached_objects)(struct super_block *,
1817                                     struct shrink_control *);
1818 };
1819 
1820 /*
1821  * Inode flags - they have no relation to superblock flags now
1822  */
1823 #define S_SYNC          1       /* Writes are synced at once */
1824 #define S_NOATIME       2       /* Do not update access times */
1825 #define S_APPEND        4       /* Append-only file */
1826 #define S_IMMUTABLE     8       /* Immutable file */
1827 #define S_DEAD          16      /* removed, but still open directory */
1828 #define S_NOQUOTA       32      /* Inode is not counted to quota */
1829 #define S_DIRSYNC       64      /* Directory modifications are synchronous */
1830 #define S_NOCMTIME      128     /* Do not update file c/mtime */
1831 #define S_SWAPFILE      256     /* Do not truncate: swapon got its bmaps */
1832 #define S_PRIVATE       512     /* Inode is fs-internal */
1833 #define S_IMA           1024    /* Inode has an associated IMA struct */
1834 #define S_AUTOMOUNT     2048    /* Automount/referral quasi-directory */
1835 #define S_NOSEC         4096    /* no suid or xattr security attributes */
1836 #ifdef CONFIG_FS_DAX
1837 #define S_DAX           8192    /* Direct Access, avoiding the page cache */
1838 #else
1839 #define S_DAX           0       /* Make all the DAX code disappear */
1840 #endif
1841 
1842 /*
1843  * Note that nosuid etc flags are inode-specific: setting some file-system
1844  * flags just means all the inodes inherit those flags by default. It might be
1845  * possible to override it selectively if you really wanted to with some
1846  * ioctl() that is not currently implemented.
1847  *
1848  * Exception: MS_RDONLY is always applied to the entire file system.
1849  *
1850  * Unfortunately, it is possible to change a filesystems flags with it mounted
1851  * with files in use.  This means that all of the inodes will not have their
1852  * i_flags updated.  Hence, i_flags no longer inherit the superblock mount
1853  * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1854  */
1855 #define __IS_FLG(inode, flg)    ((inode)->i_sb->s_flags & (flg))
1856 
1857 #define IS_RDONLY(inode)        ((inode)->i_sb->s_flags & MS_RDONLY)
1858 #define IS_SYNC(inode)          (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1859                                         ((inode)->i_flags & S_SYNC))
1860 #define IS_DIRSYNC(inode)       (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1861                                         ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1862 #define IS_MANDLOCK(inode)      __IS_FLG(inode, MS_MANDLOCK)
1863 #define IS_NOATIME(inode)       __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1864 #define IS_I_VERSION(inode)     __IS_FLG(inode, MS_I_VERSION)
1865 
1866 #define IS_NOQUOTA(inode)       ((inode)->i_flags & S_NOQUOTA)
1867 #define IS_APPEND(inode)        ((inode)->i_flags & S_APPEND)
1868 #define IS_IMMUTABLE(inode)     ((inode)->i_flags & S_IMMUTABLE)
1869 #define IS_POSIXACL(inode)      __IS_FLG(inode, MS_POSIXACL)
1870 
1871 #define IS_DEADDIR(inode)       ((inode)->i_flags & S_DEAD)
1872 #define IS_NOCMTIME(inode)      ((inode)->i_flags & S_NOCMTIME)
1873 #define IS_SWAPFILE(inode)      ((inode)->i_flags & S_SWAPFILE)
1874 #define IS_PRIVATE(inode)       ((inode)->i_flags & S_PRIVATE)
1875 #define IS_IMA(inode)           ((inode)->i_flags & S_IMA)
1876 #define IS_AUTOMOUNT(inode)     ((inode)->i_flags & S_AUTOMOUNT)
1877 #define IS_NOSEC(inode)         ((inode)->i_flags & S_NOSEC)
1878 #define IS_DAX(inode)           ((inode)->i_flags & S_DAX)
1879 
1880 #define IS_WHITEOUT(inode)      (S_ISCHR(inode->i_mode) && \
1881                                  (inode)->i_rdev == WHITEOUT_DEV)
1882 
1883 static inline bool HAS_UNMAPPED_ID(struct inode *inode)
1884 {
1885         return !uid_valid(inode->i_uid) || !gid_valid(inode->i_gid);
1886 }
1887 
1888 /*
1889  * Inode state bits.  Protected by inode->i_lock
1890  *
1891  * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1892  * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1893  *
1894  * Four bits define the lifetime of an inode.  Initially, inodes are I_NEW,
1895  * until that flag is cleared.  I_WILL_FREE, I_FREEING and I_CLEAR are set at
1896  * various stages of removing an inode.
1897  *
1898  * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1899  *
1900  * I_DIRTY_SYNC         Inode is dirty, but doesn't have to be written on
1901  *                      fdatasync().  i_atime is the usual cause.
1902  * I_DIRTY_DATASYNC     Data-related inode changes pending. We keep track of
1903  *                      these changes separately from I_DIRTY_SYNC so that we
1904  *                      don't have to write inode on fdatasync() when only
1905  *                      mtime has changed in it.
1906  * I_DIRTY_PAGES        Inode has dirty pages.  Inode itself may be clean.
1907  * I_NEW                Serves as both a mutex and completion notification.
1908  *                      New inodes set I_NEW.  If two processes both create
1909  *                      the same inode, one of them will release its inode and
1910  *                      wait for I_NEW to be released before returning.
1911  *                      Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1912  *                      also cause waiting on I_NEW, without I_NEW actually
1913  *                      being set.  find_inode() uses this to prevent returning
1914  *                      nearly-dead inodes.
1915  * I_WILL_FREE          Must be set when calling write_inode_now() if i_count
1916  *                      is zero.  I_FREEING must be set when I_WILL_FREE is
1917  *                      cleared.
1918  * I_FREEING            Set when inode is about to be freed but still has dirty
1919  *                      pages or buffers attached or the inode itself is still
1920  *                      dirty.
1921  * I_CLEAR              Added by clear_inode().  In this state the inode is
1922  *                      clean and can be destroyed.  Inode keeps I_FREEING.
1923  *
1924  *                      Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1925  *                      prohibited for many purposes.  iget() must wait for
1926  *                      the inode to be completely released, then create it
1927  *                      anew.  Other functions will just ignore such inodes,
1928  *                      if appropriate.  I_NEW is used for waiting.
1929  *
1930  * I_SYNC               Writeback of inode is running. The bit is set during
1931  *                      data writeback, and cleared with a wakeup on the bit
1932  *                      address once it is done. The bit is also used to pin
1933  *                      the inode in memory for flusher thread.
1934  *
1935  * I_REFERENCED         Marks the inode as recently references on the LRU list.
1936  *
1937  * I_DIO_WAKEUP         Never set.  Only used as a key for wait_on_bit().
1938  *
1939  * I_WB_SWITCH          Cgroup bdi_writeback switching in progress.  Used to
1940  *                      synchronize competing switching instances and to tell
1941  *                      wb stat updates to grab mapping->tree_lock.  See
1942  *                      inode_switch_wb_work_fn() for details.
1943  *
1944  * Q: What is the difference between I_WILL_FREE and I_FREEING?
1945  */
1946 #define I_DIRTY_SYNC            (1 << 0)
1947 #define I_DIRTY_DATASYNC        (1 << 1)
1948 #define I_DIRTY_PAGES           (1 << 2)
1949 #define __I_NEW                 3
1950 #define I_NEW                   (1 << __I_NEW)
1951 #define I_WILL_FREE             (1 << 4)
1952 #define I_FREEING               (1 << 5)
1953 #define I_CLEAR                 (1 << 6)
1954 #define __I_SYNC                7
1955 #define I_SYNC                  (1 << __I_SYNC)
1956 #define I_REFERENCED            (1 << 8)
1957 #define __I_DIO_WAKEUP          9
1958 #define I_DIO_WAKEUP            (1 << __I_DIO_WAKEUP)
1959 #define I_LINKABLE              (1 << 10)
1960 #define I_DIRTY_TIME            (1 << 11)
1961 #define __I_DIRTY_TIME_EXPIRED  12
1962 #define I_DIRTY_TIME_EXPIRED    (1 << __I_DIRTY_TIME_EXPIRED)
1963 #define I_WB_SWITCH             (1 << 13)
1964 
1965 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1966 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1967 
1968 extern void __mark_inode_dirty(struct inode *, int);
1969 static inline void mark_inode_dirty(struct inode *inode)
1970 {
1971         __mark_inode_dirty(inode, I_DIRTY);
1972 }
1973 
1974 static inline void mark_inode_dirty_sync(struct inode *inode)
1975 {
1976         __mark_inode_dirty(inode, I_DIRTY_SYNC);
1977 }
1978 
1979 extern void inc_nlink(struct inode *inode);
1980 extern void drop_nlink(struct inode *inode);
1981 extern void clear_nlink(struct inode *inode);
1982 extern void set_nlink(struct inode *inode, unsigned int nlink);
1983 
1984 static inline void inode_inc_link_count(struct inode *inode)
1985 {
1986         inc_nlink(inode);
1987         mark_inode_dirty(inode);
1988 }
1989 
1990 static inline void inode_dec_link_count(struct inode *inode)
1991 {
1992         drop_nlink(inode);
1993         mark_inode_dirty(inode);
1994 }
1995 
1996 /**
1997  * inode_inc_iversion - increments i_version
1998  * @inode: inode that need to be updated
1999  *
2000  * Every time the inode is modified, the i_version field will be incremented.
2001  * The filesystem has to be mounted with i_version flag
2002  */
2003 
2004 static inline void inode_inc_iversion(struct inode *inode)
2005 {
2006        spin_lock(&inode->i_lock);
2007        inode->i_version++;
2008        spin_unlock(&inode->i_lock);
2009 }
2010 
2011 enum file_time_flags {
2012         S_ATIME = 1,
2013         S_MTIME = 2,
2014         S_CTIME = 4,
2015         S_VERSION = 8,
2016 };
2017 
2018 extern void touch_atime(const struct path *);
2019 static inline void file_accessed(struct file *file)
2020 {
2021         if (!(file->f_flags & O_NOATIME))
2022                 touch_atime(&file->f_path);
2023 }
2024 
2025 int sync_inode(struct inode *inode, struct writeback_control *wbc);
2026 int sync_inode_metadata(struct inode *inode, int wait);
2027 
2028 struct file_system_type {
2029         const char *name;
2030         int fs_flags;
2031 #define FS_REQUIRES_DEV         1 
2032 #define FS_BINARY_MOUNTDATA     2
2033 #define FS_HAS_SUBTYPE          4
2034 #define FS_USERNS_MOUNT         8       /* Can be mounted by userns root */
2035 #define FS_RENAME_DOES_D_MOVE   32768   /* FS will handle d_move() during rename() internally. */
2036         struct dentry *(*mount) (struct file_system_type *, int,
2037                        const char *, void *);
2038         void (*kill_sb) (struct super_block *);
2039         struct module *owner;
2040         struct file_system_type * next;
2041         struct hlist_head fs_supers;
2042 
2043         struct lock_class_key s_lock_key;
2044         struct lock_class_key s_umount_key;
2045         struct lock_class_key s_vfs_rename_key;
2046         struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2047 
2048         struct lock_class_key i_lock_key;
2049         struct lock_class_key i_mutex_key;
2050         struct lock_class_key i_mutex_dir_key;
2051 };
2052 
2053 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2054 
2055 extern struct dentry *mount_ns(struct file_system_type *fs_type,
2056         int flags, void *data, void *ns, struct user_namespace *user_ns,
2057         int (*fill_super)(struct super_block *, void *, int));
2058 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2059         int flags, const char *dev_name, void *data,
2060         int (*fill_super)(struct super_block *, void *, int));
2061 extern struct dentry *mount_single(struct file_system_type *fs_type,
2062         int flags, void *data,
2063         int (*fill_super)(struct super_block *, void *, int));
2064 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2065         int flags, void *data,
2066         int (*fill_super)(struct super_block *, void *, int));
2067 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2068 void generic_shutdown_super(struct super_block *sb);
2069 void kill_block_super(struct super_block *sb);
2070 void kill_anon_super(struct super_block *sb);
2071 void kill_litter_super(struct super_block *sb);
2072 void deactivate_super(struct super_block *sb);
2073 void deactivate_locked_super(struct super_block *sb);
2074 int set_anon_super(struct super_block *s, void *data);
2075 int get_anon_bdev(dev_t *);
2076 void free_anon_bdev(dev_t);
2077 struct super_block *sget_userns(struct file_system_type *type,
2078                         int (*test)(struct super_block *,void *),
2079                         int (*set)(struct super_block *,void *),
2080                         int flags, struct user_namespace *user_ns,
2081                         void *data);
2082 struct super_block *sget(struct file_system_type *type,
2083                         int (*test)(struct super_block *,void *),
2084                         int (*set)(struct super_block *,void *),
2085                         int flags, void *data);
2086 extern struct dentry *mount_pseudo_xattr(struct file_system_type *, char *,
2087                                          const struct super_operations *ops,
2088                                          const struct xattr_handler **xattr,
2089                                          const struct dentry_operations *dops,
2090                                          unsigned long);
2091 
2092 static inline struct dentry *
2093 mount_pseudo(struct file_system_type *fs_type, char *name,
2094              const struct super_operations *ops,
2095              const struct dentry_operations *dops, unsigned long magic)
2096 {
2097         return mount_pseudo_xattr(fs_type, name, ops, NULL, dops, magic);
2098 }
2099 
2100 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2101 #define fops_get(fops) \
2102         (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2103 #define fops_put(fops) \
2104         do { if (fops) module_put((fops)->owner); } while(0)
2105 /*
2106  * This one is to be used *ONLY* from ->open() instances.
2107  * fops must be non-NULL, pinned down *and* module dependencies
2108  * should be sufficient to pin the caller down as well.
2109  */
2110 #define replace_fops(f, fops) \
2111         do {    \
2112                 struct file *__file = (f); \
2113                 fops_put(__file->f_op); \
2114                 BUG_ON(!(__file->f_op = (fops))); \
2115         } while(0)
2116 
2117 extern int register_filesystem(struct file_system_type *);
2118 extern int unregister_filesystem(struct file_system_type *);
2119 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
2120 #define kern_mount(type) kern_mount_data(type, NULL)
2121 extern void kern_unmount(struct vfsmount *mnt);
2122 extern int may_umount_tree(struct vfsmount *);
2123 extern int may_umount(struct vfsmount *);
2124 extern long do_mount(const char *, const char __user *,
2125                      const char *, unsigned long, void *);
2126 extern struct vfsmount *collect_mounts(struct path *);
2127 extern void drop_collected_mounts(struct vfsmount *);
2128 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2129                           struct vfsmount *);
2130 extern int vfs_statfs(struct path *, struct kstatfs *);
2131 extern int user_statfs(const char __user *, struct kstatfs *);
2132 extern int fd_statfs(int, struct kstatfs *);
2133 extern int vfs_ustat(dev_t, struct kstatfs *);
2134 extern int freeze_super(struct super_block *super);
2135 extern int thaw_super(struct super_block *super);
2136 extern bool our_mnt(struct vfsmount *mnt);
2137 
2138 extern int current_umask(void);
2139 
2140 extern void ihold(struct inode * inode);
2141 extern void iput(struct inode *);
2142 extern int generic_update_time(struct inode *, struct timespec *, int);
2143 
2144 /* /sys/fs */
2145 extern struct kobject *fs_kobj;
2146 
2147 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2148 
2149 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2150 extern int locks_mandatory_locked(struct file *);
2151 extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
2152 
2153 /*
2154  * Candidates for mandatory locking have the setgid bit set
2155  * but no group execute bit -  an otherwise meaningless combination.
2156  */
2157 
2158 static inline int __mandatory_lock(struct inode *ino)
2159 {
2160         return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2161 }
2162 
2163 /*
2164  * ... and these candidates should be on MS_MANDLOCK mounted fs,
2165  * otherwise these will be advisory locks
2166  */
2167 
2168 static inline int mandatory_lock(struct inode *ino)
2169 {
2170         return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2171 }
2172 
2173 static inline int locks_verify_locked(struct file *file)
2174 {
2175         if (mandatory_lock(locks_inode(file)))
2176                 return locks_mandatory_locked(file);
2177         return 0;
2178 }
2179 
2180 static inline int locks_verify_truncate(struct inode *inode,
2181                                     struct file *f,
2182                                     loff_t size)
2183 {
2184         if (!inode->i_flctx || !mandatory_lock(inode))
2185                 return 0;
2186 
2187         if (size < inode->i_size) {
2188                 return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2189                                 F_WRLCK);
2190         } else {
2191                 return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2192                                 F_WRLCK);
2193         }
2194 }
2195 
2196 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2197 
2198 static inline int locks_mandatory_locked(struct file *file)
2199 {
2200         return 0;
2201 }
2202 
2203 static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2204                                        loff_t start, loff_t end, unsigned char type)
2205 {
2206         return 0;
2207 }
2208 
2209 static inline int __mandatory_lock(struct inode *inode)
2210 {
2211         return 0;
2212 }
2213 
2214 static inline int mandatory_lock(struct inode *inode)
2215 {
2216         return 0;
2217 }
2218 
2219 static inline int locks_verify_locked(struct file *file)
2220 {
2221         return 0;
2222 }
2223 
2224 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2225                                         size_t size)
2226 {
2227         return 0;
2228 }
2229 
2230 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2231 
2232 
2233 #ifdef CONFIG_FILE_LOCKING
2234 static inline int break_lease(struct inode *inode, unsigned int mode)
2235 {
2236         /*
2237          * Since this check is lockless, we must ensure that any refcounts
2238          * taken are done before checking i_flctx->flc_lease. Otherwise, we
2239          * could end up racing with tasks trying to set a new lease on this
2240          * file.
2241          */
2242         smp_mb();
2243         if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2244                 return __break_lease(inode, mode, FL_LEASE);
2245         return 0;
2246 }
2247 
2248 static inline int break_deleg(struct inode *inode, unsigned int mode)
2249 {
2250         /*
2251          * Since this check is lockless, we must ensure that any refcounts
2252          * taken are done before checking i_flctx->flc_lease. Otherwise, we
2253          * could end up racing with tasks trying to set a new lease on this
2254          * file.
2255          */
2256         smp_mb();
2257         if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2258                 return __break_lease(inode, mode, FL_DELEG);
2259         return 0;
2260 }
2261 
2262 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2263 {
2264         int ret;
2265 
2266         ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2267         if (ret == -EWOULDBLOCK && delegated_inode) {
2268                 *delegated_inode = inode;
2269                 ihold(inode);
2270         }
2271         return ret;
2272 }
2273 
2274 static inline int break_deleg_wait(struct inode **delegated_inode)
2275 {
2276         int ret;
2277 
2278         ret = break_deleg(*delegated_inode, O_WRONLY);
2279         iput(*delegated_inode);
2280         *delegated_inode = NULL;
2281         return ret;
2282 }
2283 
2284 static inline int break_layout(struct inode *inode, bool wait)
2285 {
2286         smp_mb();
2287         if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2288                 return __break_lease(inode,
2289                                 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2290                                 FL_LAYOUT);
2291         return 0;
2292 }
2293 
2294 #else /* !CONFIG_FILE_LOCKING */
2295 static inline int break_lease(struct inode *inode, unsigned int mode)
2296 {
2297         return 0;
2298 }
2299 
2300 static inline int break_deleg(struct inode *inode, unsigned int mode)
2301 {
2302         return 0;
2303 }
2304 
2305 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2306 {
2307         return 0;
2308 }
2309 
2310 static inline int break_deleg_wait(struct inode **delegated_inode)
2311 {
2312         BUG();
2313         return 0;
2314 }
2315 
2316 static inline int break_layout(struct inode *inode, bool wait)
2317 {
2318         return 0;
2319 }
2320 
2321 #endif /* CONFIG_FILE_LOCKING */
2322 
2323 /* fs/open.c */
2324 struct audit_names;
2325 struct filename {
2326         const char              *name;  /* pointer to actual string */
2327         const __user char       *uptr;  /* original userland pointer */
2328         struct audit_names      *aname;
2329         int                     refcnt;
2330         const char              iname[];
2331 };
2332 
2333 extern long vfs_truncate(const struct path *, loff_t);
2334 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2335                        struct file *filp);
2336 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2337                         loff_t len);
2338 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2339                         umode_t mode);
2340 extern struct file *file_open_name(struct filename *, int, umode_t);
2341 extern struct file *filp_open(const char *, int, umode_t);
2342 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2343                                    const char *, int, umode_t);
2344 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2345 extern int filp_close(struct file *, fl_owner_t id);
2346 
2347 extern struct filename *getname_flags(const char __user *, int, int *);
2348 extern struct filename *getname(const char __user *);
2349 extern struct filename *getname_kernel(const char *);
2350 extern void putname(struct filename *name);
2351 
2352 enum {
2353         FILE_CREATED = 1,
2354         FILE_OPENED = 2
2355 };
2356 extern int finish_open(struct file *file, struct dentry *dentry,
2357                         int (*open)(struct inode *, struct file *),
2358                         int *opened);
2359 extern int finish_no_open(struct file *file, struct dentry *dentry);
2360 
2361 /* fs/ioctl.c */
2362 
2363 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2364 
2365 /* fs/dcache.c */
2366 extern void __init vfs_caches_init_early(void);
2367 extern void __init vfs_caches_init(void);
2368 
2369 extern struct kmem_cache *names_cachep;
2370 
2371 #define __getname()             kmem_cache_alloc(names_cachep, GFP_KERNEL)
2372 #define __putname(name)         kmem_cache_free(names_cachep, (void *)(name))
2373 
2374 #ifdef CONFIG_BLOCK
2375 extern int register_blkdev(unsigned int, const char *);
2376 extern void unregister_blkdev(unsigned int, const char *);
2377 extern struct block_device *bdget(dev_t);
2378 extern struct block_device *bdgrab(struct block_device *bdev);
2379 extern void bd_set_size(struct block_device *, loff_t size);
2380 extern void bd_forget(struct inode *inode);
2381 extern void bdput(struct block_device *);
2382 extern void invalidate_bdev(struct block_device *);
2383 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2384 extern int sync_blockdev(struct block_device *bdev);
2385 extern void kill_bdev(struct block_device *);
2386 extern struct super_block *freeze_bdev(struct block_device *);
2387 extern void emergency_thaw_all(void);
2388 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2389 extern int fsync_bdev(struct block_device *);
2390 
2391 extern struct super_block *blockdev_superblock;
2392 
2393 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2394 {
2395         return sb == blockdev_superblock;
2396 }
2397 #else
2398 static inline void bd_forget(struct inode *inode) {}
2399 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2400 static inline void kill_bdev(struct block_device *bdev) {}
2401 static inline void invalidate_bdev(struct block_device *bdev) {}
2402 
2403 static inline struct super_block *freeze_bdev(struct block_device *sb)
2404 {
2405         return NULL;
2406 }
2407 
2408 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2409 {
2410         return 0;
2411 }
2412 
2413 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2414 {
2415 }
2416 
2417 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2418 {
2419         return false;
2420 }
2421 #endif
2422 extern int sync_filesystem(struct super_block *);
2423 extern const struct file_operations def_blk_fops;
2424 extern const struct file_operations def_chr_fops;
2425 #ifdef CONFIG_BLOCK
2426 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2427 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2428 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2429 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2430 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2431                                                void *holder);
2432 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2433                                               void *holder);
2434 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2435 extern int __blkdev_reread_part(struct block_device *bdev);
2436 extern int blkdev_reread_part(struct block_device *bdev);
2437 
2438 #ifdef CONFIG_SYSFS
2439 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2440 extern void bd_unlink_disk_holder(struct block_device *bdev,
2441                                   struct gendisk *disk);
2442 #else
2443 static inline int bd_link_disk_holder(struct block_device *bdev,
2444                                       struct gendisk *disk)
2445 {
2446         return 0;
2447 }
2448 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2449                                          struct gendisk *disk)
2450 {
2451 }
2452 #endif
2453 #endif
2454 
2455 /* fs/char_dev.c */
2456 #define CHRDEV_MAJOR_HASH_SIZE  255
2457 /* Marks the bottom of the first segment of free char majors */
2458 #define CHRDEV_MAJOR_DYN_END 234
2459 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2460 extern int register_chrdev_region(dev_t, unsigned, const char *);
2461 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2462                              unsigned int count, const char *name,
2463                              const struct file_operations *fops);
2464 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2465                                 unsigned int count, const char *name);
2466 extern void unregister_chrdev_region(dev_t, unsigned);
2467 extern void chrdev_show(struct seq_file *,off_t);
2468 
2469 static inline int register_chrdev(unsigned int major, const char *name,
2470                                   const struct file_operations *fops)
2471 {
2472         return __register_chrdev(major, 0, 256, name, fops);
2473 }
2474 
2475 static inline void unregister_chrdev(unsigned int major, const char *name)
2476 {
2477         __unregister_chrdev(major, 0, 256, name);
2478 }
2479 
2480 /* fs/block_dev.c */
2481 #define BDEVNAME_SIZE   32      /* Largest string for a blockdev identifier */
2482 #define BDEVT_SIZE      10      /* Largest string for MAJ:MIN for blkdev */
2483 
2484 #ifdef CONFIG_BLOCK
2485 #define BLKDEV_MAJOR_HASH_SIZE  255
2486 extern const char *__bdevname(dev_t, char *buffer);
2487 extern const char *bdevname(struct block_device *bdev, char *buffer);
2488 extern struct block_device *lookup_bdev(const char *);
2489 extern void blkdev_show(struct seq_file *,off_t);
2490 
2491 #else
2492 #define BLKDEV_MAJOR_HASH_SIZE  0
2493 #endif
2494 
2495 extern void init_special_inode(struct inode *, umode_t, dev_t);
2496 
2497 /* Invalid inode operations -- fs/bad_inode.c */
2498 extern void make_bad_inode(struct inode *);
2499 extern bool is_bad_inode(struct inode *);
2500 
2501 #ifdef CONFIG_BLOCK
2502 static inline bool op_is_write(unsigned int op)
2503 {
2504         return op == REQ_OP_READ ? false : true;
2505 }
2506 
2507 /*
2508  * return data direction, READ or WRITE
2509  */
2510 static inline int bio_data_dir(struct bio *bio)
2511 {
2512         return op_is_write(bio_op(bio)) ? WRITE : READ;
2513 }
2514 
2515 extern void check_disk_size_change(struct gendisk *disk,
2516                                    struct block_device *bdev);
2517 extern int revalidate_disk(struct gendisk *);
2518 extern int check_disk_change(struct block_device *);
2519 extern int __invalidate_device(struct block_device *, bool);
2520 extern int invalidate_partition(struct gendisk *, int);
2521 #endif
2522 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2523                                         pgoff_t start, pgoff_t end);
2524 
2525 static inline void invalidate_remote_inode(struct inode *inode)
2526 {
2527         if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2528             S_ISLNK(inode->i_mode))
2529                 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2530 }
2531 extern int invalidate_inode_pages2(struct address_space *mapping);
2532 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2533                                          pgoff_t start, pgoff_t end);
2534 extern int write_inode_now(struct inode *, int);
2535 extern int filemap_fdatawrite(struct address_space *);
2536 extern int filemap_flush(struct address_space *);
2537 extern int filemap_fdatawait(struct address_space *);
2538 extern void filemap_fdatawait_keep_errors(struct address_space *);
2539 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2540                                    loff_t lend);
2541 extern int filemap_write_and_wait(struct address_space *mapping);
2542 extern int filemap_write_and_wait_range(struct address_space *mapping,
2543                                         loff_t lstart, loff_t lend);
2544 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2545                                 loff_t start, loff_t end, int sync_mode);
2546 extern int filemap_fdatawrite_range(struct address_space *mapping,
2547                                 loff_t start, loff_t end);
2548 extern int filemap_check_errors(struct address_space *mapping);
2549 
2550 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2551                            int datasync);
2552 extern int vfs_fsync(struct file *file, int datasync);
2553 
2554 /*
2555  * Sync the bytes written if this was a synchronous write.  Expect ki_pos
2556  * to already be updated for the write, and will return either the amount
2557  * of bytes passed in, or an error if syncing the file failed.
2558  */
2559 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2560 {
2561         if (iocb->ki_flags & IOCB_DSYNC) {
2562                 int ret = vfs_fsync_range(iocb->ki_filp,
2563                                 iocb->ki_pos - count, iocb->ki_pos - 1,
2564                                 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2565                 if (ret)
2566                         return ret;
2567         }
2568 
2569         return count;
2570 }
2571 
2572 extern void emergency_sync(void);
2573 extern void emergency_remount(void);
2574 #ifdef CONFIG_BLOCK
2575 extern sector_t bmap(struct inode *, sector_t);
2576 #endif
2577 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2578 extern int inode_permission(struct inode *, int);
2579 extern int __inode_permission(struct inode *, int);
2580 extern int generic_permission(struct inode *, int);
2581 extern int __check_sticky(struct inode *dir, struct inode *inode);
2582 
2583 static inline bool execute_ok(struct inode *inode)
2584 {
2585         return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2586 }
2587 
2588 static inline void file_start_write(struct file *file)
2589 {
2590         if (!S_ISREG(file_inode(file)->i_mode))
2591                 return;
2592         __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2593 }
2594 
2595 static inline bool file_start_write_trylock(struct file *file)
2596 {
2597         if (!S_ISREG(file_inode(file)->i_mode))
2598                 return true;
2599         return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2600 }
2601 
2602 static inline void file_end_write(struct file *file)
2603 {
2604         if (!S_ISREG(file_inode(file)->i_mode))
2605                 return;
2606         __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2607 }
2608 
2609 /*
2610  * get_write_access() gets write permission for a file.
2611  * put_write_access() releases this write permission.
2612  * This is used for regular files.
2613  * We cannot support write (and maybe mmap read-write shared) accesses and
2614  * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2615  * can have the following values:
2616  * 0: no writers, no VM_DENYWRITE mappings
2617  * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2618  * > 0: (i_writecount) users are writing to the file.
2619  *
2620  * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2621  * except for the cases where we don't hold i_writecount yet. Then we need to
2622  * use {get,deny}_write_access() - these functions check the sign and refuse
2623  * to do the change if sign is wrong.
2624  */
2625 static inline int get_write_access(struct inode *inode)
2626 {
2627         return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2628 }
2629 static inline int deny_write_access(struct file *file)
2630 {
2631         struct inode *inode = file_inode(file);
2632         return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2633 }
2634 static inline void put_write_access(struct inode * inode)
2635 {
2636         atomic_dec(&inode->i_writecount);
2637 }
2638 static inline void allow_write_access(struct file *file)
2639 {
2640         if (file)
2641                 atomic_inc(&file_inode(file)->i_writecount);
2642 }
2643 static inline bool inode_is_open_for_write(const struct inode *inode)
2644 {
2645         return atomic_read(&inode->i_writecount) > 0;
2646 }
2647 
2648 #ifdef CONFIG_IMA
2649 static inline void i_readcount_dec(struct inode *inode)
2650 {
2651         BUG_ON(!atomic_read(&inode->i_readcount));
2652         atomic_dec(&inode->i_readcount);
2653 }
2654 static inline void i_readcount_inc(struct inode *inode)
2655 {
2656         atomic_inc(&inode->i_readcount);
2657 }
2658 #else
2659 static inline void i_readcount_dec(struct inode *inode)
2660 {
2661         return;
2662 }
2663 static inline void i_readcount_inc(struct inode *inode)
2664 {
2665         return;
2666 }
2667 #endif
2668 extern int do_pipe_flags(int *, int);
2669 
2670 #define __kernel_read_file_id(id) \
2671         id(UNKNOWN, unknown)            \
2672         id(FIRMWARE, firmware)          \
2673         id(FIRMWARE_PREALLOC_BUFFER, firmware)  \
2674         id(MODULE, kernel-module)               \
2675         id(KEXEC_IMAGE, kexec-image)            \
2676         id(KEXEC_INITRAMFS, kexec-initramfs)    \
2677         id(POLICY, security-policy)             \
2678         id(MAX_ID, )
2679 
2680 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2681 #define __fid_stringify(dummy, str) #str,
2682 
2683 enum kernel_read_file_id {
2684         __kernel_read_file_id(__fid_enumify)
2685 };
2686 
2687 static const char * const kernel_read_file_str[] = {
2688         __kernel_read_file_id(__fid_stringify)
2689 };
2690 
2691 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id)
2692 {
2693         if (id < 0 || id >= READING_MAX_ID)
2694                 return kernel_read_file_str[READING_UNKNOWN];
2695 
2696         return kernel_read_file_str[id];
2697 }
2698 
2699 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2700 extern int kernel_read_file(struct file *, void **, loff_t *, loff_t,
2701                             enum kernel_read_file_id);
2702 extern int kernel_read_file_from_path(char *, void **, loff_t *, loff_t,
2703                                       enum kernel_read_file_id);
2704 extern int kernel_read_file_from_fd(int, void **, loff_t *, loff_t,
2705                                     enum kernel_read_file_id);
2706 extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
2707 extern ssize_t __kernel_write(struct file *, const char *, size_t, loff_t *);
2708 extern struct file * open_exec(const char *);
2709  
2710 /* fs/dcache.c -- generic fs support functions */
2711 extern bool is_subdir(struct dentry *, struct dentry *);
2712 extern bool path_is_under(struct path *, struct path *);
2713 
2714 extern char *file_path(struct file *, char *, int);
2715 
2716 #include <linux/err.h>
2717 
2718 /* needed for stackable file system support */
2719 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2720 
2721 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2722 
2723 extern int inode_init_always(struct super_block *, struct inode *);
2724 extern void inode_init_once(struct inode *);
2725 extern void address_space_init_once(struct address_space *mapping);
2726 extern struct inode * igrab(struct inode *);
2727 extern ino_t iunique(struct super_block *, ino_t);
2728 extern int inode_needs_sync(struct inode *inode);
2729 extern int generic_delete_inode(struct inode *inode);
2730 static inline int generic_drop_inode(struct inode *inode)
2731 {
2732         return !inode->i_nlink || inode_unhashed(inode);
2733 }
2734 
2735 extern struct inode *ilookup5_nowait(struct super_block *sb,
2736                 unsigned long hashval, int (*test)(struct inode *, void *),
2737                 void *data);
2738 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2739                 int (*test)(struct inode *, void *), void *data);
2740 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2741 
2742 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2743 extern struct inode * iget_locked(struct super_block *, unsigned long);
2744 extern struct inode *find_inode_nowait(struct super_block *,
2745                                        unsigned long,
2746                                        int (*match)(struct inode *,
2747                                                     unsigned long, void *),
2748                                        void *data);
2749 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2750 extern int insert_inode_locked(struct inode *);
2751 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2752 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2753 #else
2754 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2755 #endif
2756 extern void unlock_new_inode(struct inode *);
2757 extern unsigned int get_next_ino(void);
2758 
2759 extern void __iget(struct inode * inode);
2760 extern void iget_failed(struct inode *);
2761 extern void clear_inode(struct inode *);
2762 extern void __destroy_inode(struct inode *);
2763 extern struct inode *new_inode_pseudo(struct super_block *sb);
2764 extern struct inode *new_inode(struct super_block *sb);
2765 extern void free_inode_nonrcu(struct inode *inode);
2766 extern int should_remove_suid(struct dentry *);
2767 extern int file_remove_privs(struct file *);
2768 
2769 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2770 static inline void insert_inode_hash(struct inode *inode)
2771 {
2772         __insert_inode_hash(inode, inode->i_ino);
2773 }
2774 
2775 extern void __remove_inode_hash(struct inode *);
2776 static inline void remove_inode_hash(struct inode *inode)
2777 {
2778         if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
2779                 __remove_inode_hash(inode);
2780 }
2781 
2782 extern void inode_sb_list_add(struct inode *inode);
2783 
2784 #ifdef CONFIG_BLOCK
2785 extern blk_qc_t submit_bio(struct bio *);
2786 extern int bdev_read_only(struct block_device *);
2787 #endif
2788 extern int set_blocksize(struct block_device *, int);
2789 extern int sb_set_blocksize(struct super_block *, int);
2790 extern int sb_min_blocksize(struct super_block *, int);
2791 
2792 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2793 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2794 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2795 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2796 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2797 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2798 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
2799 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2800 
2801 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos);
2802 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos);
2803 
2804 /* fs/block_dev.c */
2805 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
2806 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2807 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2808                         int datasync);
2809 extern void block_sync_page(struct page *page);
2810 
2811 /* fs/splice.c */
2812 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2813                 struct pipe_inode_info *, size_t, unsigned int);
2814 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2815                 struct file *, loff_t *, size_t, unsigned int);
2816 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2817                 struct file *out, loff_t *, size_t len, unsigned int flags);
2818 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2819                 loff_t *opos, size_t len, unsigned int flags);
2820 
2821 
2822 extern void
2823 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2824 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2825 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2826 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
2827 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2828 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2829                 int whence, loff_t maxsize, loff_t eof);
2830 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2831                 int whence, loff_t size);
2832 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
2833 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
2834 extern int generic_file_open(struct inode * inode, struct file * filp);
2835 extern int nonseekable_open(struct inode * inode, struct file * filp);
2836 
2837 #ifdef CONFIG_BLOCK
2838 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
2839                             loff_t file_offset);
2840 
2841 enum {
2842         /* need locking between buffered and direct access */
2843         DIO_LOCKING     = 0x01,
2844 
2845         /* filesystem does not support filling holes */
2846         DIO_SKIP_HOLES  = 0x02,
2847 
2848         /* filesystem can handle aio writes beyond i_size */
2849         DIO_ASYNC_EXTEND = 0x04,
2850 
2851         /* inode/fs/bdev does not need truncate protection */
2852         DIO_SKIP_DIO_COUNT = 0x08,
2853 };
2854 
2855 void dio_end_io(struct bio *bio, int error);
2856 
2857 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
2858                              struct block_device *bdev, struct iov_iter *iter,
2859                              get_block_t get_block,
2860                              dio_iodone_t end_io, dio_submit_t submit_io,
2861                              int flags);
2862 
2863 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
2864                                          struct inode *inode,
2865                                          struct iov_iter *iter,
2866                                          get_block_t get_block)
2867 {
2868         return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
2869                         get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
2870 }
2871 #endif
2872 
2873 void inode_dio_wait(struct inode *inode);
2874 
2875 /*
2876  * inode_dio_begin - signal start of a direct I/O requests
2877  * @inode: inode the direct I/O happens on
2878  *
2879  * This is called once we've finished processing a direct I/O request,
2880  * and is used to wake up callers waiting for direct I/O to be quiesced.
2881  */
2882 static inline void inode_dio_begin(struct inode *inode)
2883 {
2884         atomic_inc(&inode->i_dio_count);
2885 }
2886 
2887 /*
2888  * inode_dio_end - signal finish of a direct I/O requests
2889  * @inode: inode the direct I/O happens on
2890  *
2891  * This is called once we've finished processing a direct I/O request,
2892  * and is used to wake up callers waiting for direct I/O to be quiesced.
2893  */
2894 static inline void inode_dio_end(struct inode *inode)
2895 {
2896         if (atomic_dec_and_test(&inode->i_dio_count))
2897                 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
2898 }
2899 
2900 extern void inode_set_flags(struct inode *inode, unsigned int flags,
2901                             unsigned int mask);
2902 
2903 extern const struct file_operations generic_ro_fops;
2904 
2905 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2906 
2907 extern int readlink_copy(char __user *, int, const char *);
2908 extern int page_readlink(struct dentry *, char __user *, int);
2909 extern const char *page_get_link(struct dentry *, struct inode *,
2910                                  struct delayed_call *);
2911 extern void page_put_link(void *);
2912 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2913                 int nofs);
2914 extern int page_symlink(struct inode *inode, const char *symname, int len);
2915 extern const struct inode_operations page_symlink_inode_operations;
2916 extern void kfree_link(void *);
2917 extern int generic_readlink(struct dentry *, char __user *, int);
2918 extern void generic_fillattr(struct inode *, struct kstat *);
2919 int vfs_getattr_nosec(struct path *path, struct kstat *stat);
2920 extern int vfs_getattr(struct path *, struct kstat *);
2921 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2922 void inode_add_bytes(struct inode *inode, loff_t bytes);
2923 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
2924 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2925 loff_t inode_get_bytes(struct inode *inode);
2926 void inode_set_bytes(struct inode *inode, loff_t bytes);
2927 const char *simple_get_link(struct dentry *, struct inode *,
2928                             struct delayed_call *);
2929 extern const struct inode_operations simple_symlink_inode_operations;
2930 
2931 extern int iterate_dir(struct file *, struct dir_context *);
2932 
2933 extern int vfs_stat(const char __user *, struct kstat *);
2934 extern int vfs_lstat(const char __user *, struct kstat *);
2935 extern int vfs_fstat(unsigned int, struct kstat *);
2936 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2937 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
2938 
2939 extern int __generic_block_fiemap(struct inode *inode,
2940                                   struct fiemap_extent_info *fieinfo,
2941                                   loff_t start, loff_t len,
2942                                   get_block_t *get_block);
2943 extern int generic_block_fiemap(struct inode *inode,
2944                                 struct fiemap_extent_info *fieinfo, u64 start,
2945                                 u64 len, get_block_t *get_block);
2946 
2947 extern void get_filesystem(struct file_system_type *fs);
2948 extern void put_filesystem(struct file_system_type *fs);
2949 extern struct file_system_type *get_fs_type(const char *name);
2950 extern struct super_block *get_super(struct block_device *);
2951 extern struct super_block *get_super_thawed(struct block_device *);
2952 extern struct super_block *get_active_super(struct block_device *bdev);
2953 extern void drop_super(struct super_block *sb);
2954 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2955 extern void iterate_supers_type(struct file_system_type *,
2956                                 void (*)(struct super_block *, void *), void *);
2957 
2958 extern int dcache_dir_open(struct inode *, struct file *);
2959 extern int dcache_dir_close(struct inode *, struct file *);
2960 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2961 extern int dcache_readdir(struct file *, struct dir_context *);
2962 extern int simple_setattr(struct dentry *, struct iattr *);
2963 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2964 extern int simple_statfs(struct dentry *, struct kstatfs *);
2965 extern int simple_open(struct inode *inode, struct file *file);
2966 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2967 extern int simple_unlink(struct inode *, struct dentry *);
2968 extern int simple_rmdir(struct inode *, struct dentry *);
2969 extern int simple_rename(struct inode *, struct dentry *,
2970                          struct inode *, struct dentry *, unsigned int);
2971 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2972 extern int simple_empty(struct dentry *);
2973 extern int simple_readpage(struct file *file, struct page *page);
2974 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2975                         loff_t pos, unsigned len, unsigned flags,
2976                         struct page **pagep, void **fsdata);
2977 extern int simple_write_end(struct file *file, struct address_space *mapping,
2978                         loff_t pos, unsigned len, unsigned copied,
2979                         struct page *page, void *fsdata);
2980 extern int always_delete_dentry(const struct dentry *);
2981 extern struct inode *alloc_anon_inode(struct super_block *);
2982 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
2983 extern const struct dentry_operations simple_dentry_operations;
2984 
2985 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
2986 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2987 extern const struct file_operations simple_dir_operations;
2988 extern const struct inode_operations simple_dir_inode_operations;
2989 extern void make_empty_dir_inode(struct inode *inode);
2990 extern bool is_empty_dir_inode(struct inode *inode);
2991 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2992 struct dentry *d_alloc_name(struct dentry *, const char *);
2993 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2994 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2995 extern void simple_release_fs(struct vfsmount **mount, int *count);
2996 
2997 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2998                         loff_t *ppos, const void *from, size_t available);
2999 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3000                 const void __user *from, size_t count);
3001 
3002 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3003 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3004 
3005 extern int generic_check_addressable(unsigned, u64);
3006 
3007 #ifdef CONFIG_MIGRATION
3008 extern int buffer_migrate_page(struct address_space *,
3009                                 struct page *, struct page *,
3010                                 enum migrate_mode);
3011 #else
3012 #define buffer_migrate_page NULL
3013 #endif
3014 
3015 extern int setattr_prepare(struct dentry *, struct iattr *);
3016 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3017 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
3018 
3019 extern int file_update_time(struct file *file);
3020 
3021 extern int generic_show_options(struct seq_file *m, struct dentry *root);
3022 extern void save_mount_options(struct super_block *sb, char *options);
3023 extern void replace_mount_options(struct super_block *sb, char *options);
3024 
3025 static inline bool io_is_direct(struct file *filp)
3026 {
3027         return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host);
3028 }
3029 
3030 static inline int iocb_flags(struct file *file)
3031 {
3032         int res = 0;
3033         if (file->f_flags & O_APPEND)
3034                 res |= IOCB_APPEND;
3035         if (io_is_direct(file))
3036                 res |= IOCB_DIRECT;
3037         if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3038                 res |= IOCB_DSYNC;
3039         if (file->f_flags & __O_SYNC)
3040                 res |= IOCB_SYNC;
3041         return res;
3042 }
3043 
3044 static inline ino_t parent_ino(struct dentry *dentry)
3045 {
3046         ino_t res;
3047 
3048         /*
3049          * Don't strictly need d_lock here? If the parent ino could change
3050          * then surely we'd have a deeper race in the caller?
3051          */
3052         spin_lock(&dentry->d_lock);
3053         res = dentry->d_parent->d_inode->i_ino;
3054         spin_unlock(&dentry->d_lock);
3055         return res;
3056 }
3057 
3058 /* Transaction based IO helpers */
3059 
3060 /*
3061  * An argresp is stored in an allocated page and holds the
3062  * size of the argument or response, along with its content
3063  */
3064 struct simple_transaction_argresp {
3065         ssize_t size;
3066         char data[0];
3067 };
3068 
3069 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3070 
3071 char *simple_transaction_get(struct file *file, const char __user *buf,
3072                                 size_t size);
3073 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3074                                 size_t size, loff_t *pos);
3075 int simple_transaction_release(struct inode *inode, struct file *file);
3076 
3077 void simple_transaction_set(struct file *file, size_t n);
3078 
3079 /*
3080  * simple attribute files
3081  *
3082  * These attributes behave similar to those in sysfs:
3083  *
3084  * Writing to an attribute immediately sets a value, an open file can be
3085  * written to multiple times.
3086  *
3087  * Reading from an attribute creates a buffer from the value that might get
3088  * read with multiple read calls. When the attribute has been read
3089  * completely, no further read calls are possible until the file is opened
3090  * again.
3091  *
3092  * All attributes contain a text representation of a numeric value
3093  * that are accessed with the get() and set() functions.
3094  */
3095 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt)            \
3096 static int __fops ## _open(struct inode *inode, struct file *file)      \
3097 {                                                                       \
3098         __simple_attr_check_format(__fmt, 0ull);                        \
3099         return simple_attr_open(inode, file, __get, __set, __fmt);      \
3100 }                                                                       \
3101 static const struct file_operations __fops = {                          \
3102         .owner   = THIS_MODULE,                                         \
3103         .open    = __fops ## _open,                                     \
3104         .release = simple_attr_release,                                 \
3105         .read    = simple_attr_read,                                    \
3106         .write   = simple_attr_write,                                   \
3107         .llseek  = generic_file_llseek,                                 \
3108 }
3109 
3110 static inline __printf(1, 2)
3111 void __simple_attr_check_format(const char *fmt, ...)
3112 {
3113         /* don't do anything, just let the compiler check the arguments; */
3114 }
3115 
3116 int simple_attr_open(struct inode *inode, struct file *file,
3117                      int (*get)(void *, u64 *), int (*set)(void *, u64),
3118                      const char *fmt);
3119 int simple_attr_release(struct inode *inode, struct file *file);
3120 ssize_t simple_attr_read(struct file *file, char __user *buf,
3121                          size_t len, loff_t *ppos);
3122 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3123                           size_t len, loff_t *ppos);
3124 
3125 struct ctl_table;
3126 int proc_nr_files(struct ctl_table *table, int write,
3127                   void __user *buffer, size_t *lenp, loff_t *ppos);
3128 int proc_nr_dentry(struct ctl_table *table, int write,
3129                   void __user *buffer, size_t *lenp, loff_t *ppos);
3130 int proc_nr_inodes(struct ctl_table *table, int write,
3131                    void __user *buffer, size_t *lenp, loff_t *ppos);
3132 int __init get_filesystem_list(char *buf);
3133 
3134 #define __FMODE_EXEC            ((__force int) FMODE_EXEC)
3135 #define __FMODE_NONOTIFY        ((__force int) FMODE_NONOTIFY)
3136 
3137 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3138 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3139                                             (flag & __FMODE_NONOTIFY)))
3140 
3141 static inline bool is_sxid(umode_t mode)
3142 {
3143         return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3144 }
3145 
3146 static inline int check_sticky(struct inode *dir, struct inode *inode)
3147 {
3148         if (!(dir->i_mode & S_ISVTX))
3149                 return 0;
3150 
3151         return __check_sticky(dir, inode);
3152 }
3153 
3154 static inline void inode_has_no_xattr(struct inode *inode)
3155 {
3156         if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
3157                 inode->i_flags |= S_NOSEC;
3158 }
3159 
3160 static inline bool is_root_inode(struct inode *inode)
3161 {
3162         return inode == inode->i_sb->s_root->d_inode;
3163 }
3164 
3165 static inline bool dir_emit(struct dir_context *ctx,
3166                             const char *name, int namelen,
3167                             u64 ino, unsigned type)
3168 {
3169         return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3170 }
3171 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3172 {
3173         return ctx->actor(ctx, ".", 1, ctx->pos,
3174                           file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3175 }
3176 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3177 {
3178         return ctx->actor(ctx, "..", 2, ctx->pos,
3179                           parent_ino(file->f_path.dentry), DT_DIR) == 0;
3180 }
3181 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3182 {
3183         if (ctx->pos == 0) {
3184                 if (!dir_emit_dot(file, ctx))
3185                         return false;
3186                 ctx->pos = 1;
3187         }
3188         if (ctx->pos == 1) {
3189                 if (!dir_emit_dotdot(file, ctx))
3190                         return false;
3191                 ctx->pos = 2;
3192         }
3193         return true;
3194 }
3195 static inline bool dir_relax(struct inode *inode)
3196 {
3197         inode_unlock(inode);
3198         inode_lock(inode);
3199         return !IS_DEADDIR(inode);
3200 }
3201 
3202 static inline bool dir_relax_shared(struct inode *inode)
3203 {
3204         inode_unlock_shared(inode);
3205         inode_lock_shared(inode);
3206         return !IS_DEADDIR(inode);
3207 }
3208 
3209 extern bool path_noexec(const struct path *path);
3210 extern void inode_nohighmem(struct inode *inode);
3211 
3212 #endif /* _LINUX_FS_H */
3213 

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