]> git.proxmox.com Git - mirror_ubuntu-jammy-kernel.git/blob - include/linux/fs.h
switch generic_file_splice_read() to use of ->read_iter()
[mirror_ubuntu-jammy-kernel.git] / 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 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
228 /*
229 * Whiteout is represented by a char device. The following constants define the
230 * mode and device number to use.
231 */
232 #define WHITEOUT_MODE 0
233 #define WHITEOUT_DEV 0
234
235 /*
236 * This is the Inode Attributes structure, used for notify_change(). It
237 * uses the above definitions as flags, to know which values have changed.
238 * Also, in this manner, a Filesystem can look at only the values it cares
239 * about. Basically, these are the attributes that the VFS layer can
240 * request to change from the FS layer.
241 *
242 * Derek Atkins <warlord@MIT.EDU> 94-10-20
243 */
244 struct iattr {
245 unsigned int ia_valid;
246 umode_t ia_mode;
247 kuid_t ia_uid;
248 kgid_t ia_gid;
249 loff_t ia_size;
250 struct timespec ia_atime;
251 struct timespec ia_mtime;
252 struct timespec ia_ctime;
253
254 /*
255 * Not an attribute, but an auxiliary info for filesystems wanting to
256 * implement an ftruncate() like method. NOTE: filesystem should
257 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
258 */
259 struct file *ia_file;
260 };
261
262 /*
263 * Includes for diskquotas.
264 */
265 #include <linux/quota.h>
266
267 /*
268 * Maximum number of layers of fs stack. Needs to be limited to
269 * prevent kernel stack overflow
270 */
271 #define FILESYSTEM_MAX_STACK_DEPTH 2
272
273 /**
274 * enum positive_aop_returns - aop return codes with specific semantics
275 *
276 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
277 * completed, that the page is still locked, and
278 * should be considered active. The VM uses this hint
279 * to return the page to the active list -- it won't
280 * be a candidate for writeback again in the near
281 * future. Other callers must be careful to unlock
282 * the page if they get this return. Returned by
283 * writepage();
284 *
285 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
286 * unlocked it and the page might have been truncated.
287 * The caller should back up to acquiring a new page and
288 * trying again. The aop will be taking reasonable
289 * precautions not to livelock. If the caller held a page
290 * reference, it should drop it before retrying. Returned
291 * by readpage().
292 *
293 * address_space_operation functions return these large constants to indicate
294 * special semantics to the caller. These are much larger than the bytes in a
295 * page to allow for functions that return the number of bytes operated on in a
296 * given page.
297 */
298
299 enum positive_aop_returns {
300 AOP_WRITEPAGE_ACTIVATE = 0x80000,
301 AOP_TRUNCATED_PAGE = 0x80001,
302 };
303
304 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
305 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
306 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
307 * helper code (eg buffer layer)
308 * to clear GFP_FS from alloc */
309
310 /*
311 * oh the beauties of C type declarations.
312 */
313 struct page;
314 struct address_space;
315 struct writeback_control;
316
317 #define IOCB_EVENTFD (1 << 0)
318 #define IOCB_APPEND (1 << 1)
319 #define IOCB_DIRECT (1 << 2)
320 #define IOCB_HIPRI (1 << 3)
321 #define IOCB_DSYNC (1 << 4)
322 #define IOCB_SYNC (1 << 5)
323
324 struct kiocb {
325 struct file *ki_filp;
326 loff_t ki_pos;
327 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
328 void *private;
329 int ki_flags;
330 };
331
332 static inline bool is_sync_kiocb(struct kiocb *kiocb)
333 {
334 return kiocb->ki_complete == NULL;
335 }
336
337 static inline int iocb_flags(struct file *file);
338
339 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
340 {
341 *kiocb = (struct kiocb) {
342 .ki_filp = filp,
343 .ki_flags = iocb_flags(filp),
344 };
345 }
346
347 /*
348 * "descriptor" for what we're up to with a read.
349 * This allows us to use the same read code yet
350 * have multiple different users of the data that
351 * we read from a file.
352 *
353 * The simplest case just copies the data to user
354 * mode.
355 */
356 typedef struct {
357 size_t written;
358 size_t count;
359 union {
360 char __user *buf;
361 void *data;
362 } arg;
363 int error;
364 } read_descriptor_t;
365
366 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
367 unsigned long, unsigned long);
368
369 struct address_space_operations {
370 int (*writepage)(struct page *page, struct writeback_control *wbc);
371 int (*readpage)(struct file *, struct page *);
372
373 /* Write back some dirty pages from this mapping. */
374 int (*writepages)(struct address_space *, struct writeback_control *);
375
376 /* Set a page dirty. Return true if this dirtied it */
377 int (*set_page_dirty)(struct page *page);
378
379 int (*readpages)(struct file *filp, struct address_space *mapping,
380 struct list_head *pages, unsigned nr_pages);
381
382 int (*write_begin)(struct file *, struct address_space *mapping,
383 loff_t pos, unsigned len, unsigned flags,
384 struct page **pagep, void **fsdata);
385 int (*write_end)(struct file *, struct address_space *mapping,
386 loff_t pos, unsigned len, unsigned copied,
387 struct page *page, void *fsdata);
388
389 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
390 sector_t (*bmap)(struct address_space *, sector_t);
391 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
392 int (*releasepage) (struct page *, gfp_t);
393 void (*freepage)(struct page *);
394 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
395 /*
396 * migrate the contents of a page to the specified target. If
397 * migrate_mode is MIGRATE_ASYNC, it must not block.
398 */
399 int (*migratepage) (struct address_space *,
400 struct page *, struct page *, enum migrate_mode);
401 bool (*isolate_page)(struct page *, isolate_mode_t);
402 void (*putback_page)(struct page *);
403 int (*launder_page) (struct page *);
404 int (*is_partially_uptodate) (struct page *, unsigned long,
405 unsigned long);
406 void (*is_dirty_writeback) (struct page *, bool *, bool *);
407 int (*error_remove_page)(struct address_space *, struct page *);
408
409 /* swapfile support */
410 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
411 sector_t *span);
412 void (*swap_deactivate)(struct file *file);
413 };
414
415 extern const struct address_space_operations empty_aops;
416
417 /*
418 * pagecache_write_begin/pagecache_write_end must be used by general code
419 * to write into the pagecache.
420 */
421 int pagecache_write_begin(struct file *, struct address_space *mapping,
422 loff_t pos, unsigned len, unsigned flags,
423 struct page **pagep, void **fsdata);
424
425 int pagecache_write_end(struct file *, struct address_space *mapping,
426 loff_t pos, unsigned len, unsigned copied,
427 struct page *page, void *fsdata);
428
429 struct address_space {
430 struct inode *host; /* owner: inode, block_device */
431 struct radix_tree_root page_tree; /* radix tree of all pages */
432 spinlock_t tree_lock; /* and lock protecting it */
433 atomic_t i_mmap_writable;/* count VM_SHARED mappings */
434 struct rb_root i_mmap; /* tree of private and shared mappings */
435 struct rw_semaphore i_mmap_rwsem; /* protect tree, count, list */
436 /* Protected by tree_lock together with the radix tree */
437 unsigned long nrpages; /* number of total pages */
438 /* number of shadow or DAX exceptional entries */
439 unsigned long nrexceptional;
440 pgoff_t writeback_index;/* writeback starts here */
441 const struct address_space_operations *a_ops; /* methods */
442 unsigned long flags; /* error bits/gfp mask */
443 spinlock_t private_lock; /* for use by the address_space */
444 struct list_head private_list; /* ditto */
445 void *private_data; /* ditto */
446 } __attribute__((aligned(sizeof(long))));
447 /*
448 * On most architectures that alignment is already the case; but
449 * must be enforced here for CRIS, to let the least significant bit
450 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
451 */
452 struct request_queue;
453
454 struct block_device {
455 dev_t bd_dev; /* not a kdev_t - it's a search key */
456 int bd_openers;
457 struct inode * bd_inode; /* will die */
458 struct super_block * bd_super;
459 struct mutex bd_mutex; /* open/close mutex */
460 void * bd_claiming;
461 void * bd_holder;
462 int bd_holders;
463 bool bd_write_holder;
464 #ifdef CONFIG_SYSFS
465 struct list_head bd_holder_disks;
466 #endif
467 struct block_device * bd_contains;
468 unsigned bd_block_size;
469 struct hd_struct * bd_part;
470 /* number of times partitions within this device have been opened. */
471 unsigned bd_part_count;
472 int bd_invalidated;
473 struct gendisk * bd_disk;
474 struct request_queue * bd_queue;
475 struct list_head bd_list;
476 /*
477 * Private data. You must have bd_claim'ed the block_device
478 * to use this. NOTE: bd_claim allows an owner to claim
479 * the same device multiple times, the owner must take special
480 * care to not mess up bd_private for that case.
481 */
482 unsigned long bd_private;
483
484 /* The counter of freeze processes */
485 int bd_fsfreeze_count;
486 /* Mutex for freeze */
487 struct mutex bd_fsfreeze_mutex;
488 };
489
490 /*
491 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
492 * radix trees
493 */
494 #define PAGECACHE_TAG_DIRTY 0
495 #define PAGECACHE_TAG_WRITEBACK 1
496 #define PAGECACHE_TAG_TOWRITE 2
497
498 int mapping_tagged(struct address_space *mapping, int tag);
499
500 static inline void i_mmap_lock_write(struct address_space *mapping)
501 {
502 down_write(&mapping->i_mmap_rwsem);
503 }
504
505 static inline void i_mmap_unlock_write(struct address_space *mapping)
506 {
507 up_write(&mapping->i_mmap_rwsem);
508 }
509
510 static inline void i_mmap_lock_read(struct address_space *mapping)
511 {
512 down_read(&mapping->i_mmap_rwsem);
513 }
514
515 static inline void i_mmap_unlock_read(struct address_space *mapping)
516 {
517 up_read(&mapping->i_mmap_rwsem);
518 }
519
520 /*
521 * Might pages of this file be mapped into userspace?
522 */
523 static inline int mapping_mapped(struct address_space *mapping)
524 {
525 return !RB_EMPTY_ROOT(&mapping->i_mmap);
526 }
527
528 /*
529 * Might pages of this file have been modified in userspace?
530 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
531 * marks vma as VM_SHARED if it is shared, and the file was opened for
532 * writing i.e. vma may be mprotected writable even if now readonly.
533 *
534 * If i_mmap_writable is negative, no new writable mappings are allowed. You
535 * can only deny writable mappings, if none exists right now.
536 */
537 static inline int mapping_writably_mapped(struct address_space *mapping)
538 {
539 return atomic_read(&mapping->i_mmap_writable) > 0;
540 }
541
542 static inline int mapping_map_writable(struct address_space *mapping)
543 {
544 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
545 0 : -EPERM;
546 }
547
548 static inline void mapping_unmap_writable(struct address_space *mapping)
549 {
550 atomic_dec(&mapping->i_mmap_writable);
551 }
552
553 static inline int mapping_deny_writable(struct address_space *mapping)
554 {
555 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
556 0 : -EBUSY;
557 }
558
559 static inline void mapping_allow_writable(struct address_space *mapping)
560 {
561 atomic_inc(&mapping->i_mmap_writable);
562 }
563
564 /*
565 * Use sequence counter to get consistent i_size on 32-bit processors.
566 */
567 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
568 #include <linux/seqlock.h>
569 #define __NEED_I_SIZE_ORDERED
570 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
571 #else
572 #define i_size_ordered_init(inode) do { } while (0)
573 #endif
574
575 struct posix_acl;
576 #define ACL_NOT_CACHED ((void *)(-1))
577 #define ACL_DONT_CACHE ((void *)(-3))
578
579 static inline struct posix_acl *
580 uncached_acl_sentinel(struct task_struct *task)
581 {
582 return (void *)task + 1;
583 }
584
585 static inline bool
586 is_uncached_acl(struct posix_acl *acl)
587 {
588 return (long)acl & 1;
589 }
590
591 #define IOP_FASTPERM 0x0001
592 #define IOP_LOOKUP 0x0002
593 #define IOP_NOFOLLOW 0x0004
594
595 /*
596 * Keep mostly read-only and often accessed (especially for
597 * the RCU path lookup and 'stat' data) fields at the beginning
598 * of the 'struct inode'
599 */
600 struct inode {
601 umode_t i_mode;
602 unsigned short i_opflags;
603 kuid_t i_uid;
604 kgid_t i_gid;
605 unsigned int i_flags;
606
607 #ifdef CONFIG_FS_POSIX_ACL
608 struct posix_acl *i_acl;
609 struct posix_acl *i_default_acl;
610 #endif
611
612 const struct inode_operations *i_op;
613 struct super_block *i_sb;
614 struct address_space *i_mapping;
615
616 #ifdef CONFIG_SECURITY
617 void *i_security;
618 #endif
619
620 /* Stat data, not accessed from path walking */
621 unsigned long i_ino;
622 /*
623 * Filesystems may only read i_nlink directly. They shall use the
624 * following functions for modification:
625 *
626 * (set|clear|inc|drop)_nlink
627 * inode_(inc|dec)_link_count
628 */
629 union {
630 const unsigned int i_nlink;
631 unsigned int __i_nlink;
632 };
633 dev_t i_rdev;
634 loff_t i_size;
635 struct timespec i_atime;
636 struct timespec i_mtime;
637 struct timespec i_ctime;
638 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
639 unsigned short i_bytes;
640 unsigned int i_blkbits;
641 blkcnt_t i_blocks;
642
643 #ifdef __NEED_I_SIZE_ORDERED
644 seqcount_t i_size_seqcount;
645 #endif
646
647 /* Misc */
648 unsigned long i_state;
649 struct rw_semaphore i_rwsem;
650
651 unsigned long dirtied_when; /* jiffies of first dirtying */
652 unsigned long dirtied_time_when;
653
654 struct hlist_node i_hash;
655 struct list_head i_io_list; /* backing dev IO list */
656 #ifdef CONFIG_CGROUP_WRITEBACK
657 struct bdi_writeback *i_wb; /* the associated cgroup wb */
658
659 /* foreign inode detection, see wbc_detach_inode() */
660 int i_wb_frn_winner;
661 u16 i_wb_frn_avg_time;
662 u16 i_wb_frn_history;
663 #endif
664 struct list_head i_lru; /* inode LRU list */
665 struct list_head i_sb_list;
666 struct list_head i_wb_list; /* backing dev writeback list */
667 union {
668 struct hlist_head i_dentry;
669 struct rcu_head i_rcu;
670 };
671 u64 i_version;
672 atomic_t i_count;
673 atomic_t i_dio_count;
674 atomic_t i_writecount;
675 #ifdef CONFIG_IMA
676 atomic_t i_readcount; /* struct files open RO */
677 #endif
678 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
679 struct file_lock_context *i_flctx;
680 struct address_space i_data;
681 struct list_head i_devices;
682 union {
683 struct pipe_inode_info *i_pipe;
684 struct block_device *i_bdev;
685 struct cdev *i_cdev;
686 char *i_link;
687 unsigned i_dir_seq;
688 };
689
690 __u32 i_generation;
691
692 #ifdef CONFIG_FSNOTIFY
693 __u32 i_fsnotify_mask; /* all events this inode cares about */
694 struct hlist_head i_fsnotify_marks;
695 #endif
696
697 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
698 struct fscrypt_info *i_crypt_info;
699 #endif
700
701 void *i_private; /* fs or device private pointer */
702 };
703
704 static inline int inode_unhashed(struct inode *inode)
705 {
706 return hlist_unhashed(&inode->i_hash);
707 }
708
709 /*
710 * inode->i_mutex nesting subclasses for the lock validator:
711 *
712 * 0: the object of the current VFS operation
713 * 1: parent
714 * 2: child/target
715 * 3: xattr
716 * 4: second non-directory
717 * 5: second parent (when locking independent directories in rename)
718 *
719 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
720 * non-directories at once.
721 *
722 * The locking order between these classes is
723 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
724 */
725 enum inode_i_mutex_lock_class
726 {
727 I_MUTEX_NORMAL,
728 I_MUTEX_PARENT,
729 I_MUTEX_CHILD,
730 I_MUTEX_XATTR,
731 I_MUTEX_NONDIR2,
732 I_MUTEX_PARENT2,
733 };
734
735 static inline void inode_lock(struct inode *inode)
736 {
737 down_write(&inode->i_rwsem);
738 }
739
740 static inline void inode_unlock(struct inode *inode)
741 {
742 up_write(&inode->i_rwsem);
743 }
744
745 static inline void inode_lock_shared(struct inode *inode)
746 {
747 down_read(&inode->i_rwsem);
748 }
749
750 static inline void inode_unlock_shared(struct inode *inode)
751 {
752 up_read(&inode->i_rwsem);
753 }
754
755 static inline int inode_trylock(struct inode *inode)
756 {
757 return down_write_trylock(&inode->i_rwsem);
758 }
759
760 static inline int inode_trylock_shared(struct inode *inode)
761 {
762 return down_read_trylock(&inode->i_rwsem);
763 }
764
765 static inline int inode_is_locked(struct inode *inode)
766 {
767 return rwsem_is_locked(&inode->i_rwsem);
768 }
769
770 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
771 {
772 down_write_nested(&inode->i_rwsem, subclass);
773 }
774
775 void lock_two_nondirectories(struct inode *, struct inode*);
776 void unlock_two_nondirectories(struct inode *, struct inode*);
777
778 /*
779 * NOTE: in a 32bit arch with a preemptable kernel and
780 * an UP compile the i_size_read/write must be atomic
781 * with respect to the local cpu (unlike with preempt disabled),
782 * but they don't need to be atomic with respect to other cpus like in
783 * true SMP (so they need either to either locally disable irq around
784 * the read or for example on x86 they can be still implemented as a
785 * cmpxchg8b without the need of the lock prefix). For SMP compiles
786 * and 64bit archs it makes no difference if preempt is enabled or not.
787 */
788 static inline loff_t i_size_read(const struct inode *inode)
789 {
790 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
791 loff_t i_size;
792 unsigned int seq;
793
794 do {
795 seq = read_seqcount_begin(&inode->i_size_seqcount);
796 i_size = inode->i_size;
797 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
798 return i_size;
799 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
800 loff_t i_size;
801
802 preempt_disable();
803 i_size = inode->i_size;
804 preempt_enable();
805 return i_size;
806 #else
807 return inode->i_size;
808 #endif
809 }
810
811 /*
812 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
813 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
814 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
815 */
816 static inline void i_size_write(struct inode *inode, loff_t i_size)
817 {
818 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
819 preempt_disable();
820 write_seqcount_begin(&inode->i_size_seqcount);
821 inode->i_size = i_size;
822 write_seqcount_end(&inode->i_size_seqcount);
823 preempt_enable();
824 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
825 preempt_disable();
826 inode->i_size = i_size;
827 preempt_enable();
828 #else
829 inode->i_size = i_size;
830 #endif
831 }
832
833 static inline unsigned iminor(const struct inode *inode)
834 {
835 return MINOR(inode->i_rdev);
836 }
837
838 static inline unsigned imajor(const struct inode *inode)
839 {
840 return MAJOR(inode->i_rdev);
841 }
842
843 extern struct block_device *I_BDEV(struct inode *inode);
844
845 struct fown_struct {
846 rwlock_t lock; /* protects pid, uid, euid fields */
847 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
848 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
849 kuid_t uid, euid; /* uid/euid of process setting the owner */
850 int signum; /* posix.1b rt signal to be delivered on IO */
851 };
852
853 /*
854 * Track a single file's readahead state
855 */
856 struct file_ra_state {
857 pgoff_t start; /* where readahead started */
858 unsigned int size; /* # of readahead pages */
859 unsigned int async_size; /* do asynchronous readahead when
860 there are only # of pages ahead */
861
862 unsigned int ra_pages; /* Maximum readahead window */
863 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
864 loff_t prev_pos; /* Cache last read() position */
865 };
866
867 /*
868 * Check if @index falls in the readahead windows.
869 */
870 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
871 {
872 return (index >= ra->start &&
873 index < ra->start + ra->size);
874 }
875
876 struct file {
877 union {
878 struct llist_node fu_llist;
879 struct rcu_head fu_rcuhead;
880 } f_u;
881 struct path f_path;
882 struct inode *f_inode; /* cached value */
883 const struct file_operations *f_op;
884
885 /*
886 * Protects f_ep_links, f_flags.
887 * Must not be taken from IRQ context.
888 */
889 spinlock_t f_lock;
890 atomic_long_t f_count;
891 unsigned int f_flags;
892 fmode_t f_mode;
893 struct mutex f_pos_lock;
894 loff_t f_pos;
895 struct fown_struct f_owner;
896 const struct cred *f_cred;
897 struct file_ra_state f_ra;
898
899 u64 f_version;
900 #ifdef CONFIG_SECURITY
901 void *f_security;
902 #endif
903 /* needed for tty driver, and maybe others */
904 void *private_data;
905
906 #ifdef CONFIG_EPOLL
907 /* Used by fs/eventpoll.c to link all the hooks to this file */
908 struct list_head f_ep_links;
909 struct list_head f_tfile_llink;
910 #endif /* #ifdef CONFIG_EPOLL */
911 struct address_space *f_mapping;
912 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
913
914 struct file_handle {
915 __u32 handle_bytes;
916 int handle_type;
917 /* file identifier */
918 unsigned char f_handle[0];
919 };
920
921 static inline struct file *get_file(struct file *f)
922 {
923 atomic_long_inc(&f->f_count);
924 return f;
925 }
926 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
927 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
928 #define file_count(x) atomic_long_read(&(x)->f_count)
929
930 #define MAX_NON_LFS ((1UL<<31) - 1)
931
932 /* Page cache limit. The filesystems should put that into their s_maxbytes
933 limits, otherwise bad things can happen in VM. */
934 #if BITS_PER_LONG==32
935 #define MAX_LFS_FILESIZE (((loff_t)PAGE_SIZE << (BITS_PER_LONG-1))-1)
936 #elif BITS_PER_LONG==64
937 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
938 #endif
939
940 #define FL_POSIX 1
941 #define FL_FLOCK 2
942 #define FL_DELEG 4 /* NFSv4 delegation */
943 #define FL_ACCESS 8 /* not trying to lock, just looking */
944 #define FL_EXISTS 16 /* when unlocking, test for existence */
945 #define FL_LEASE 32 /* lease held on this file */
946 #define FL_CLOSE 64 /* unlock on close */
947 #define FL_SLEEP 128 /* A blocking lock */
948 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
949 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
950 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
951 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
952
953 /*
954 * Special return value from posix_lock_file() and vfs_lock_file() for
955 * asynchronous locking.
956 */
957 #define FILE_LOCK_DEFERRED 1
958
959 /* legacy typedef, should eventually be removed */
960 typedef void *fl_owner_t;
961
962 struct file_lock;
963
964 struct file_lock_operations {
965 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
966 void (*fl_release_private)(struct file_lock *);
967 };
968
969 struct lock_manager_operations {
970 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
971 unsigned long (*lm_owner_key)(struct file_lock *);
972 fl_owner_t (*lm_get_owner)(fl_owner_t);
973 void (*lm_put_owner)(fl_owner_t);
974 void (*lm_notify)(struct file_lock *); /* unblock callback */
975 int (*lm_grant)(struct file_lock *, int);
976 bool (*lm_break)(struct file_lock *);
977 int (*lm_change)(struct file_lock *, int, struct list_head *);
978 void (*lm_setup)(struct file_lock *, void **);
979 };
980
981 struct lock_manager {
982 struct list_head list;
983 /*
984 * NFSv4 and up also want opens blocked during the grace period;
985 * NLM doesn't care:
986 */
987 bool block_opens;
988 };
989
990 struct net;
991 void locks_start_grace(struct net *, struct lock_manager *);
992 void locks_end_grace(struct lock_manager *);
993 int locks_in_grace(struct net *);
994 int opens_in_grace(struct net *);
995
996 /* that will die - we need it for nfs_lock_info */
997 #include <linux/nfs_fs_i.h>
998
999 /*
1000 * struct file_lock represents a generic "file lock". It's used to represent
1001 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1002 * note that the same struct is used to represent both a request for a lock and
1003 * the lock itself, but the same object is never used for both.
1004 *
1005 * FIXME: should we create a separate "struct lock_request" to help distinguish
1006 * these two uses?
1007 *
1008 * The varous i_flctx lists are ordered by:
1009 *
1010 * 1) lock owner
1011 * 2) lock range start
1012 * 3) lock range end
1013 *
1014 * Obviously, the last two criteria only matter for POSIX locks.
1015 */
1016 struct file_lock {
1017 struct file_lock *fl_next; /* singly linked list for this inode */
1018 struct list_head fl_list; /* link into file_lock_context */
1019 struct hlist_node fl_link; /* node in global lists */
1020 struct list_head fl_block; /* circular list of blocked processes */
1021 fl_owner_t fl_owner;
1022 unsigned int fl_flags;
1023 unsigned char fl_type;
1024 unsigned int fl_pid;
1025 int fl_link_cpu; /* what cpu's list is this on? */
1026 struct pid *fl_nspid;
1027 wait_queue_head_t fl_wait;
1028 struct file *fl_file;
1029 loff_t fl_start;
1030 loff_t fl_end;
1031
1032 struct fasync_struct * fl_fasync; /* for lease break notifications */
1033 /* for lease breaks: */
1034 unsigned long fl_break_time;
1035 unsigned long fl_downgrade_time;
1036
1037 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1038 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1039 union {
1040 struct nfs_lock_info nfs_fl;
1041 struct nfs4_lock_info nfs4_fl;
1042 struct {
1043 struct list_head link; /* link in AFS vnode's pending_locks list */
1044 int state; /* state of grant or error if -ve */
1045 } afs;
1046 } fl_u;
1047 };
1048
1049 struct file_lock_context {
1050 spinlock_t flc_lock;
1051 struct list_head flc_flock;
1052 struct list_head flc_posix;
1053 struct list_head flc_lease;
1054 };
1055
1056 /* The following constant reflects the upper bound of the file/locking space */
1057 #ifndef OFFSET_MAX
1058 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1059 #define OFFSET_MAX INT_LIMIT(loff_t)
1060 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1061 #endif
1062
1063 #include <linux/fcntl.h>
1064
1065 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1066
1067 #ifdef CONFIG_FILE_LOCKING
1068 extern int fcntl_getlk(struct file *, unsigned int, struct flock __user *);
1069 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1070 struct flock __user *);
1071
1072 #if BITS_PER_LONG == 32
1073 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 __user *);
1074 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1075 struct flock64 __user *);
1076 #endif
1077
1078 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1079 extern int fcntl_getlease(struct file *filp);
1080
1081 /* fs/locks.c */
1082 void locks_free_lock_context(struct inode *inode);
1083 void locks_free_lock(struct file_lock *fl);
1084 extern void locks_init_lock(struct file_lock *);
1085 extern struct file_lock * locks_alloc_lock(void);
1086 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1087 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1088 extern void locks_remove_posix(struct file *, fl_owner_t);
1089 extern void locks_remove_file(struct file *);
1090 extern void locks_release_private(struct file_lock *);
1091 extern void posix_test_lock(struct file *, struct file_lock *);
1092 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1093 extern int posix_unblock_lock(struct file_lock *);
1094 extern int vfs_test_lock(struct file *, struct file_lock *);
1095 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1096 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1097 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1098 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1099 extern void lease_get_mtime(struct inode *, struct timespec *time);
1100 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1101 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1102 extern int lease_modify(struct file_lock *, int, struct list_head *);
1103 struct files_struct;
1104 extern void show_fd_locks(struct seq_file *f,
1105 struct file *filp, struct files_struct *files);
1106 #else /* !CONFIG_FILE_LOCKING */
1107 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1108 struct flock __user *user)
1109 {
1110 return -EINVAL;
1111 }
1112
1113 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1114 unsigned int cmd, struct flock __user *user)
1115 {
1116 return -EACCES;
1117 }
1118
1119 #if BITS_PER_LONG == 32
1120 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1121 struct flock64 __user *user)
1122 {
1123 return -EINVAL;
1124 }
1125
1126 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1127 unsigned int cmd, struct flock64 __user *user)
1128 {
1129 return -EACCES;
1130 }
1131 #endif
1132 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1133 {
1134 return -EINVAL;
1135 }
1136
1137 static inline int fcntl_getlease(struct file *filp)
1138 {
1139 return F_UNLCK;
1140 }
1141
1142 static inline void
1143 locks_free_lock_context(struct inode *inode)
1144 {
1145 }
1146
1147 static inline void locks_init_lock(struct file_lock *fl)
1148 {
1149 return;
1150 }
1151
1152 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1153 {
1154 return;
1155 }
1156
1157 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1158 {
1159 return;
1160 }
1161
1162 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1163 {
1164 return;
1165 }
1166
1167 static inline void locks_remove_file(struct file *filp)
1168 {
1169 return;
1170 }
1171
1172 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1173 {
1174 return;
1175 }
1176
1177 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1178 struct file_lock *conflock)
1179 {
1180 return -ENOLCK;
1181 }
1182
1183 static inline int posix_unblock_lock(struct file_lock *waiter)
1184 {
1185 return -ENOENT;
1186 }
1187
1188 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1189 {
1190 return 0;
1191 }
1192
1193 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1194 struct file_lock *fl, struct file_lock *conf)
1195 {
1196 return -ENOLCK;
1197 }
1198
1199 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1200 {
1201 return 0;
1202 }
1203
1204 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1205 {
1206 return -ENOLCK;
1207 }
1208
1209 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1210 {
1211 return 0;
1212 }
1213
1214 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1215 {
1216 return;
1217 }
1218
1219 static inline int generic_setlease(struct file *filp, long arg,
1220 struct file_lock **flp, void **priv)
1221 {
1222 return -EINVAL;
1223 }
1224
1225 static inline int vfs_setlease(struct file *filp, long arg,
1226 struct file_lock **lease, void **priv)
1227 {
1228 return -EINVAL;
1229 }
1230
1231 static inline int lease_modify(struct file_lock *fl, int arg,
1232 struct list_head *dispose)
1233 {
1234 return -EINVAL;
1235 }
1236
1237 struct files_struct;
1238 static inline void show_fd_locks(struct seq_file *f,
1239 struct file *filp, struct files_struct *files) {}
1240 #endif /* !CONFIG_FILE_LOCKING */
1241
1242 static inline struct inode *file_inode(const struct file *f)
1243 {
1244 return f->f_inode;
1245 }
1246
1247 static inline struct dentry *file_dentry(const struct file *file)
1248 {
1249 return d_real(file->f_path.dentry, file_inode(file), 0);
1250 }
1251
1252 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1253 {
1254 return locks_lock_inode_wait(file_inode(filp), fl);
1255 }
1256
1257 struct fasync_struct {
1258 spinlock_t fa_lock;
1259 int magic;
1260 int fa_fd;
1261 struct fasync_struct *fa_next; /* singly linked list */
1262 struct file *fa_file;
1263 struct rcu_head fa_rcu;
1264 };
1265
1266 #define FASYNC_MAGIC 0x4601
1267
1268 /* SMP safe fasync helpers: */
1269 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1270 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1271 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1272 extern struct fasync_struct *fasync_alloc(void);
1273 extern void fasync_free(struct fasync_struct *);
1274
1275 /* can be called from interrupts */
1276 extern void kill_fasync(struct fasync_struct **, int, int);
1277
1278 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1279 extern void f_setown(struct file *filp, unsigned long arg, int force);
1280 extern void f_delown(struct file *filp);
1281 extern pid_t f_getown(struct file *filp);
1282 extern int send_sigurg(struct fown_struct *fown);
1283
1284 struct mm_struct;
1285
1286 /*
1287 * Umount options
1288 */
1289
1290 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1291 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1292 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1293 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1294 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1295
1296 /* sb->s_iflags */
1297 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1298 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1299 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1300
1301 /* sb->s_iflags to limit user namespace mounts */
1302 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1303
1304 /* Possible states of 'frozen' field */
1305 enum {
1306 SB_UNFROZEN = 0, /* FS is unfrozen */
1307 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1308 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1309 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1310 * internal threads if needed) */
1311 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1312 };
1313
1314 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1315
1316 struct sb_writers {
1317 int frozen; /* Is sb frozen? */
1318 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1319 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1320 };
1321
1322 struct super_block {
1323 struct list_head s_list; /* Keep this first */
1324 dev_t s_dev; /* search index; _not_ kdev_t */
1325 unsigned char s_blocksize_bits;
1326 unsigned long s_blocksize;
1327 loff_t s_maxbytes; /* Max file size */
1328 struct file_system_type *s_type;
1329 const struct super_operations *s_op;
1330 const struct dquot_operations *dq_op;
1331 const struct quotactl_ops *s_qcop;
1332 const struct export_operations *s_export_op;
1333 unsigned long s_flags;
1334 unsigned long s_iflags; /* internal SB_I_* flags */
1335 unsigned long s_magic;
1336 struct dentry *s_root;
1337 struct rw_semaphore s_umount;
1338 int s_count;
1339 atomic_t s_active;
1340 #ifdef CONFIG_SECURITY
1341 void *s_security;
1342 #endif
1343 const struct xattr_handler **s_xattr;
1344
1345 const struct fscrypt_operations *s_cop;
1346
1347 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1348 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1349 struct block_device *s_bdev;
1350 struct backing_dev_info *s_bdi;
1351 struct mtd_info *s_mtd;
1352 struct hlist_node s_instances;
1353 unsigned int s_quota_types; /* Bitmask of supported quota types */
1354 struct quota_info s_dquot; /* Diskquota specific options */
1355
1356 struct sb_writers s_writers;
1357
1358 char s_id[32]; /* Informational name */
1359 u8 s_uuid[16]; /* UUID */
1360
1361 void *s_fs_info; /* Filesystem private info */
1362 unsigned int s_max_links;
1363 fmode_t s_mode;
1364
1365 /* Granularity of c/m/atime in ns.
1366 Cannot be worse than a second */
1367 u32 s_time_gran;
1368
1369 /*
1370 * The next field is for VFS *only*. No filesystems have any business
1371 * even looking at it. You had been warned.
1372 */
1373 struct mutex s_vfs_rename_mutex; /* Kludge */
1374
1375 /*
1376 * Filesystem subtype. If non-empty the filesystem type field
1377 * in /proc/mounts will be "type.subtype"
1378 */
1379 char *s_subtype;
1380
1381 /*
1382 * Saved mount options for lazy filesystems using
1383 * generic_show_options()
1384 */
1385 char __rcu *s_options;
1386 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1387
1388 /*
1389 * Saved pool identifier for cleancache (-1 means none)
1390 */
1391 int cleancache_poolid;
1392
1393 struct shrinker s_shrink; /* per-sb shrinker handle */
1394
1395 /* Number of inodes with nlink == 0 but still referenced */
1396 atomic_long_t s_remove_count;
1397
1398 /* Being remounted read-only */
1399 int s_readonly_remount;
1400
1401 /* AIO completions deferred from interrupt context */
1402 struct workqueue_struct *s_dio_done_wq;
1403 struct hlist_head s_pins;
1404
1405 /*
1406 * Owning user namespace and default context in which to
1407 * interpret filesystem uids, gids, quotas, device nodes,
1408 * xattrs and security labels.
1409 */
1410 struct user_namespace *s_user_ns;
1411
1412 /*
1413 * Keep the lru lists last in the structure so they always sit on their
1414 * own individual cachelines.
1415 */
1416 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1417 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
1418 struct rcu_head rcu;
1419 struct work_struct destroy_work;
1420
1421 struct mutex s_sync_lock; /* sync serialisation lock */
1422
1423 /*
1424 * Indicates how deep in a filesystem stack this SB is
1425 */
1426 int s_stack_depth;
1427
1428 /* s_inode_list_lock protects s_inodes */
1429 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1430 struct list_head s_inodes; /* all inodes */
1431
1432 spinlock_t s_inode_wblist_lock;
1433 struct list_head s_inodes_wb; /* writeback inodes */
1434 };
1435
1436 /* Helper functions so that in most cases filesystems will
1437 * not need to deal directly with kuid_t and kgid_t and can
1438 * instead deal with the raw numeric values that are stored
1439 * in the filesystem.
1440 */
1441 static inline uid_t i_uid_read(const struct inode *inode)
1442 {
1443 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid);
1444 }
1445
1446 static inline gid_t i_gid_read(const struct inode *inode)
1447 {
1448 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid);
1449 }
1450
1451 static inline void i_uid_write(struct inode *inode, uid_t uid)
1452 {
1453 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid);
1454 }
1455
1456 static inline void i_gid_write(struct inode *inode, gid_t gid)
1457 {
1458 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid);
1459 }
1460
1461 extern struct timespec current_fs_time(struct super_block *sb);
1462
1463 /*
1464 * Snapshotting support.
1465 */
1466
1467 void __sb_end_write(struct super_block *sb, int level);
1468 int __sb_start_write(struct super_block *sb, int level, bool wait);
1469
1470 #define __sb_writers_acquired(sb, lev) \
1471 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1472 #define __sb_writers_release(sb, lev) \
1473 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1474
1475 /**
1476 * sb_end_write - drop write access to a superblock
1477 * @sb: the super we wrote to
1478 *
1479 * Decrement number of writers to the filesystem. Wake up possible waiters
1480 * wanting to freeze the filesystem.
1481 */
1482 static inline void sb_end_write(struct super_block *sb)
1483 {
1484 __sb_end_write(sb, SB_FREEZE_WRITE);
1485 }
1486
1487 /**
1488 * sb_end_pagefault - drop write access to a superblock from a page fault
1489 * @sb: the super we wrote to
1490 *
1491 * Decrement number of processes handling write page fault to the filesystem.
1492 * Wake up possible waiters wanting to freeze the filesystem.
1493 */
1494 static inline void sb_end_pagefault(struct super_block *sb)
1495 {
1496 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1497 }
1498
1499 /**
1500 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1501 * @sb: the super we wrote to
1502 *
1503 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1504 * waiters wanting to freeze the filesystem.
1505 */
1506 static inline void sb_end_intwrite(struct super_block *sb)
1507 {
1508 __sb_end_write(sb, SB_FREEZE_FS);
1509 }
1510
1511 /**
1512 * sb_start_write - get write access to a superblock
1513 * @sb: the super we write to
1514 *
1515 * When a process wants to write data or metadata to a file system (i.e. dirty
1516 * a page or an inode), it should embed the operation in a sb_start_write() -
1517 * sb_end_write() pair to get exclusion against file system freezing. This
1518 * function increments number of writers preventing freezing. If the file
1519 * system is already frozen, the function waits until the file system is
1520 * thawed.
1521 *
1522 * Since freeze protection behaves as a lock, users have to preserve
1523 * ordering of freeze protection and other filesystem locks. Generally,
1524 * freeze protection should be the outermost lock. In particular, we have:
1525 *
1526 * sb_start_write
1527 * -> i_mutex (write path, truncate, directory ops, ...)
1528 * -> s_umount (freeze_super, thaw_super)
1529 */
1530 static inline void sb_start_write(struct super_block *sb)
1531 {
1532 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1533 }
1534
1535 static inline int sb_start_write_trylock(struct super_block *sb)
1536 {
1537 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1538 }
1539
1540 /**
1541 * sb_start_pagefault - get write access to a superblock from a page fault
1542 * @sb: the super we write to
1543 *
1544 * When a process starts handling write page fault, it should embed the
1545 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1546 * exclusion against file system freezing. This is needed since the page fault
1547 * is going to dirty a page. This function increments number of running page
1548 * faults preventing freezing. If the file system is already frozen, the
1549 * function waits until the file system is thawed.
1550 *
1551 * Since page fault freeze protection behaves as a lock, users have to preserve
1552 * ordering of freeze protection and other filesystem locks. It is advised to
1553 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1554 * handling code implies lock dependency:
1555 *
1556 * mmap_sem
1557 * -> sb_start_pagefault
1558 */
1559 static inline void sb_start_pagefault(struct super_block *sb)
1560 {
1561 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1562 }
1563
1564 /*
1565 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1566 * @sb: the super we write to
1567 *
1568 * This is the third level of protection against filesystem freezing. It is
1569 * free for use by a filesystem. The only requirement is that it must rank
1570 * below sb_start_pagefault.
1571 *
1572 * For example filesystem can call sb_start_intwrite() when starting a
1573 * transaction which somewhat eases handling of freezing for internal sources
1574 * of filesystem changes (internal fs threads, discarding preallocation on file
1575 * close, etc.).
1576 */
1577 static inline void sb_start_intwrite(struct super_block *sb)
1578 {
1579 __sb_start_write(sb, SB_FREEZE_FS, true);
1580 }
1581
1582
1583 extern bool inode_owner_or_capable(const struct inode *inode);
1584
1585 /*
1586 * VFS helper functions..
1587 */
1588 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1589 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1590 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1591 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1592 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1593 extern int vfs_rmdir(struct inode *, struct dentry *);
1594 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1595 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1596 extern int vfs_whiteout(struct inode *, struct dentry *);
1597
1598 /*
1599 * VFS file helper functions.
1600 */
1601 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1602 umode_t mode);
1603 extern bool may_open_dev(const struct path *path);
1604 /*
1605 * VFS FS_IOC_FIEMAP helper definitions.
1606 */
1607 struct fiemap_extent_info {
1608 unsigned int fi_flags; /* Flags as passed from user */
1609 unsigned int fi_extents_mapped; /* Number of mapped extents */
1610 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1611 struct fiemap_extent __user *fi_extents_start; /* Start of
1612 fiemap_extent array */
1613 };
1614 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1615 u64 phys, u64 len, u32 flags);
1616 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1617
1618 /*
1619 * File types
1620 *
1621 * NOTE! These match bits 12..15 of stat.st_mode
1622 * (ie "(i_mode >> 12) & 15").
1623 */
1624 #define DT_UNKNOWN 0
1625 #define DT_FIFO 1
1626 #define DT_CHR 2
1627 #define DT_DIR 4
1628 #define DT_BLK 6
1629 #define DT_REG 8
1630 #define DT_LNK 10
1631 #define DT_SOCK 12
1632 #define DT_WHT 14
1633
1634 /*
1635 * This is the "filldir" function type, used by readdir() to let
1636 * the kernel specify what kind of dirent layout it wants to have.
1637 * This allows the kernel to read directories into kernel space or
1638 * to have different dirent layouts depending on the binary type.
1639 */
1640 struct dir_context;
1641 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1642 unsigned);
1643
1644 struct dir_context {
1645 const filldir_t actor;
1646 loff_t pos;
1647 };
1648
1649 struct block_device_operations;
1650
1651 /* These macros are for out of kernel modules to test that
1652 * the kernel supports the unlocked_ioctl and compat_ioctl
1653 * fields in struct file_operations. */
1654 #define HAVE_COMPAT_IOCTL 1
1655 #define HAVE_UNLOCKED_IOCTL 1
1656
1657 /*
1658 * These flags let !MMU mmap() govern direct device mapping vs immediate
1659 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1660 *
1661 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1662 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1663 * NOMMU_MAP_READ: Can be mapped for reading
1664 * NOMMU_MAP_WRITE: Can be mapped for writing
1665 * NOMMU_MAP_EXEC: Can be mapped for execution
1666 */
1667 #define NOMMU_MAP_COPY 0x00000001
1668 #define NOMMU_MAP_DIRECT 0x00000008
1669 #define NOMMU_MAP_READ VM_MAYREAD
1670 #define NOMMU_MAP_WRITE VM_MAYWRITE
1671 #define NOMMU_MAP_EXEC VM_MAYEXEC
1672
1673 #define NOMMU_VMFLAGS \
1674 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1675
1676
1677 struct iov_iter;
1678
1679 struct file_operations {
1680 struct module *owner;
1681 loff_t (*llseek) (struct file *, loff_t, int);
1682 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1683 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1684 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1685 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1686 int (*iterate) (struct file *, struct dir_context *);
1687 int (*iterate_shared) (struct file *, struct dir_context *);
1688 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1689 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1690 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1691 int (*mmap) (struct file *, struct vm_area_struct *);
1692 int (*open) (struct inode *, struct file *);
1693 int (*flush) (struct file *, fl_owner_t id);
1694 int (*release) (struct inode *, struct file *);
1695 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1696 int (*aio_fsync) (struct kiocb *, int datasync);
1697 int (*fasync) (int, struct file *, int);
1698 int (*lock) (struct file *, int, struct file_lock *);
1699 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1700 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1701 int (*check_flags)(int);
1702 int (*flock) (struct file *, int, struct file_lock *);
1703 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1704 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1705 int (*setlease)(struct file *, long, struct file_lock **, void **);
1706 long (*fallocate)(struct file *file, int mode, loff_t offset,
1707 loff_t len);
1708 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1709 #ifndef CONFIG_MMU
1710 unsigned (*mmap_capabilities)(struct file *);
1711 #endif
1712 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1713 loff_t, size_t, unsigned int);
1714 int (*clone_file_range)(struct file *, loff_t, struct file *, loff_t,
1715 u64);
1716 ssize_t (*dedupe_file_range)(struct file *, u64, u64, struct file *,
1717 u64);
1718 };
1719
1720 struct inode_operations {
1721 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1722 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
1723 int (*permission) (struct inode *, int);
1724 struct posix_acl * (*get_acl)(struct inode *, int);
1725
1726 int (*readlink) (struct dentry *, char __user *,int);
1727
1728 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1729 int (*link) (struct dentry *,struct inode *,struct dentry *);
1730 int (*unlink) (struct inode *,struct dentry *);
1731 int (*symlink) (struct inode *,struct dentry *,const char *);
1732 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1733 int (*rmdir) (struct inode *,struct dentry *);
1734 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1735 int (*rename) (struct inode *, struct dentry *,
1736 struct inode *, struct dentry *);
1737 int (*rename2) (struct inode *, struct dentry *,
1738 struct inode *, struct dentry *, unsigned int);
1739 int (*setattr) (struct dentry *, struct iattr *);
1740 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1741 int (*setxattr) (struct dentry *, struct inode *,
1742 const char *, const void *, size_t, int);
1743 ssize_t (*getxattr) (struct dentry *, struct inode *,
1744 const char *, void *, size_t);
1745 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1746 int (*removexattr) (struct dentry *, const char *);
1747 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1748 u64 len);
1749 int (*update_time)(struct inode *, struct timespec *, int);
1750 int (*atomic_open)(struct inode *, struct dentry *,
1751 struct file *, unsigned open_flag,
1752 umode_t create_mode, int *opened);
1753 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1754 int (*set_acl)(struct inode *, struct posix_acl *, int);
1755 } ____cacheline_aligned;
1756
1757 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1758 unsigned long nr_segs, unsigned long fast_segs,
1759 struct iovec *fast_pointer,
1760 struct iovec **ret_pointer);
1761
1762 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1763 extern ssize_t __vfs_write(struct file *, const char __user *, size_t, loff_t *);
1764 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1765 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1766 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1767 unsigned long, loff_t *, int);
1768 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1769 unsigned long, loff_t *, int);
1770 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1771 loff_t, size_t, unsigned int);
1772 extern int vfs_clone_file_range(struct file *file_in, loff_t pos_in,
1773 struct file *file_out, loff_t pos_out, u64 len);
1774 extern int vfs_dedupe_file_range(struct file *file,
1775 struct file_dedupe_range *same);
1776
1777 struct super_operations {
1778 struct inode *(*alloc_inode)(struct super_block *sb);
1779 void (*destroy_inode)(struct inode *);
1780
1781 void (*dirty_inode) (struct inode *, int flags);
1782 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1783 int (*drop_inode) (struct inode *);
1784 void (*evict_inode) (struct inode *);
1785 void (*put_super) (struct super_block *);
1786 int (*sync_fs)(struct super_block *sb, int wait);
1787 int (*freeze_super) (struct super_block *);
1788 int (*freeze_fs) (struct super_block *);
1789 int (*thaw_super) (struct super_block *);
1790 int (*unfreeze_fs) (struct super_block *);
1791 int (*statfs) (struct dentry *, struct kstatfs *);
1792 int (*remount_fs) (struct super_block *, int *, char *);
1793 void (*umount_begin) (struct super_block *);
1794
1795 int (*show_options)(struct seq_file *, struct dentry *);
1796 int (*show_devname)(struct seq_file *, struct dentry *);
1797 int (*show_path)(struct seq_file *, struct dentry *);
1798 int (*show_stats)(struct seq_file *, struct dentry *);
1799 #ifdef CONFIG_QUOTA
1800 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1801 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1802 struct dquot **(*get_dquots)(struct inode *);
1803 #endif
1804 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1805 long (*nr_cached_objects)(struct super_block *,
1806 struct shrink_control *);
1807 long (*free_cached_objects)(struct super_block *,
1808 struct shrink_control *);
1809 };
1810
1811 /*
1812 * Inode flags - they have no relation to superblock flags now
1813 */
1814 #define S_SYNC 1 /* Writes are synced at once */
1815 #define S_NOATIME 2 /* Do not update access times */
1816 #define S_APPEND 4 /* Append-only file */
1817 #define S_IMMUTABLE 8 /* Immutable file */
1818 #define S_DEAD 16 /* removed, but still open directory */
1819 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1820 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1821 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1822 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1823 #define S_PRIVATE 512 /* Inode is fs-internal */
1824 #define S_IMA 1024 /* Inode has an associated IMA struct */
1825 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1826 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1827 #ifdef CONFIG_FS_DAX
1828 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1829 #else
1830 #define S_DAX 0 /* Make all the DAX code disappear */
1831 #endif
1832
1833 /*
1834 * Note that nosuid etc flags are inode-specific: setting some file-system
1835 * flags just means all the inodes inherit those flags by default. It might be
1836 * possible to override it selectively if you really wanted to with some
1837 * ioctl() that is not currently implemented.
1838 *
1839 * Exception: MS_RDONLY is always applied to the entire file system.
1840 *
1841 * Unfortunately, it is possible to change a filesystems flags with it mounted
1842 * with files in use. This means that all of the inodes will not have their
1843 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1844 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1845 */
1846 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1847
1848 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1849 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1850 ((inode)->i_flags & S_SYNC))
1851 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1852 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1853 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1854 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1855 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1856
1857 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1858 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1859 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1860 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1861
1862 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1863 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1864 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1865 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1866 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1867 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1868 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1869 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1870
1871 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1872 (inode)->i_rdev == WHITEOUT_DEV)
1873
1874 static inline bool HAS_UNMAPPED_ID(struct inode *inode)
1875 {
1876 return !uid_valid(inode->i_uid) || !gid_valid(inode->i_gid);
1877 }
1878
1879 /*
1880 * Inode state bits. Protected by inode->i_lock
1881 *
1882 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1883 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1884 *
1885 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1886 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1887 * various stages of removing an inode.
1888 *
1889 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1890 *
1891 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1892 * fdatasync(). i_atime is the usual cause.
1893 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1894 * these changes separately from I_DIRTY_SYNC so that we
1895 * don't have to write inode on fdatasync() when only
1896 * mtime has changed in it.
1897 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1898 * I_NEW Serves as both a mutex and completion notification.
1899 * New inodes set I_NEW. If two processes both create
1900 * the same inode, one of them will release its inode and
1901 * wait for I_NEW to be released before returning.
1902 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1903 * also cause waiting on I_NEW, without I_NEW actually
1904 * being set. find_inode() uses this to prevent returning
1905 * nearly-dead inodes.
1906 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1907 * is zero. I_FREEING must be set when I_WILL_FREE is
1908 * cleared.
1909 * I_FREEING Set when inode is about to be freed but still has dirty
1910 * pages or buffers attached or the inode itself is still
1911 * dirty.
1912 * I_CLEAR Added by clear_inode(). In this state the inode is
1913 * clean and can be destroyed. Inode keeps I_FREEING.
1914 *
1915 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1916 * prohibited for many purposes. iget() must wait for
1917 * the inode to be completely released, then create it
1918 * anew. Other functions will just ignore such inodes,
1919 * if appropriate. I_NEW is used for waiting.
1920 *
1921 * I_SYNC Writeback of inode is running. The bit is set during
1922 * data writeback, and cleared with a wakeup on the bit
1923 * address once it is done. The bit is also used to pin
1924 * the inode in memory for flusher thread.
1925 *
1926 * I_REFERENCED Marks the inode as recently references on the LRU list.
1927 *
1928 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1929 *
1930 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
1931 * synchronize competing switching instances and to tell
1932 * wb stat updates to grab mapping->tree_lock. See
1933 * inode_switch_wb_work_fn() for details.
1934 *
1935 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1936 */
1937 #define I_DIRTY_SYNC (1 << 0)
1938 #define I_DIRTY_DATASYNC (1 << 1)
1939 #define I_DIRTY_PAGES (1 << 2)
1940 #define __I_NEW 3
1941 #define I_NEW (1 << __I_NEW)
1942 #define I_WILL_FREE (1 << 4)
1943 #define I_FREEING (1 << 5)
1944 #define I_CLEAR (1 << 6)
1945 #define __I_SYNC 7
1946 #define I_SYNC (1 << __I_SYNC)
1947 #define I_REFERENCED (1 << 8)
1948 #define __I_DIO_WAKEUP 9
1949 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
1950 #define I_LINKABLE (1 << 10)
1951 #define I_DIRTY_TIME (1 << 11)
1952 #define __I_DIRTY_TIME_EXPIRED 12
1953 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
1954 #define I_WB_SWITCH (1 << 13)
1955
1956 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1957 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1958
1959 extern void __mark_inode_dirty(struct inode *, int);
1960 static inline void mark_inode_dirty(struct inode *inode)
1961 {
1962 __mark_inode_dirty(inode, I_DIRTY);
1963 }
1964
1965 static inline void mark_inode_dirty_sync(struct inode *inode)
1966 {
1967 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1968 }
1969
1970 extern void inc_nlink(struct inode *inode);
1971 extern void drop_nlink(struct inode *inode);
1972 extern void clear_nlink(struct inode *inode);
1973 extern void set_nlink(struct inode *inode, unsigned int nlink);
1974
1975 static inline void inode_inc_link_count(struct inode *inode)
1976 {
1977 inc_nlink(inode);
1978 mark_inode_dirty(inode);
1979 }
1980
1981 static inline void inode_dec_link_count(struct inode *inode)
1982 {
1983 drop_nlink(inode);
1984 mark_inode_dirty(inode);
1985 }
1986
1987 /**
1988 * inode_inc_iversion - increments i_version
1989 * @inode: inode that need to be updated
1990 *
1991 * Every time the inode is modified, the i_version field will be incremented.
1992 * The filesystem has to be mounted with i_version flag
1993 */
1994
1995 static inline void inode_inc_iversion(struct inode *inode)
1996 {
1997 spin_lock(&inode->i_lock);
1998 inode->i_version++;
1999 spin_unlock(&inode->i_lock);
2000 }
2001
2002 enum file_time_flags {
2003 S_ATIME = 1,
2004 S_MTIME = 2,
2005 S_CTIME = 4,
2006 S_VERSION = 8,
2007 };
2008
2009 extern bool atime_needs_update(const struct path *, struct inode *);
2010 extern void touch_atime(const struct path *);
2011 static inline void file_accessed(struct file *file)
2012 {
2013 if (!(file->f_flags & O_NOATIME))
2014 touch_atime(&file->f_path);
2015 }
2016
2017 int sync_inode(struct inode *inode, struct writeback_control *wbc);
2018 int sync_inode_metadata(struct inode *inode, int wait);
2019
2020 struct file_system_type {
2021 const char *name;
2022 int fs_flags;
2023 #define FS_REQUIRES_DEV 1
2024 #define FS_BINARY_MOUNTDATA 2
2025 #define FS_HAS_SUBTYPE 4
2026 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2027 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2028 struct dentry *(*mount) (struct file_system_type *, int,
2029 const char *, void *);
2030 void (*kill_sb) (struct super_block *);
2031 struct module *owner;
2032 struct file_system_type * next;
2033 struct hlist_head fs_supers;
2034
2035 struct lock_class_key s_lock_key;
2036 struct lock_class_key s_umount_key;
2037 struct lock_class_key s_vfs_rename_key;
2038 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2039
2040 struct lock_class_key i_lock_key;
2041 struct lock_class_key i_mutex_key;
2042 struct lock_class_key i_mutex_dir_key;
2043 };
2044
2045 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2046
2047 extern struct dentry *mount_ns(struct file_system_type *fs_type,
2048 int flags, void *data, void *ns, struct user_namespace *user_ns,
2049 int (*fill_super)(struct super_block *, void *, int));
2050 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2051 int flags, const char *dev_name, void *data,
2052 int (*fill_super)(struct super_block *, void *, int));
2053 extern struct dentry *mount_single(struct file_system_type *fs_type,
2054 int flags, void *data,
2055 int (*fill_super)(struct super_block *, void *, int));
2056 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2057 int flags, void *data,
2058 int (*fill_super)(struct super_block *, void *, int));
2059 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2060 void generic_shutdown_super(struct super_block *sb);
2061 void kill_block_super(struct super_block *sb);
2062 void kill_anon_super(struct super_block *sb);
2063 void kill_litter_super(struct super_block *sb);
2064 void deactivate_super(struct super_block *sb);
2065 void deactivate_locked_super(struct super_block *sb);
2066 int set_anon_super(struct super_block *s, void *data);
2067 int get_anon_bdev(dev_t *);
2068 void free_anon_bdev(dev_t);
2069 struct super_block *sget_userns(struct file_system_type *type,
2070 int (*test)(struct super_block *,void *),
2071 int (*set)(struct super_block *,void *),
2072 int flags, struct user_namespace *user_ns,
2073 void *data);
2074 struct super_block *sget(struct file_system_type *type,
2075 int (*test)(struct super_block *,void *),
2076 int (*set)(struct super_block *,void *),
2077 int flags, void *data);
2078 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
2079 const struct super_operations *ops,
2080 const struct dentry_operations *dops,
2081 unsigned long);
2082
2083 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2084 #define fops_get(fops) \
2085 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2086 #define fops_put(fops) \
2087 do { if (fops) module_put((fops)->owner); } while(0)
2088 /*
2089 * This one is to be used *ONLY* from ->open() instances.
2090 * fops must be non-NULL, pinned down *and* module dependencies
2091 * should be sufficient to pin the caller down as well.
2092 */
2093 #define replace_fops(f, fops) \
2094 do { \
2095 struct file *__file = (f); \
2096 fops_put(__file->f_op); \
2097 BUG_ON(!(__file->f_op = (fops))); \
2098 } while(0)
2099
2100 extern int register_filesystem(struct file_system_type *);
2101 extern int unregister_filesystem(struct file_system_type *);
2102 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
2103 #define kern_mount(type) kern_mount_data(type, NULL)
2104 extern void kern_unmount(struct vfsmount *mnt);
2105 extern int may_umount_tree(struct vfsmount *);
2106 extern int may_umount(struct vfsmount *);
2107 extern long do_mount(const char *, const char __user *,
2108 const char *, unsigned long, void *);
2109 extern struct vfsmount *collect_mounts(struct path *);
2110 extern void drop_collected_mounts(struct vfsmount *);
2111 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2112 struct vfsmount *);
2113 extern int vfs_statfs(struct path *, struct kstatfs *);
2114 extern int user_statfs(const char __user *, struct kstatfs *);
2115 extern int fd_statfs(int, struct kstatfs *);
2116 extern int vfs_ustat(dev_t, struct kstatfs *);
2117 extern int freeze_super(struct super_block *super);
2118 extern int thaw_super(struct super_block *super);
2119 extern bool our_mnt(struct vfsmount *mnt);
2120
2121 extern int current_umask(void);
2122
2123 extern void ihold(struct inode * inode);
2124 extern void iput(struct inode *);
2125 extern int generic_update_time(struct inode *, struct timespec *, int);
2126
2127 /* /sys/fs */
2128 extern struct kobject *fs_kobj;
2129
2130 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2131
2132 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2133 extern int locks_mandatory_locked(struct file *);
2134 extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
2135
2136 /*
2137 * Candidates for mandatory locking have the setgid bit set
2138 * but no group execute bit - an otherwise meaningless combination.
2139 */
2140
2141 static inline int __mandatory_lock(struct inode *ino)
2142 {
2143 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2144 }
2145
2146 /*
2147 * ... and these candidates should be on MS_MANDLOCK mounted fs,
2148 * otherwise these will be advisory locks
2149 */
2150
2151 static inline int mandatory_lock(struct inode *ino)
2152 {
2153 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2154 }
2155
2156 static inline int locks_verify_locked(struct file *file)
2157 {
2158 if (mandatory_lock(file_inode(file)))
2159 return locks_mandatory_locked(file);
2160 return 0;
2161 }
2162
2163 static inline int locks_verify_truncate(struct inode *inode,
2164 struct file *f,
2165 loff_t size)
2166 {
2167 if (!inode->i_flctx || !mandatory_lock(inode))
2168 return 0;
2169
2170 if (size < inode->i_size) {
2171 return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2172 F_WRLCK);
2173 } else {
2174 return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2175 F_WRLCK);
2176 }
2177 }
2178
2179 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2180
2181 static inline int locks_mandatory_locked(struct file *file)
2182 {
2183 return 0;
2184 }
2185
2186 static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2187 loff_t start, loff_t end, unsigned char type)
2188 {
2189 return 0;
2190 }
2191
2192 static inline int __mandatory_lock(struct inode *inode)
2193 {
2194 return 0;
2195 }
2196
2197 static inline int mandatory_lock(struct inode *inode)
2198 {
2199 return 0;
2200 }
2201
2202 static inline int locks_verify_locked(struct file *file)
2203 {
2204 return 0;
2205 }
2206
2207 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2208 size_t size)
2209 {
2210 return 0;
2211 }
2212
2213 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2214
2215
2216 #ifdef CONFIG_FILE_LOCKING
2217 static inline int break_lease(struct inode *inode, unsigned int mode)
2218 {
2219 /*
2220 * Since this check is lockless, we must ensure that any refcounts
2221 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2222 * could end up racing with tasks trying to set a new lease on this
2223 * file.
2224 */
2225 smp_mb();
2226 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2227 return __break_lease(inode, mode, FL_LEASE);
2228 return 0;
2229 }
2230
2231 static inline int break_deleg(struct inode *inode, unsigned int mode)
2232 {
2233 /*
2234 * Since this check is lockless, we must ensure that any refcounts
2235 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2236 * could end up racing with tasks trying to set a new lease on this
2237 * file.
2238 */
2239 smp_mb();
2240 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2241 return __break_lease(inode, mode, FL_DELEG);
2242 return 0;
2243 }
2244
2245 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2246 {
2247 int ret;
2248
2249 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2250 if (ret == -EWOULDBLOCK && delegated_inode) {
2251 *delegated_inode = inode;
2252 ihold(inode);
2253 }
2254 return ret;
2255 }
2256
2257 static inline int break_deleg_wait(struct inode **delegated_inode)
2258 {
2259 int ret;
2260
2261 ret = break_deleg(*delegated_inode, O_WRONLY);
2262 iput(*delegated_inode);
2263 *delegated_inode = NULL;
2264 return ret;
2265 }
2266
2267 static inline int break_layout(struct inode *inode, bool wait)
2268 {
2269 smp_mb();
2270 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2271 return __break_lease(inode,
2272 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2273 FL_LAYOUT);
2274 return 0;
2275 }
2276
2277 #else /* !CONFIG_FILE_LOCKING */
2278 static inline int break_lease(struct inode *inode, unsigned int mode)
2279 {
2280 return 0;
2281 }
2282
2283 static inline int break_deleg(struct inode *inode, unsigned int mode)
2284 {
2285 return 0;
2286 }
2287
2288 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2289 {
2290 return 0;
2291 }
2292
2293 static inline int break_deleg_wait(struct inode **delegated_inode)
2294 {
2295 BUG();
2296 return 0;
2297 }
2298
2299 static inline int break_layout(struct inode *inode, bool wait)
2300 {
2301 return 0;
2302 }
2303
2304 #endif /* CONFIG_FILE_LOCKING */
2305
2306 /* fs/open.c */
2307 struct audit_names;
2308 struct filename {
2309 const char *name; /* pointer to actual string */
2310 const __user char *uptr; /* original userland pointer */
2311 struct audit_names *aname;
2312 int refcnt;
2313 const char iname[];
2314 };
2315
2316 extern long vfs_truncate(const struct path *, loff_t);
2317 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2318 struct file *filp);
2319 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2320 loff_t len);
2321 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2322 umode_t mode);
2323 extern struct file *file_open_name(struct filename *, int, umode_t);
2324 extern struct file *filp_open(const char *, int, umode_t);
2325 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2326 const char *, int, umode_t);
2327 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2328 extern int filp_close(struct file *, fl_owner_t id);
2329
2330 extern struct filename *getname_flags(const char __user *, int, int *);
2331 extern struct filename *getname(const char __user *);
2332 extern struct filename *getname_kernel(const char *);
2333 extern void putname(struct filename *name);
2334
2335 enum {
2336 FILE_CREATED = 1,
2337 FILE_OPENED = 2
2338 };
2339 extern int finish_open(struct file *file, struct dentry *dentry,
2340 int (*open)(struct inode *, struct file *),
2341 int *opened);
2342 extern int finish_no_open(struct file *file, struct dentry *dentry);
2343
2344 /* fs/ioctl.c */
2345
2346 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2347
2348 /* fs/dcache.c */
2349 extern void __init vfs_caches_init_early(void);
2350 extern void __init vfs_caches_init(void);
2351
2352 extern struct kmem_cache *names_cachep;
2353
2354 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2355 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2356
2357 #ifdef CONFIG_BLOCK
2358 extern int register_blkdev(unsigned int, const char *);
2359 extern void unregister_blkdev(unsigned int, const char *);
2360 extern struct block_device *bdget(dev_t);
2361 extern struct block_device *bdgrab(struct block_device *bdev);
2362 extern void bd_set_size(struct block_device *, loff_t size);
2363 extern void bd_forget(struct inode *inode);
2364 extern void bdput(struct block_device *);
2365 extern void invalidate_bdev(struct block_device *);
2366 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2367 extern int sync_blockdev(struct block_device *bdev);
2368 extern void kill_bdev(struct block_device *);
2369 extern struct super_block *freeze_bdev(struct block_device *);
2370 extern void emergency_thaw_all(void);
2371 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2372 extern int fsync_bdev(struct block_device *);
2373
2374 extern struct super_block *blockdev_superblock;
2375
2376 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2377 {
2378 return sb == blockdev_superblock;
2379 }
2380 #else
2381 static inline void bd_forget(struct inode *inode) {}
2382 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2383 static inline void kill_bdev(struct block_device *bdev) {}
2384 static inline void invalidate_bdev(struct block_device *bdev) {}
2385
2386 static inline struct super_block *freeze_bdev(struct block_device *sb)
2387 {
2388 return NULL;
2389 }
2390
2391 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2392 {
2393 return 0;
2394 }
2395
2396 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2397 {
2398 }
2399
2400 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2401 {
2402 return false;
2403 }
2404 #endif
2405 extern int sync_filesystem(struct super_block *);
2406 extern const struct file_operations def_blk_fops;
2407 extern const struct file_operations def_chr_fops;
2408 #ifdef CONFIG_BLOCK
2409 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2410 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2411 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2412 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2413 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2414 void *holder);
2415 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2416 void *holder);
2417 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2418 extern int __blkdev_reread_part(struct block_device *bdev);
2419 extern int blkdev_reread_part(struct block_device *bdev);
2420
2421 #ifdef CONFIG_SYSFS
2422 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2423 extern void bd_unlink_disk_holder(struct block_device *bdev,
2424 struct gendisk *disk);
2425 #else
2426 static inline int bd_link_disk_holder(struct block_device *bdev,
2427 struct gendisk *disk)
2428 {
2429 return 0;
2430 }
2431 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2432 struct gendisk *disk)
2433 {
2434 }
2435 #endif
2436 #endif
2437
2438 /* fs/char_dev.c */
2439 #define CHRDEV_MAJOR_HASH_SIZE 255
2440 /* Marks the bottom of the first segment of free char majors */
2441 #define CHRDEV_MAJOR_DYN_END 234
2442 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2443 extern int register_chrdev_region(dev_t, unsigned, const char *);
2444 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2445 unsigned int count, const char *name,
2446 const struct file_operations *fops);
2447 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2448 unsigned int count, const char *name);
2449 extern void unregister_chrdev_region(dev_t, unsigned);
2450 extern void chrdev_show(struct seq_file *,off_t);
2451
2452 static inline int register_chrdev(unsigned int major, const char *name,
2453 const struct file_operations *fops)
2454 {
2455 return __register_chrdev(major, 0, 256, name, fops);
2456 }
2457
2458 static inline void unregister_chrdev(unsigned int major, const char *name)
2459 {
2460 __unregister_chrdev(major, 0, 256, name);
2461 }
2462
2463 /* fs/block_dev.c */
2464 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2465 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2466
2467 #ifdef CONFIG_BLOCK
2468 #define BLKDEV_MAJOR_HASH_SIZE 255
2469 extern const char *__bdevname(dev_t, char *buffer);
2470 extern const char *bdevname(struct block_device *bdev, char *buffer);
2471 extern struct block_device *lookup_bdev(const char *);
2472 extern void blkdev_show(struct seq_file *,off_t);
2473
2474 #else
2475 #define BLKDEV_MAJOR_HASH_SIZE 0
2476 #endif
2477
2478 extern void init_special_inode(struct inode *, umode_t, dev_t);
2479
2480 /* Invalid inode operations -- fs/bad_inode.c */
2481 extern void make_bad_inode(struct inode *);
2482 extern bool is_bad_inode(struct inode *);
2483
2484 #ifdef CONFIG_BLOCK
2485 static inline bool op_is_write(unsigned int op)
2486 {
2487 return op == REQ_OP_READ ? false : true;
2488 }
2489
2490 /*
2491 * return data direction, READ or WRITE
2492 */
2493 static inline int bio_data_dir(struct bio *bio)
2494 {
2495 return op_is_write(bio_op(bio)) ? WRITE : READ;
2496 }
2497
2498 extern void check_disk_size_change(struct gendisk *disk,
2499 struct block_device *bdev);
2500 extern int revalidate_disk(struct gendisk *);
2501 extern int check_disk_change(struct block_device *);
2502 extern int __invalidate_device(struct block_device *, bool);
2503 extern int invalidate_partition(struct gendisk *, int);
2504 #endif
2505 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2506 pgoff_t start, pgoff_t end);
2507
2508 static inline void invalidate_remote_inode(struct inode *inode)
2509 {
2510 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2511 S_ISLNK(inode->i_mode))
2512 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2513 }
2514 extern int invalidate_inode_pages2(struct address_space *mapping);
2515 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2516 pgoff_t start, pgoff_t end);
2517 extern int write_inode_now(struct inode *, int);
2518 extern int filemap_fdatawrite(struct address_space *);
2519 extern int filemap_flush(struct address_space *);
2520 extern int filemap_fdatawait(struct address_space *);
2521 extern void filemap_fdatawait_keep_errors(struct address_space *);
2522 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2523 loff_t lend);
2524 extern int filemap_write_and_wait(struct address_space *mapping);
2525 extern int filemap_write_and_wait_range(struct address_space *mapping,
2526 loff_t lstart, loff_t lend);
2527 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2528 loff_t start, loff_t end, int sync_mode);
2529 extern int filemap_fdatawrite_range(struct address_space *mapping,
2530 loff_t start, loff_t end);
2531 extern int filemap_check_errors(struct address_space *mapping);
2532
2533 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2534 int datasync);
2535 extern int vfs_fsync(struct file *file, int datasync);
2536
2537 /*
2538 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2539 * to already be updated for the write, and will return either the amount
2540 * of bytes passed in, or an error if syncing the file failed.
2541 */
2542 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2543 {
2544 if (iocb->ki_flags & IOCB_DSYNC) {
2545 int ret = vfs_fsync_range(iocb->ki_filp,
2546 iocb->ki_pos - count, iocb->ki_pos - 1,
2547 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2548 if (ret)
2549 return ret;
2550 }
2551
2552 return count;
2553 }
2554
2555 extern void emergency_sync(void);
2556 extern void emergency_remount(void);
2557 #ifdef CONFIG_BLOCK
2558 extern sector_t bmap(struct inode *, sector_t);
2559 #endif
2560 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2561 extern int inode_permission(struct inode *, int);
2562 extern int __inode_permission(struct inode *, int);
2563 extern int generic_permission(struct inode *, int);
2564 extern int __check_sticky(struct inode *dir, struct inode *inode);
2565
2566 static inline bool execute_ok(struct inode *inode)
2567 {
2568 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2569 }
2570
2571 static inline void file_start_write(struct file *file)
2572 {
2573 if (!S_ISREG(file_inode(file)->i_mode))
2574 return;
2575 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2576 }
2577
2578 static inline bool file_start_write_trylock(struct file *file)
2579 {
2580 if (!S_ISREG(file_inode(file)->i_mode))
2581 return true;
2582 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2583 }
2584
2585 static inline void file_end_write(struct file *file)
2586 {
2587 if (!S_ISREG(file_inode(file)->i_mode))
2588 return;
2589 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2590 }
2591
2592 /*
2593 * get_write_access() gets write permission for a file.
2594 * put_write_access() releases this write permission.
2595 * This is used for regular files.
2596 * We cannot support write (and maybe mmap read-write shared) accesses and
2597 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2598 * can have the following values:
2599 * 0: no writers, no VM_DENYWRITE mappings
2600 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2601 * > 0: (i_writecount) users are writing to the file.
2602 *
2603 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2604 * except for the cases where we don't hold i_writecount yet. Then we need to
2605 * use {get,deny}_write_access() - these functions check the sign and refuse
2606 * to do the change if sign is wrong.
2607 */
2608 static inline int get_write_access(struct inode *inode)
2609 {
2610 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2611 }
2612 static inline int deny_write_access(struct file *file)
2613 {
2614 struct inode *inode = file_inode(file);
2615 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2616 }
2617 static inline void put_write_access(struct inode * inode)
2618 {
2619 atomic_dec(&inode->i_writecount);
2620 }
2621 static inline void allow_write_access(struct file *file)
2622 {
2623 if (file)
2624 atomic_inc(&file_inode(file)->i_writecount);
2625 }
2626 static inline bool inode_is_open_for_write(const struct inode *inode)
2627 {
2628 return atomic_read(&inode->i_writecount) > 0;
2629 }
2630
2631 #ifdef CONFIG_IMA
2632 static inline void i_readcount_dec(struct inode *inode)
2633 {
2634 BUG_ON(!atomic_read(&inode->i_readcount));
2635 atomic_dec(&inode->i_readcount);
2636 }
2637 static inline void i_readcount_inc(struct inode *inode)
2638 {
2639 atomic_inc(&inode->i_readcount);
2640 }
2641 #else
2642 static inline void i_readcount_dec(struct inode *inode)
2643 {
2644 return;
2645 }
2646 static inline void i_readcount_inc(struct inode *inode)
2647 {
2648 return;
2649 }
2650 #endif
2651 extern int do_pipe_flags(int *, int);
2652
2653 #define __kernel_read_file_id(id) \
2654 id(UNKNOWN, unknown) \
2655 id(FIRMWARE, firmware) \
2656 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
2657 id(MODULE, kernel-module) \
2658 id(KEXEC_IMAGE, kexec-image) \
2659 id(KEXEC_INITRAMFS, kexec-initramfs) \
2660 id(POLICY, security-policy) \
2661 id(MAX_ID, )
2662
2663 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2664 #define __fid_stringify(dummy, str) #str,
2665
2666 enum kernel_read_file_id {
2667 __kernel_read_file_id(__fid_enumify)
2668 };
2669
2670 static const char * const kernel_read_file_str[] = {
2671 __kernel_read_file_id(__fid_stringify)
2672 };
2673
2674 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id)
2675 {
2676 if (id < 0 || id >= READING_MAX_ID)
2677 return kernel_read_file_str[READING_UNKNOWN];
2678
2679 return kernel_read_file_str[id];
2680 }
2681
2682 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2683 extern int kernel_read_file(struct file *, void **, loff_t *, loff_t,
2684 enum kernel_read_file_id);
2685 extern int kernel_read_file_from_path(char *, void **, loff_t *, loff_t,
2686 enum kernel_read_file_id);
2687 extern int kernel_read_file_from_fd(int, void **, loff_t *, loff_t,
2688 enum kernel_read_file_id);
2689 extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
2690 extern ssize_t __kernel_write(struct file *, const char *, size_t, loff_t *);
2691 extern struct file * open_exec(const char *);
2692
2693 /* fs/dcache.c -- generic fs support functions */
2694 extern bool is_subdir(struct dentry *, struct dentry *);
2695 extern bool path_is_under(struct path *, struct path *);
2696
2697 extern char *file_path(struct file *, char *, int);
2698
2699 #include <linux/err.h>
2700
2701 /* needed for stackable file system support */
2702 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2703
2704 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2705
2706 extern int inode_init_always(struct super_block *, struct inode *);
2707 extern void inode_init_once(struct inode *);
2708 extern void address_space_init_once(struct address_space *mapping);
2709 extern struct inode * igrab(struct inode *);
2710 extern ino_t iunique(struct super_block *, ino_t);
2711 extern int inode_needs_sync(struct inode *inode);
2712 extern int generic_delete_inode(struct inode *inode);
2713 static inline int generic_drop_inode(struct inode *inode)
2714 {
2715 return !inode->i_nlink || inode_unhashed(inode);
2716 }
2717
2718 extern struct inode *ilookup5_nowait(struct super_block *sb,
2719 unsigned long hashval, int (*test)(struct inode *, void *),
2720 void *data);
2721 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2722 int (*test)(struct inode *, void *), void *data);
2723 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2724
2725 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2726 extern struct inode * iget_locked(struct super_block *, unsigned long);
2727 extern struct inode *find_inode_nowait(struct super_block *,
2728 unsigned long,
2729 int (*match)(struct inode *,
2730 unsigned long, void *),
2731 void *data);
2732 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2733 extern int insert_inode_locked(struct inode *);
2734 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2735 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2736 #else
2737 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2738 #endif
2739 extern void unlock_new_inode(struct inode *);
2740 extern unsigned int get_next_ino(void);
2741
2742 extern void __iget(struct inode * inode);
2743 extern void iget_failed(struct inode *);
2744 extern void clear_inode(struct inode *);
2745 extern void __destroy_inode(struct inode *);
2746 extern struct inode *new_inode_pseudo(struct super_block *sb);
2747 extern struct inode *new_inode(struct super_block *sb);
2748 extern void free_inode_nonrcu(struct inode *inode);
2749 extern int should_remove_suid(struct dentry *);
2750 extern int file_remove_privs(struct file *);
2751
2752 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2753 static inline void insert_inode_hash(struct inode *inode)
2754 {
2755 __insert_inode_hash(inode, inode->i_ino);
2756 }
2757
2758 extern void __remove_inode_hash(struct inode *);
2759 static inline void remove_inode_hash(struct inode *inode)
2760 {
2761 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
2762 __remove_inode_hash(inode);
2763 }
2764
2765 extern void inode_sb_list_add(struct inode *inode);
2766
2767 #ifdef CONFIG_BLOCK
2768 extern blk_qc_t submit_bio(struct bio *);
2769 extern int bdev_read_only(struct block_device *);
2770 #endif
2771 extern int set_blocksize(struct block_device *, int);
2772 extern int sb_set_blocksize(struct super_block *, int);
2773 extern int sb_min_blocksize(struct super_block *, int);
2774
2775 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2776 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2777 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2778 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2779 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2780 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2781 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
2782 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2783
2784 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos);
2785 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos);
2786
2787 /* fs/block_dev.c */
2788 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
2789 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2790 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2791 int datasync);
2792 extern void block_sync_page(struct page *page);
2793
2794 /* fs/splice.c */
2795 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2796 struct pipe_inode_info *, size_t, unsigned int);
2797 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2798 struct file *, loff_t *, size_t, unsigned int);
2799 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2800 struct file *out, loff_t *, size_t len, unsigned int flags);
2801 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2802 loff_t *opos, size_t len, unsigned int flags);
2803
2804
2805 extern void
2806 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2807 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2808 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2809 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
2810 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2811 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2812 int whence, loff_t maxsize, loff_t eof);
2813 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2814 int whence, loff_t size);
2815 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
2816 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
2817 extern int generic_file_open(struct inode * inode, struct file * filp);
2818 extern int nonseekable_open(struct inode * inode, struct file * filp);
2819
2820 #ifdef CONFIG_BLOCK
2821 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
2822 loff_t file_offset);
2823
2824 enum {
2825 /* need locking between buffered and direct access */
2826 DIO_LOCKING = 0x01,
2827
2828 /* filesystem does not support filling holes */
2829 DIO_SKIP_HOLES = 0x02,
2830
2831 /* filesystem can handle aio writes beyond i_size */
2832 DIO_ASYNC_EXTEND = 0x04,
2833
2834 /* inode/fs/bdev does not need truncate protection */
2835 DIO_SKIP_DIO_COUNT = 0x08,
2836 };
2837
2838 void dio_end_io(struct bio *bio, int error);
2839
2840 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
2841 struct block_device *bdev, struct iov_iter *iter,
2842 get_block_t get_block,
2843 dio_iodone_t end_io, dio_submit_t submit_io,
2844 int flags);
2845
2846 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
2847 struct inode *inode,
2848 struct iov_iter *iter,
2849 get_block_t get_block)
2850 {
2851 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
2852 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
2853 }
2854 #endif
2855
2856 void inode_dio_wait(struct inode *inode);
2857
2858 /*
2859 * inode_dio_begin - signal start of a direct I/O requests
2860 * @inode: inode the direct I/O happens on
2861 *
2862 * This is called once we've finished processing a direct I/O request,
2863 * and is used to wake up callers waiting for direct I/O to be quiesced.
2864 */
2865 static inline void inode_dio_begin(struct inode *inode)
2866 {
2867 atomic_inc(&inode->i_dio_count);
2868 }
2869
2870 /*
2871 * inode_dio_end - signal finish of a direct I/O requests
2872 * @inode: inode the direct I/O happens on
2873 *
2874 * This is called once we've finished processing a direct I/O request,
2875 * and is used to wake up callers waiting for direct I/O to be quiesced.
2876 */
2877 static inline void inode_dio_end(struct inode *inode)
2878 {
2879 if (atomic_dec_and_test(&inode->i_dio_count))
2880 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
2881 }
2882
2883 extern void inode_set_flags(struct inode *inode, unsigned int flags,
2884 unsigned int mask);
2885
2886 extern const struct file_operations generic_ro_fops;
2887
2888 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2889
2890 extern int readlink_copy(char __user *, int, const char *);
2891 extern int page_readlink(struct dentry *, char __user *, int);
2892 extern const char *page_get_link(struct dentry *, struct inode *,
2893 struct delayed_call *);
2894 extern void page_put_link(void *);
2895 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2896 int nofs);
2897 extern int page_symlink(struct inode *inode, const char *symname, int len);
2898 extern const struct inode_operations page_symlink_inode_operations;
2899 extern void kfree_link(void *);
2900 extern int generic_readlink(struct dentry *, char __user *, int);
2901 extern void generic_fillattr(struct inode *, struct kstat *);
2902 int vfs_getattr_nosec(struct path *path, struct kstat *stat);
2903 extern int vfs_getattr(struct path *, struct kstat *);
2904 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2905 void inode_add_bytes(struct inode *inode, loff_t bytes);
2906 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
2907 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2908 loff_t inode_get_bytes(struct inode *inode);
2909 void inode_set_bytes(struct inode *inode, loff_t bytes);
2910 const char *simple_get_link(struct dentry *, struct inode *,
2911 struct delayed_call *);
2912 extern const struct inode_operations simple_symlink_inode_operations;
2913
2914 extern int iterate_dir(struct file *, struct dir_context *);
2915
2916 extern int vfs_stat(const char __user *, struct kstat *);
2917 extern int vfs_lstat(const char __user *, struct kstat *);
2918 extern int vfs_fstat(unsigned int, struct kstat *);
2919 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2920
2921 extern int __generic_block_fiemap(struct inode *inode,
2922 struct fiemap_extent_info *fieinfo,
2923 loff_t start, loff_t len,
2924 get_block_t *get_block);
2925 extern int generic_block_fiemap(struct inode *inode,
2926 struct fiemap_extent_info *fieinfo, u64 start,
2927 u64 len, get_block_t *get_block);
2928
2929 extern void get_filesystem(struct file_system_type *fs);
2930 extern void put_filesystem(struct file_system_type *fs);
2931 extern struct file_system_type *get_fs_type(const char *name);
2932 extern struct super_block *get_super(struct block_device *);
2933 extern struct super_block *get_super_thawed(struct block_device *);
2934 extern struct super_block *get_active_super(struct block_device *bdev);
2935 extern void drop_super(struct super_block *sb);
2936 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2937 extern void iterate_supers_type(struct file_system_type *,
2938 void (*)(struct super_block *, void *), void *);
2939
2940 extern int dcache_dir_open(struct inode *, struct file *);
2941 extern int dcache_dir_close(struct inode *, struct file *);
2942 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2943 extern int dcache_readdir(struct file *, struct dir_context *);
2944 extern int simple_setattr(struct dentry *, struct iattr *);
2945 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2946 extern int simple_statfs(struct dentry *, struct kstatfs *);
2947 extern int simple_open(struct inode *inode, struct file *file);
2948 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2949 extern int simple_unlink(struct inode *, struct dentry *);
2950 extern int simple_rmdir(struct inode *, struct dentry *);
2951 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2952 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2953 extern int simple_empty(struct dentry *);
2954 extern int simple_readpage(struct file *file, struct page *page);
2955 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2956 loff_t pos, unsigned len, unsigned flags,
2957 struct page **pagep, void **fsdata);
2958 extern int simple_write_end(struct file *file, struct address_space *mapping,
2959 loff_t pos, unsigned len, unsigned copied,
2960 struct page *page, void *fsdata);
2961 extern int always_delete_dentry(const struct dentry *);
2962 extern struct inode *alloc_anon_inode(struct super_block *);
2963 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
2964 extern const struct dentry_operations simple_dentry_operations;
2965
2966 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
2967 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2968 extern const struct file_operations simple_dir_operations;
2969 extern const struct inode_operations simple_dir_inode_operations;
2970 extern void make_empty_dir_inode(struct inode *inode);
2971 extern bool is_empty_dir_inode(struct inode *inode);
2972 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2973 struct dentry *d_alloc_name(struct dentry *, const char *);
2974 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2975 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2976 extern void simple_release_fs(struct vfsmount **mount, int *count);
2977
2978 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2979 loff_t *ppos, const void *from, size_t available);
2980 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2981 const void __user *from, size_t count);
2982
2983 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
2984 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2985
2986 extern int generic_check_addressable(unsigned, u64);
2987
2988 #ifdef CONFIG_MIGRATION
2989 extern int buffer_migrate_page(struct address_space *,
2990 struct page *, struct page *,
2991 enum migrate_mode);
2992 #else
2993 #define buffer_migrate_page NULL
2994 #endif
2995
2996 extern int inode_change_ok(const struct inode *, struct iattr *);
2997 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2998 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2999
3000 extern int file_update_time(struct file *file);
3001
3002 extern int generic_show_options(struct seq_file *m, struct dentry *root);
3003 extern void save_mount_options(struct super_block *sb, char *options);
3004 extern void replace_mount_options(struct super_block *sb, char *options);
3005
3006 static inline bool io_is_direct(struct file *filp)
3007 {
3008 return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host);
3009 }
3010
3011 static inline int iocb_flags(struct file *file)
3012 {
3013 int res = 0;
3014 if (file->f_flags & O_APPEND)
3015 res |= IOCB_APPEND;
3016 if (io_is_direct(file))
3017 res |= IOCB_DIRECT;
3018 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3019 res |= IOCB_DSYNC;
3020 if (file->f_flags & __O_SYNC)
3021 res |= IOCB_SYNC;
3022 return res;
3023 }
3024
3025 static inline ino_t parent_ino(struct dentry *dentry)
3026 {
3027 ino_t res;
3028
3029 /*
3030 * Don't strictly need d_lock here? If the parent ino could change
3031 * then surely we'd have a deeper race in the caller?
3032 */
3033 spin_lock(&dentry->d_lock);
3034 res = dentry->d_parent->d_inode->i_ino;
3035 spin_unlock(&dentry->d_lock);
3036 return res;
3037 }
3038
3039 /* Transaction based IO helpers */
3040
3041 /*
3042 * An argresp is stored in an allocated page and holds the
3043 * size of the argument or response, along with its content
3044 */
3045 struct simple_transaction_argresp {
3046 ssize_t size;
3047 char data[0];
3048 };
3049
3050 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3051
3052 char *simple_transaction_get(struct file *file, const char __user *buf,
3053 size_t size);
3054 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3055 size_t size, loff_t *pos);
3056 int simple_transaction_release(struct inode *inode, struct file *file);
3057
3058 void simple_transaction_set(struct file *file, size_t n);
3059
3060 /*
3061 * simple attribute files
3062 *
3063 * These attributes behave similar to those in sysfs:
3064 *
3065 * Writing to an attribute immediately sets a value, an open file can be
3066 * written to multiple times.
3067 *
3068 * Reading from an attribute creates a buffer from the value that might get
3069 * read with multiple read calls. When the attribute has been read
3070 * completely, no further read calls are possible until the file is opened
3071 * again.
3072 *
3073 * All attributes contain a text representation of a numeric value
3074 * that are accessed with the get() and set() functions.
3075 */
3076 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3077 static int __fops ## _open(struct inode *inode, struct file *file) \
3078 { \
3079 __simple_attr_check_format(__fmt, 0ull); \
3080 return simple_attr_open(inode, file, __get, __set, __fmt); \
3081 } \
3082 static const struct file_operations __fops = { \
3083 .owner = THIS_MODULE, \
3084 .open = __fops ## _open, \
3085 .release = simple_attr_release, \
3086 .read = simple_attr_read, \
3087 .write = simple_attr_write, \
3088 .llseek = generic_file_llseek, \
3089 }
3090
3091 static inline __printf(1, 2)
3092 void __simple_attr_check_format(const char *fmt, ...)
3093 {
3094 /* don't do anything, just let the compiler check the arguments; */
3095 }
3096
3097 int simple_attr_open(struct inode *inode, struct file *file,
3098 int (*get)(void *, u64 *), int (*set)(void *, u64),
3099 const char *fmt);
3100 int simple_attr_release(struct inode *inode, struct file *file);
3101 ssize_t simple_attr_read(struct file *file, char __user *buf,
3102 size_t len, loff_t *ppos);
3103 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3104 size_t len, loff_t *ppos);
3105
3106 struct ctl_table;
3107 int proc_nr_files(struct ctl_table *table, int write,
3108 void __user *buffer, size_t *lenp, loff_t *ppos);
3109 int proc_nr_dentry(struct ctl_table *table, int write,
3110 void __user *buffer, size_t *lenp, loff_t *ppos);
3111 int proc_nr_inodes(struct ctl_table *table, int write,
3112 void __user *buffer, size_t *lenp, loff_t *ppos);
3113 int __init get_filesystem_list(char *buf);
3114
3115 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3116 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3117
3118 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3119 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3120 (flag & __FMODE_NONOTIFY)))
3121
3122 static inline bool is_sxid(umode_t mode)
3123 {
3124 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3125 }
3126
3127 static inline int check_sticky(struct inode *dir, struct inode *inode)
3128 {
3129 if (!(dir->i_mode & S_ISVTX))
3130 return 0;
3131
3132 return __check_sticky(dir, inode);
3133 }
3134
3135 static inline void inode_has_no_xattr(struct inode *inode)
3136 {
3137 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
3138 inode->i_flags |= S_NOSEC;
3139 }
3140
3141 static inline bool is_root_inode(struct inode *inode)
3142 {
3143 return inode == inode->i_sb->s_root->d_inode;
3144 }
3145
3146 static inline bool dir_emit(struct dir_context *ctx,
3147 const char *name, int namelen,
3148 u64 ino, unsigned type)
3149 {
3150 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3151 }
3152 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3153 {
3154 return ctx->actor(ctx, ".", 1, ctx->pos,
3155 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3156 }
3157 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3158 {
3159 return ctx->actor(ctx, "..", 2, ctx->pos,
3160 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3161 }
3162 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3163 {
3164 if (ctx->pos == 0) {
3165 if (!dir_emit_dot(file, ctx))
3166 return false;
3167 ctx->pos = 1;
3168 }
3169 if (ctx->pos == 1) {
3170 if (!dir_emit_dotdot(file, ctx))
3171 return false;
3172 ctx->pos = 2;
3173 }
3174 return true;
3175 }
3176 static inline bool dir_relax(struct inode *inode)
3177 {
3178 inode_unlock(inode);
3179 inode_lock(inode);
3180 return !IS_DEADDIR(inode);
3181 }
3182
3183 static inline bool dir_relax_shared(struct inode *inode)
3184 {
3185 inode_unlock_shared(inode);
3186 inode_lock_shared(inode);
3187 return !IS_DEADDIR(inode);
3188 }
3189
3190 extern bool path_noexec(const struct path *path);
3191 extern void inode_nohighmem(struct inode *inode);
3192
3193 #endif /* _LINUX_FS_H */