]> git.proxmox.com Git - mirror_ubuntu-focal-kernel.git/blob - include/linux/fs.h
Input: wm97xx: add new AC97 bus support
[mirror_ubuntu-focal-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_bit.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/mm_types.h>
22 #include <linux/capability.h>
23 #include <linux/semaphore.h>
24 #include <linux/fcntl.h>
25 #include <linux/fiemap.h>
26 #include <linux/rculist_bl.h>
27 #include <linux/atomic.h>
28 #include <linux/shrinker.h>
29 #include <linux/migrate_mode.h>
30 #include <linux/uidgid.h>
31 #include <linux/lockdep.h>
32 #include <linux/percpu-rwsem.h>
33 #include <linux/workqueue.h>
34 #include <linux/delayed_call.h>
35 #include <linux/uuid.h>
36 #include <linux/errseq.h>
37
38 #include <asm/byteorder.h>
39 #include <uapi/linux/fs.h>
40
41 struct backing_dev_info;
42 struct bdi_writeback;
43 struct bio;
44 struct export_operations;
45 struct hd_geometry;
46 struct iovec;
47 struct kiocb;
48 struct kobject;
49 struct pipe_inode_info;
50 struct poll_table_struct;
51 struct kstatfs;
52 struct vm_area_struct;
53 struct vfsmount;
54 struct cred;
55 struct swap_info_struct;
56 struct seq_file;
57 struct workqueue_struct;
58 struct iov_iter;
59 struct fscrypt_info;
60 struct fscrypt_operations;
61
62 extern void __init inode_init(void);
63 extern void __init inode_init_early(void);
64 extern void __init files_init(void);
65 extern void __init files_maxfiles_init(void);
66
67 extern struct files_stat_struct files_stat;
68 extern unsigned long get_max_files(void);
69 extern unsigned int sysctl_nr_open;
70 extern struct inodes_stat_t inodes_stat;
71 extern int leases_enable, lease_break_time;
72 extern int sysctl_protected_symlinks;
73 extern int sysctl_protected_hardlinks;
74
75 typedef __kernel_rwf_t rwf_t;
76
77 struct buffer_head;
78 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
79 struct buffer_head *bh_result, int create);
80 typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
81 ssize_t bytes, void *private);
82
83 #define MAY_EXEC 0x00000001
84 #define MAY_WRITE 0x00000002
85 #define MAY_READ 0x00000004
86 #define MAY_APPEND 0x00000008
87 #define MAY_ACCESS 0x00000010
88 #define MAY_OPEN 0x00000020
89 #define MAY_CHDIR 0x00000040
90 /* called from RCU mode, don't block */
91 #define MAY_NOT_BLOCK 0x00000080
92
93 /*
94 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
95 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
96 */
97
98 /* file is open for reading */
99 #define FMODE_READ ((__force fmode_t)0x1)
100 /* file is open for writing */
101 #define FMODE_WRITE ((__force fmode_t)0x2)
102 /* file is seekable */
103 #define FMODE_LSEEK ((__force fmode_t)0x4)
104 /* file can be accessed using pread */
105 #define FMODE_PREAD ((__force fmode_t)0x8)
106 /* file can be accessed using pwrite */
107 #define FMODE_PWRITE ((__force fmode_t)0x10)
108 /* File is opened for execution with sys_execve / sys_uselib */
109 #define FMODE_EXEC ((__force fmode_t)0x20)
110 /* File is opened with O_NDELAY (only set for block devices) */
111 #define FMODE_NDELAY ((__force fmode_t)0x40)
112 /* File is opened with O_EXCL (only set for block devices) */
113 #define FMODE_EXCL ((__force fmode_t)0x80)
114 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
115 (specialy hack for floppy.c) */
116 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
117 /* 32bit hashes as llseek() offset (for directories) */
118 #define FMODE_32BITHASH ((__force fmode_t)0x200)
119 /* 64bit hashes as llseek() offset (for directories) */
120 #define FMODE_64BITHASH ((__force fmode_t)0x400)
121
122 /*
123 * Don't update ctime and mtime.
124 *
125 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
126 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
127 */
128 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
129
130 /* Expect random access pattern */
131 #define FMODE_RANDOM ((__force fmode_t)0x1000)
132
133 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
134 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
135
136 /* File is opened with O_PATH; almost nothing can be done with it */
137 #define FMODE_PATH ((__force fmode_t)0x4000)
138
139 /* File needs atomic accesses to f_pos */
140 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
141 /* Write access to underlying fs */
142 #define FMODE_WRITER ((__force fmode_t)0x10000)
143 /* Has read method(s) */
144 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
145 /* Has write method(s) */
146 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
147
148 /* File was opened by fanotify and shouldn't generate fanotify events */
149 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
150
151 /* File is capable of returning -EAGAIN if I/O will block */
152 #define FMODE_NOWAIT ((__force fmode_t)0x8000000)
153
154 /*
155 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
156 * that indicates that they should check the contents of the iovec are
157 * valid, but not check the memory that the iovec elements
158 * points too.
159 */
160 #define CHECK_IOVEC_ONLY -1
161
162 /*
163 * Attribute flags. These should be or-ed together to figure out what
164 * has been changed!
165 */
166 #define ATTR_MODE (1 << 0)
167 #define ATTR_UID (1 << 1)
168 #define ATTR_GID (1 << 2)
169 #define ATTR_SIZE (1 << 3)
170 #define ATTR_ATIME (1 << 4)
171 #define ATTR_MTIME (1 << 5)
172 #define ATTR_CTIME (1 << 6)
173 #define ATTR_ATIME_SET (1 << 7)
174 #define ATTR_MTIME_SET (1 << 8)
175 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
176 #define ATTR_ATTR_FLAG (1 << 10)
177 #define ATTR_KILL_SUID (1 << 11)
178 #define ATTR_KILL_SGID (1 << 12)
179 #define ATTR_FILE (1 << 13)
180 #define ATTR_KILL_PRIV (1 << 14)
181 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
182 #define ATTR_TIMES_SET (1 << 16)
183 #define ATTR_TOUCH (1 << 17)
184
185 /*
186 * Whiteout is represented by a char device. The following constants define the
187 * mode and device number to use.
188 */
189 #define WHITEOUT_MODE 0
190 #define WHITEOUT_DEV 0
191
192 /*
193 * This is the Inode Attributes structure, used for notify_change(). It
194 * uses the above definitions as flags, to know which values have changed.
195 * Also, in this manner, a Filesystem can look at only the values it cares
196 * about. Basically, these are the attributes that the VFS layer can
197 * request to change from the FS layer.
198 *
199 * Derek Atkins <warlord@MIT.EDU> 94-10-20
200 */
201 struct iattr {
202 unsigned int ia_valid;
203 umode_t ia_mode;
204 kuid_t ia_uid;
205 kgid_t ia_gid;
206 loff_t ia_size;
207 struct timespec ia_atime;
208 struct timespec ia_mtime;
209 struct timespec ia_ctime;
210
211 /*
212 * Not an attribute, but an auxiliary info for filesystems wanting to
213 * implement an ftruncate() like method. NOTE: filesystem should
214 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
215 */
216 struct file *ia_file;
217 };
218
219 /*
220 * Includes for diskquotas.
221 */
222 #include <linux/quota.h>
223
224 /*
225 * Maximum number of layers of fs stack. Needs to be limited to
226 * prevent kernel stack overflow
227 */
228 #define FILESYSTEM_MAX_STACK_DEPTH 2
229
230 /**
231 * enum positive_aop_returns - aop return codes with specific semantics
232 *
233 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
234 * completed, that the page is still locked, and
235 * should be considered active. The VM uses this hint
236 * to return the page to the active list -- it won't
237 * be a candidate for writeback again in the near
238 * future. Other callers must be careful to unlock
239 * the page if they get this return. Returned by
240 * writepage();
241 *
242 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
243 * unlocked it and the page might have been truncated.
244 * The caller should back up to acquiring a new page and
245 * trying again. The aop will be taking reasonable
246 * precautions not to livelock. If the caller held a page
247 * reference, it should drop it before retrying. Returned
248 * by readpage().
249 *
250 * address_space_operation functions return these large constants to indicate
251 * special semantics to the caller. These are much larger than the bytes in a
252 * page to allow for functions that return the number of bytes operated on in a
253 * given page.
254 */
255
256 enum positive_aop_returns {
257 AOP_WRITEPAGE_ACTIVATE = 0x80000,
258 AOP_TRUNCATED_PAGE = 0x80001,
259 };
260
261 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
262 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
263 * helper code (eg buffer layer)
264 * to clear GFP_FS from alloc */
265
266 /*
267 * oh the beauties of C type declarations.
268 */
269 struct page;
270 struct address_space;
271 struct writeback_control;
272
273 /*
274 * Write life time hint values.
275 */
276 enum rw_hint {
277 WRITE_LIFE_NOT_SET = 0,
278 WRITE_LIFE_NONE = RWH_WRITE_LIFE_NONE,
279 WRITE_LIFE_SHORT = RWH_WRITE_LIFE_SHORT,
280 WRITE_LIFE_MEDIUM = RWH_WRITE_LIFE_MEDIUM,
281 WRITE_LIFE_LONG = RWH_WRITE_LIFE_LONG,
282 WRITE_LIFE_EXTREME = RWH_WRITE_LIFE_EXTREME,
283 };
284
285 #define IOCB_EVENTFD (1 << 0)
286 #define IOCB_APPEND (1 << 1)
287 #define IOCB_DIRECT (1 << 2)
288 #define IOCB_HIPRI (1 << 3)
289 #define IOCB_DSYNC (1 << 4)
290 #define IOCB_SYNC (1 << 5)
291 #define IOCB_WRITE (1 << 6)
292 #define IOCB_NOWAIT (1 << 7)
293
294 struct kiocb {
295 struct file *ki_filp;
296 loff_t ki_pos;
297 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
298 void *private;
299 int ki_flags;
300 enum rw_hint ki_hint;
301 } __randomize_layout;
302
303 static inline bool is_sync_kiocb(struct kiocb *kiocb)
304 {
305 return kiocb->ki_complete == NULL;
306 }
307
308 /*
309 * "descriptor" for what we're up to with a read.
310 * This allows us to use the same read code yet
311 * have multiple different users of the data that
312 * we read from a file.
313 *
314 * The simplest case just copies the data to user
315 * mode.
316 */
317 typedef struct {
318 size_t written;
319 size_t count;
320 union {
321 char __user *buf;
322 void *data;
323 } arg;
324 int error;
325 } read_descriptor_t;
326
327 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
328 unsigned long, unsigned long);
329
330 struct address_space_operations {
331 int (*writepage)(struct page *page, struct writeback_control *wbc);
332 int (*readpage)(struct file *, struct page *);
333
334 /* Write back some dirty pages from this mapping. */
335 int (*writepages)(struct address_space *, struct writeback_control *);
336
337 /* Set a page dirty. Return true if this dirtied it */
338 int (*set_page_dirty)(struct page *page);
339
340 int (*readpages)(struct file *filp, struct address_space *mapping,
341 struct list_head *pages, unsigned nr_pages);
342
343 int (*write_begin)(struct file *, struct address_space *mapping,
344 loff_t pos, unsigned len, unsigned flags,
345 struct page **pagep, void **fsdata);
346 int (*write_end)(struct file *, struct address_space *mapping,
347 loff_t pos, unsigned len, unsigned copied,
348 struct page *page, void *fsdata);
349
350 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
351 sector_t (*bmap)(struct address_space *, sector_t);
352 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
353 int (*releasepage) (struct page *, gfp_t);
354 void (*freepage)(struct page *);
355 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
356 /*
357 * migrate the contents of a page to the specified target. If
358 * migrate_mode is MIGRATE_ASYNC, it must not block.
359 */
360 int (*migratepage) (struct address_space *,
361 struct page *, struct page *, enum migrate_mode);
362 bool (*isolate_page)(struct page *, isolate_mode_t);
363 void (*putback_page)(struct page *);
364 int (*launder_page) (struct page *);
365 int (*is_partially_uptodate) (struct page *, unsigned long,
366 unsigned long);
367 void (*is_dirty_writeback) (struct page *, bool *, bool *);
368 int (*error_remove_page)(struct address_space *, struct page *);
369
370 /* swapfile support */
371 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
372 sector_t *span);
373 void (*swap_deactivate)(struct file *file);
374 };
375
376 extern const struct address_space_operations empty_aops;
377
378 /*
379 * pagecache_write_begin/pagecache_write_end must be used by general code
380 * to write into the pagecache.
381 */
382 int pagecache_write_begin(struct file *, struct address_space *mapping,
383 loff_t pos, unsigned len, unsigned flags,
384 struct page **pagep, void **fsdata);
385
386 int pagecache_write_end(struct file *, struct address_space *mapping,
387 loff_t pos, unsigned len, unsigned copied,
388 struct page *page, void *fsdata);
389
390 struct address_space {
391 struct inode *host; /* owner: inode, block_device */
392 struct radix_tree_root page_tree; /* radix tree of all pages */
393 spinlock_t tree_lock; /* and lock protecting it */
394 atomic_t i_mmap_writable;/* count VM_SHARED mappings */
395 struct rb_root_cached i_mmap; /* tree of private and shared mappings */
396 struct rw_semaphore i_mmap_rwsem; /* protect tree, count, list */
397 /* Protected by tree_lock together with the radix tree */
398 unsigned long nrpages; /* number of total pages */
399 /* number of shadow or DAX exceptional entries */
400 unsigned long nrexceptional;
401 pgoff_t writeback_index;/* writeback starts here */
402 const struct address_space_operations *a_ops; /* methods */
403 unsigned long flags; /* error bits */
404 spinlock_t private_lock; /* for use by the address_space */
405 gfp_t gfp_mask; /* implicit gfp mask for allocations */
406 struct list_head private_list; /* ditto */
407 void *private_data; /* ditto */
408 errseq_t wb_err;
409 } __attribute__((aligned(sizeof(long)))) __randomize_layout;
410 /*
411 * On most architectures that alignment is already the case; but
412 * must be enforced here for CRIS, to let the least significant bit
413 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
414 */
415 struct request_queue;
416
417 struct block_device {
418 dev_t bd_dev; /* not a kdev_t - it's a search key */
419 int bd_openers;
420 struct inode * bd_inode; /* will die */
421 struct super_block * bd_super;
422 struct mutex bd_mutex; /* open/close mutex */
423 void * bd_claiming;
424 void * bd_holder;
425 int bd_holders;
426 bool bd_write_holder;
427 #ifdef CONFIG_SYSFS
428 struct list_head bd_holder_disks;
429 #endif
430 struct block_device * bd_contains;
431 unsigned bd_block_size;
432 u8 bd_partno;
433 struct hd_struct * bd_part;
434 /* number of times partitions within this device have been opened. */
435 unsigned bd_part_count;
436 int bd_invalidated;
437 struct gendisk * bd_disk;
438 struct request_queue * bd_queue;
439 struct backing_dev_info *bd_bdi;
440 struct list_head bd_list;
441 /*
442 * Private data. You must have bd_claim'ed the block_device
443 * to use this. NOTE: bd_claim allows an owner to claim
444 * the same device multiple times, the owner must take special
445 * care to not mess up bd_private for that case.
446 */
447 unsigned long bd_private;
448
449 /* The counter of freeze processes */
450 int bd_fsfreeze_count;
451 /* Mutex for freeze */
452 struct mutex bd_fsfreeze_mutex;
453 } __randomize_layout;
454
455 /*
456 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
457 * radix trees
458 */
459 #define PAGECACHE_TAG_DIRTY 0
460 #define PAGECACHE_TAG_WRITEBACK 1
461 #define PAGECACHE_TAG_TOWRITE 2
462
463 int mapping_tagged(struct address_space *mapping, int tag);
464
465 static inline void i_mmap_lock_write(struct address_space *mapping)
466 {
467 down_write(&mapping->i_mmap_rwsem);
468 }
469
470 static inline void i_mmap_unlock_write(struct address_space *mapping)
471 {
472 up_write(&mapping->i_mmap_rwsem);
473 }
474
475 static inline void i_mmap_lock_read(struct address_space *mapping)
476 {
477 down_read(&mapping->i_mmap_rwsem);
478 }
479
480 static inline void i_mmap_unlock_read(struct address_space *mapping)
481 {
482 up_read(&mapping->i_mmap_rwsem);
483 }
484
485 /*
486 * Might pages of this file be mapped into userspace?
487 */
488 static inline int mapping_mapped(struct address_space *mapping)
489 {
490 return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root);
491 }
492
493 /*
494 * Might pages of this file have been modified in userspace?
495 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
496 * marks vma as VM_SHARED if it is shared, and the file was opened for
497 * writing i.e. vma may be mprotected writable even if now readonly.
498 *
499 * If i_mmap_writable is negative, no new writable mappings are allowed. You
500 * can only deny writable mappings, if none exists right now.
501 */
502 static inline int mapping_writably_mapped(struct address_space *mapping)
503 {
504 return atomic_read(&mapping->i_mmap_writable) > 0;
505 }
506
507 static inline int mapping_map_writable(struct address_space *mapping)
508 {
509 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
510 0 : -EPERM;
511 }
512
513 static inline void mapping_unmap_writable(struct address_space *mapping)
514 {
515 atomic_dec(&mapping->i_mmap_writable);
516 }
517
518 static inline int mapping_deny_writable(struct address_space *mapping)
519 {
520 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
521 0 : -EBUSY;
522 }
523
524 static inline void mapping_allow_writable(struct address_space *mapping)
525 {
526 atomic_inc(&mapping->i_mmap_writable);
527 }
528
529 /*
530 * Use sequence counter to get consistent i_size on 32-bit processors.
531 */
532 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
533 #include <linux/seqlock.h>
534 #define __NEED_I_SIZE_ORDERED
535 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
536 #else
537 #define i_size_ordered_init(inode) do { } while (0)
538 #endif
539
540 struct posix_acl;
541 #define ACL_NOT_CACHED ((void *)(-1))
542 #define ACL_DONT_CACHE ((void *)(-3))
543
544 static inline struct posix_acl *
545 uncached_acl_sentinel(struct task_struct *task)
546 {
547 return (void *)task + 1;
548 }
549
550 static inline bool
551 is_uncached_acl(struct posix_acl *acl)
552 {
553 return (long)acl & 1;
554 }
555
556 #define IOP_FASTPERM 0x0001
557 #define IOP_LOOKUP 0x0002
558 #define IOP_NOFOLLOW 0x0004
559 #define IOP_XATTR 0x0008
560 #define IOP_DEFAULT_READLINK 0x0010
561
562 struct fsnotify_mark_connector;
563
564 /*
565 * Keep mostly read-only and often accessed (especially for
566 * the RCU path lookup and 'stat' data) fields at the beginning
567 * of the 'struct inode'
568 */
569 struct inode {
570 umode_t i_mode;
571 unsigned short i_opflags;
572 kuid_t i_uid;
573 kgid_t i_gid;
574 unsigned int i_flags;
575
576 #ifdef CONFIG_FS_POSIX_ACL
577 struct posix_acl *i_acl;
578 struct posix_acl *i_default_acl;
579 #endif
580
581 const struct inode_operations *i_op;
582 struct super_block *i_sb;
583 struct address_space *i_mapping;
584
585 #ifdef CONFIG_SECURITY
586 void *i_security;
587 #endif
588
589 /* Stat data, not accessed from path walking */
590 unsigned long i_ino;
591 /*
592 * Filesystems may only read i_nlink directly. They shall use the
593 * following functions for modification:
594 *
595 * (set|clear|inc|drop)_nlink
596 * inode_(inc|dec)_link_count
597 */
598 union {
599 const unsigned int i_nlink;
600 unsigned int __i_nlink;
601 };
602 dev_t i_rdev;
603 loff_t i_size;
604 struct timespec i_atime;
605 struct timespec i_mtime;
606 struct timespec i_ctime;
607 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
608 unsigned short i_bytes;
609 unsigned int i_blkbits;
610 enum rw_hint i_write_hint;
611 blkcnt_t i_blocks;
612
613 #ifdef __NEED_I_SIZE_ORDERED
614 seqcount_t i_size_seqcount;
615 #endif
616
617 /* Misc */
618 unsigned long i_state;
619 struct rw_semaphore i_rwsem;
620
621 unsigned long dirtied_when; /* jiffies of first dirtying */
622 unsigned long dirtied_time_when;
623
624 struct hlist_node i_hash;
625 struct list_head i_io_list; /* backing dev IO list */
626 #ifdef CONFIG_CGROUP_WRITEBACK
627 struct bdi_writeback *i_wb; /* the associated cgroup wb */
628
629 /* foreign inode detection, see wbc_detach_inode() */
630 int i_wb_frn_winner;
631 u16 i_wb_frn_avg_time;
632 u16 i_wb_frn_history;
633 #endif
634 struct list_head i_lru; /* inode LRU list */
635 struct list_head i_sb_list;
636 struct list_head i_wb_list; /* backing dev writeback list */
637 union {
638 struct hlist_head i_dentry;
639 struct rcu_head i_rcu;
640 };
641 u64 i_version;
642 atomic_t i_count;
643 atomic_t i_dio_count;
644 atomic_t i_writecount;
645 #ifdef CONFIG_IMA
646 atomic_t i_readcount; /* struct files open RO */
647 #endif
648 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
649 struct file_lock_context *i_flctx;
650 struct address_space i_data;
651 struct list_head i_devices;
652 union {
653 struct pipe_inode_info *i_pipe;
654 struct block_device *i_bdev;
655 struct cdev *i_cdev;
656 char *i_link;
657 unsigned i_dir_seq;
658 };
659
660 __u32 i_generation;
661
662 #ifdef CONFIG_FSNOTIFY
663 __u32 i_fsnotify_mask; /* all events this inode cares about */
664 struct fsnotify_mark_connector __rcu *i_fsnotify_marks;
665 #endif
666
667 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
668 struct fscrypt_info *i_crypt_info;
669 #endif
670
671 void *i_private; /* fs or device private pointer */
672 } __randomize_layout;
673
674 static inline unsigned int i_blocksize(const struct inode *node)
675 {
676 return (1 << node->i_blkbits);
677 }
678
679 static inline int inode_unhashed(struct inode *inode)
680 {
681 return hlist_unhashed(&inode->i_hash);
682 }
683
684 /*
685 * inode->i_mutex nesting subclasses for the lock validator:
686 *
687 * 0: the object of the current VFS operation
688 * 1: parent
689 * 2: child/target
690 * 3: xattr
691 * 4: second non-directory
692 * 5: second parent (when locking independent directories in rename)
693 *
694 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
695 * non-directories at once.
696 *
697 * The locking order between these classes is
698 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
699 */
700 enum inode_i_mutex_lock_class
701 {
702 I_MUTEX_NORMAL,
703 I_MUTEX_PARENT,
704 I_MUTEX_CHILD,
705 I_MUTEX_XATTR,
706 I_MUTEX_NONDIR2,
707 I_MUTEX_PARENT2,
708 };
709
710 static inline void inode_lock(struct inode *inode)
711 {
712 down_write(&inode->i_rwsem);
713 }
714
715 static inline void inode_unlock(struct inode *inode)
716 {
717 up_write(&inode->i_rwsem);
718 }
719
720 static inline void inode_lock_shared(struct inode *inode)
721 {
722 down_read(&inode->i_rwsem);
723 }
724
725 static inline void inode_unlock_shared(struct inode *inode)
726 {
727 up_read(&inode->i_rwsem);
728 }
729
730 static inline int inode_trylock(struct inode *inode)
731 {
732 return down_write_trylock(&inode->i_rwsem);
733 }
734
735 static inline int inode_trylock_shared(struct inode *inode)
736 {
737 return down_read_trylock(&inode->i_rwsem);
738 }
739
740 static inline int inode_is_locked(struct inode *inode)
741 {
742 return rwsem_is_locked(&inode->i_rwsem);
743 }
744
745 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
746 {
747 down_write_nested(&inode->i_rwsem, subclass);
748 }
749
750 void lock_two_nondirectories(struct inode *, struct inode*);
751 void unlock_two_nondirectories(struct inode *, struct inode*);
752
753 /*
754 * NOTE: in a 32bit arch with a preemptable kernel and
755 * an UP compile the i_size_read/write must be atomic
756 * with respect to the local cpu (unlike with preempt disabled),
757 * but they don't need to be atomic with respect to other cpus like in
758 * true SMP (so they need either to either locally disable irq around
759 * the read or for example on x86 they can be still implemented as a
760 * cmpxchg8b without the need of the lock prefix). For SMP compiles
761 * and 64bit archs it makes no difference if preempt is enabled or not.
762 */
763 static inline loff_t i_size_read(const struct inode *inode)
764 {
765 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
766 loff_t i_size;
767 unsigned int seq;
768
769 do {
770 seq = read_seqcount_begin(&inode->i_size_seqcount);
771 i_size = inode->i_size;
772 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
773 return i_size;
774 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
775 loff_t i_size;
776
777 preempt_disable();
778 i_size = inode->i_size;
779 preempt_enable();
780 return i_size;
781 #else
782 return inode->i_size;
783 #endif
784 }
785
786 /*
787 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
788 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
789 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
790 */
791 static inline void i_size_write(struct inode *inode, loff_t i_size)
792 {
793 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
794 preempt_disable();
795 write_seqcount_begin(&inode->i_size_seqcount);
796 inode->i_size = i_size;
797 write_seqcount_end(&inode->i_size_seqcount);
798 preempt_enable();
799 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
800 preempt_disable();
801 inode->i_size = i_size;
802 preempt_enable();
803 #else
804 inode->i_size = i_size;
805 #endif
806 }
807
808 static inline unsigned iminor(const struct inode *inode)
809 {
810 return MINOR(inode->i_rdev);
811 }
812
813 static inline unsigned imajor(const struct inode *inode)
814 {
815 return MAJOR(inode->i_rdev);
816 }
817
818 extern struct block_device *I_BDEV(struct inode *inode);
819
820 struct fown_struct {
821 rwlock_t lock; /* protects pid, uid, euid fields */
822 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
823 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
824 kuid_t uid, euid; /* uid/euid of process setting the owner */
825 int signum; /* posix.1b rt signal to be delivered on IO */
826 };
827
828 /*
829 * Track a single file's readahead state
830 */
831 struct file_ra_state {
832 pgoff_t start; /* where readahead started */
833 unsigned int size; /* # of readahead pages */
834 unsigned int async_size; /* do asynchronous readahead when
835 there are only # of pages ahead */
836
837 unsigned int ra_pages; /* Maximum readahead window */
838 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
839 loff_t prev_pos; /* Cache last read() position */
840 };
841
842 /*
843 * Check if @index falls in the readahead windows.
844 */
845 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
846 {
847 return (index >= ra->start &&
848 index < ra->start + ra->size);
849 }
850
851 struct file {
852 union {
853 struct llist_node fu_llist;
854 struct rcu_head fu_rcuhead;
855 } f_u;
856 struct path f_path;
857 struct inode *f_inode; /* cached value */
858 const struct file_operations *f_op;
859
860 /*
861 * Protects f_ep_links, f_flags.
862 * Must not be taken from IRQ context.
863 */
864 spinlock_t f_lock;
865 enum rw_hint f_write_hint;
866 atomic_long_t f_count;
867 unsigned int f_flags;
868 fmode_t f_mode;
869 struct mutex f_pos_lock;
870 loff_t f_pos;
871 struct fown_struct f_owner;
872 const struct cred *f_cred;
873 struct file_ra_state f_ra;
874
875 u64 f_version;
876 #ifdef CONFIG_SECURITY
877 void *f_security;
878 #endif
879 /* needed for tty driver, and maybe others */
880 void *private_data;
881
882 #ifdef CONFIG_EPOLL
883 /* Used by fs/eventpoll.c to link all the hooks to this file */
884 struct list_head f_ep_links;
885 struct list_head f_tfile_llink;
886 #endif /* #ifdef CONFIG_EPOLL */
887 struct address_space *f_mapping;
888 errseq_t f_wb_err;
889 } __randomize_layout
890 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
891
892 struct file_handle {
893 __u32 handle_bytes;
894 int handle_type;
895 /* file identifier */
896 unsigned char f_handle[0];
897 };
898
899 static inline struct file *get_file(struct file *f)
900 {
901 atomic_long_inc(&f->f_count);
902 return f;
903 }
904 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
905 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
906 #define file_count(x) atomic_long_read(&(x)->f_count)
907
908 #define MAX_NON_LFS ((1UL<<31) - 1)
909
910 /* Page cache limit. The filesystems should put that into their s_maxbytes
911 limits, otherwise bad things can happen in VM. */
912 #if BITS_PER_LONG==32
913 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
914 #elif BITS_PER_LONG==64
915 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
916 #endif
917
918 #define FL_POSIX 1
919 #define FL_FLOCK 2
920 #define FL_DELEG 4 /* NFSv4 delegation */
921 #define FL_ACCESS 8 /* not trying to lock, just looking */
922 #define FL_EXISTS 16 /* when unlocking, test for existence */
923 #define FL_LEASE 32 /* lease held on this file */
924 #define FL_CLOSE 64 /* unlock on close */
925 #define FL_SLEEP 128 /* A blocking lock */
926 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
927 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
928 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
929 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
930
931 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
932
933 /*
934 * Special return value from posix_lock_file() and vfs_lock_file() for
935 * asynchronous locking.
936 */
937 #define FILE_LOCK_DEFERRED 1
938
939 /* legacy typedef, should eventually be removed */
940 typedef void *fl_owner_t;
941
942 struct file_lock;
943
944 struct file_lock_operations {
945 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
946 void (*fl_release_private)(struct file_lock *);
947 };
948
949 struct lock_manager_operations {
950 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
951 unsigned long (*lm_owner_key)(struct file_lock *);
952 fl_owner_t (*lm_get_owner)(fl_owner_t);
953 void (*lm_put_owner)(fl_owner_t);
954 void (*lm_notify)(struct file_lock *); /* unblock callback */
955 int (*lm_grant)(struct file_lock *, int);
956 bool (*lm_break)(struct file_lock *);
957 int (*lm_change)(struct file_lock *, int, struct list_head *);
958 void (*lm_setup)(struct file_lock *, void **);
959 };
960
961 struct lock_manager {
962 struct list_head list;
963 /*
964 * NFSv4 and up also want opens blocked during the grace period;
965 * NLM doesn't care:
966 */
967 bool block_opens;
968 };
969
970 struct net;
971 void locks_start_grace(struct net *, struct lock_manager *);
972 void locks_end_grace(struct lock_manager *);
973 int locks_in_grace(struct net *);
974 int opens_in_grace(struct net *);
975
976 /* that will die - we need it for nfs_lock_info */
977 #include <linux/nfs_fs_i.h>
978
979 /*
980 * struct file_lock represents a generic "file lock". It's used to represent
981 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
982 * note that the same struct is used to represent both a request for a lock and
983 * the lock itself, but the same object is never used for both.
984 *
985 * FIXME: should we create a separate "struct lock_request" to help distinguish
986 * these two uses?
987 *
988 * The varous i_flctx lists are ordered by:
989 *
990 * 1) lock owner
991 * 2) lock range start
992 * 3) lock range end
993 *
994 * Obviously, the last two criteria only matter for POSIX locks.
995 */
996 struct file_lock {
997 struct file_lock *fl_next; /* singly linked list for this inode */
998 struct list_head fl_list; /* link into file_lock_context */
999 struct hlist_node fl_link; /* node in global lists */
1000 struct list_head fl_block; /* circular list of blocked processes */
1001 fl_owner_t fl_owner;
1002 unsigned int fl_flags;
1003 unsigned char fl_type;
1004 unsigned int fl_pid;
1005 int fl_link_cpu; /* what cpu's list is this on? */
1006 wait_queue_head_t fl_wait;
1007 struct file *fl_file;
1008 loff_t fl_start;
1009 loff_t fl_end;
1010
1011 struct fasync_struct * fl_fasync; /* for lease break notifications */
1012 /* for lease breaks: */
1013 unsigned long fl_break_time;
1014 unsigned long fl_downgrade_time;
1015
1016 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1017 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1018 union {
1019 struct nfs_lock_info nfs_fl;
1020 struct nfs4_lock_info nfs4_fl;
1021 struct {
1022 struct list_head link; /* link in AFS vnode's pending_locks list */
1023 int state; /* state of grant or error if -ve */
1024 } afs;
1025 } fl_u;
1026 } __randomize_layout;
1027
1028 struct file_lock_context {
1029 spinlock_t flc_lock;
1030 struct list_head flc_flock;
1031 struct list_head flc_posix;
1032 struct list_head flc_lease;
1033 };
1034
1035 /* The following constant reflects the upper bound of the file/locking space */
1036 #ifndef OFFSET_MAX
1037 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1038 #define OFFSET_MAX INT_LIMIT(loff_t)
1039 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1040 #endif
1041
1042 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1043
1044 /*
1045 * Return the inode to use for locking
1046 *
1047 * For overlayfs this should be the overlay inode, not the real inode returned
1048 * by file_inode(). For any other fs file_inode(filp) and locks_inode(filp) are
1049 * equal.
1050 */
1051 static inline struct inode *locks_inode(const struct file *f)
1052 {
1053 return f->f_path.dentry->d_inode;
1054 }
1055
1056 #ifdef CONFIG_FILE_LOCKING
1057 extern int fcntl_getlk(struct file *, unsigned int, struct flock *);
1058 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1059 struct flock *);
1060
1061 #if BITS_PER_LONG == 32
1062 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *);
1063 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1064 struct flock64 *);
1065 #endif
1066
1067 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1068 extern int fcntl_getlease(struct file *filp);
1069
1070 /* fs/locks.c */
1071 void locks_free_lock_context(struct inode *inode);
1072 void locks_free_lock(struct file_lock *fl);
1073 extern void locks_init_lock(struct file_lock *);
1074 extern struct file_lock * locks_alloc_lock(void);
1075 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1076 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1077 extern void locks_remove_posix(struct file *, fl_owner_t);
1078 extern void locks_remove_file(struct file *);
1079 extern void locks_release_private(struct file_lock *);
1080 extern void posix_test_lock(struct file *, struct file_lock *);
1081 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1082 extern int posix_unblock_lock(struct file_lock *);
1083 extern int vfs_test_lock(struct file *, struct file_lock *);
1084 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1085 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1086 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1087 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1088 extern void lease_get_mtime(struct inode *, struct timespec *time);
1089 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1090 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1091 extern int lease_modify(struct file_lock *, int, struct list_head *);
1092 struct files_struct;
1093 extern void show_fd_locks(struct seq_file *f,
1094 struct file *filp, struct files_struct *files);
1095 #else /* !CONFIG_FILE_LOCKING */
1096 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1097 struct flock __user *user)
1098 {
1099 return -EINVAL;
1100 }
1101
1102 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1103 unsigned int cmd, struct flock __user *user)
1104 {
1105 return -EACCES;
1106 }
1107
1108 #if BITS_PER_LONG == 32
1109 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1110 struct flock64 __user *user)
1111 {
1112 return -EINVAL;
1113 }
1114
1115 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1116 unsigned int cmd, struct flock64 __user *user)
1117 {
1118 return -EACCES;
1119 }
1120 #endif
1121 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1122 {
1123 return -EINVAL;
1124 }
1125
1126 static inline int fcntl_getlease(struct file *filp)
1127 {
1128 return F_UNLCK;
1129 }
1130
1131 static inline void
1132 locks_free_lock_context(struct inode *inode)
1133 {
1134 }
1135
1136 static inline void locks_init_lock(struct file_lock *fl)
1137 {
1138 return;
1139 }
1140
1141 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1142 {
1143 return;
1144 }
1145
1146 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1147 {
1148 return;
1149 }
1150
1151 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1152 {
1153 return;
1154 }
1155
1156 static inline void locks_remove_file(struct file *filp)
1157 {
1158 return;
1159 }
1160
1161 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1162 {
1163 return;
1164 }
1165
1166 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1167 struct file_lock *conflock)
1168 {
1169 return -ENOLCK;
1170 }
1171
1172 static inline int posix_unblock_lock(struct file_lock *waiter)
1173 {
1174 return -ENOENT;
1175 }
1176
1177 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1178 {
1179 return 0;
1180 }
1181
1182 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1183 struct file_lock *fl, struct file_lock *conf)
1184 {
1185 return -ENOLCK;
1186 }
1187
1188 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1189 {
1190 return 0;
1191 }
1192
1193 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1194 {
1195 return -ENOLCK;
1196 }
1197
1198 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1199 {
1200 return 0;
1201 }
1202
1203 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1204 {
1205 return;
1206 }
1207
1208 static inline int generic_setlease(struct file *filp, long arg,
1209 struct file_lock **flp, void **priv)
1210 {
1211 return -EINVAL;
1212 }
1213
1214 static inline int vfs_setlease(struct file *filp, long arg,
1215 struct file_lock **lease, void **priv)
1216 {
1217 return -EINVAL;
1218 }
1219
1220 static inline int lease_modify(struct file_lock *fl, int arg,
1221 struct list_head *dispose)
1222 {
1223 return -EINVAL;
1224 }
1225
1226 struct files_struct;
1227 static inline void show_fd_locks(struct seq_file *f,
1228 struct file *filp, struct files_struct *files) {}
1229 #endif /* !CONFIG_FILE_LOCKING */
1230
1231 static inline struct inode *file_inode(const struct file *f)
1232 {
1233 return f->f_inode;
1234 }
1235
1236 static inline struct dentry *file_dentry(const struct file *file)
1237 {
1238 return d_real(file->f_path.dentry, file_inode(file), 0, 0);
1239 }
1240
1241 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1242 {
1243 return locks_lock_inode_wait(locks_inode(filp), fl);
1244 }
1245
1246 struct fasync_struct {
1247 spinlock_t fa_lock;
1248 int magic;
1249 int fa_fd;
1250 struct fasync_struct *fa_next; /* singly linked list */
1251 struct file *fa_file;
1252 struct rcu_head fa_rcu;
1253 };
1254
1255 #define FASYNC_MAGIC 0x4601
1256
1257 /* SMP safe fasync helpers: */
1258 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1259 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1260 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1261 extern struct fasync_struct *fasync_alloc(void);
1262 extern void fasync_free(struct fasync_struct *);
1263
1264 /* can be called from interrupts */
1265 extern void kill_fasync(struct fasync_struct **, int, int);
1266
1267 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1268 extern int f_setown(struct file *filp, unsigned long arg, int force);
1269 extern void f_delown(struct file *filp);
1270 extern pid_t f_getown(struct file *filp);
1271 extern int send_sigurg(struct fown_struct *fown);
1272
1273 /*
1274 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1275 * represented in both.
1276 */
1277 #define SB_RDONLY 1 /* Mount read-only */
1278 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1279 #define SB_NODEV 4 /* Disallow access to device special files */
1280 #define SB_NOEXEC 8 /* Disallow program execution */
1281 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1282 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1283 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1284 #define SB_NOATIME 1024 /* Do not update access times. */
1285 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1286 #define SB_SILENT 32768
1287 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1288 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1289 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1290 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1291
1292 /* These sb flags are internal to the kernel */
1293 #define SB_SUBMOUNT (1<<26)
1294 #define SB_NOREMOTELOCK (1<<27)
1295 #define SB_NOSEC (1<<28)
1296 #define SB_BORN (1<<29)
1297 #define SB_ACTIVE (1<<30)
1298 #define SB_NOUSER (1<<31)
1299
1300 /*
1301 * Umount options
1302 */
1303
1304 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1305 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1306 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1307 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1308 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1309
1310 /* sb->s_iflags */
1311 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1312 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1313 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1314
1315 /* sb->s_iflags to limit user namespace mounts */
1316 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1317
1318 /* Possible states of 'frozen' field */
1319 enum {
1320 SB_UNFROZEN = 0, /* FS is unfrozen */
1321 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1322 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1323 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1324 * internal threads if needed) */
1325 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1326 };
1327
1328 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1329
1330 struct sb_writers {
1331 int frozen; /* Is sb frozen? */
1332 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1333 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1334 };
1335
1336 struct super_block {
1337 struct list_head s_list; /* Keep this first */
1338 dev_t s_dev; /* search index; _not_ kdev_t */
1339 unsigned char s_blocksize_bits;
1340 unsigned long s_blocksize;
1341 loff_t s_maxbytes; /* Max file size */
1342 struct file_system_type *s_type;
1343 const struct super_operations *s_op;
1344 const struct dquot_operations *dq_op;
1345 const struct quotactl_ops *s_qcop;
1346 const struct export_operations *s_export_op;
1347 unsigned long s_flags;
1348 unsigned long s_iflags; /* internal SB_I_* flags */
1349 unsigned long s_magic;
1350 struct dentry *s_root;
1351 struct rw_semaphore s_umount;
1352 int s_count;
1353 atomic_t s_active;
1354 #ifdef CONFIG_SECURITY
1355 void *s_security;
1356 #endif
1357 const struct xattr_handler **s_xattr;
1358
1359 const struct fscrypt_operations *s_cop;
1360
1361 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1362 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1363 struct block_device *s_bdev;
1364 struct backing_dev_info *s_bdi;
1365 struct mtd_info *s_mtd;
1366 struct hlist_node s_instances;
1367 unsigned int s_quota_types; /* Bitmask of supported quota types */
1368 struct quota_info s_dquot; /* Diskquota specific options */
1369
1370 struct sb_writers s_writers;
1371
1372 char s_id[32]; /* Informational name */
1373 uuid_t s_uuid; /* UUID */
1374
1375 void *s_fs_info; /* Filesystem private info */
1376 unsigned int s_max_links;
1377 fmode_t s_mode;
1378
1379 /* Granularity of c/m/atime in ns.
1380 Cannot be worse than a second */
1381 u32 s_time_gran;
1382
1383 /*
1384 * The next field is for VFS *only*. No filesystems have any business
1385 * even looking at it. You had been warned.
1386 */
1387 struct mutex s_vfs_rename_mutex; /* Kludge */
1388
1389 /*
1390 * Filesystem subtype. If non-empty the filesystem type field
1391 * in /proc/mounts will be "type.subtype"
1392 */
1393 char *s_subtype;
1394
1395 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1396
1397 /*
1398 * Saved pool identifier for cleancache (-1 means none)
1399 */
1400 int cleancache_poolid;
1401
1402 struct shrinker s_shrink; /* per-sb shrinker handle */
1403
1404 /* Number of inodes with nlink == 0 but still referenced */
1405 atomic_long_t s_remove_count;
1406
1407 /* Being remounted read-only */
1408 int s_readonly_remount;
1409
1410 /* AIO completions deferred from interrupt context */
1411 struct workqueue_struct *s_dio_done_wq;
1412 struct hlist_head s_pins;
1413
1414 /*
1415 * Owning user namespace and default context in which to
1416 * interpret filesystem uids, gids, quotas, device nodes,
1417 * xattrs and security labels.
1418 */
1419 struct user_namespace *s_user_ns;
1420
1421 /*
1422 * Keep the lru lists last in the structure so they always sit on their
1423 * own individual cachelines.
1424 */
1425 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1426 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
1427 struct rcu_head rcu;
1428 struct work_struct destroy_work;
1429
1430 struct mutex s_sync_lock; /* sync serialisation lock */
1431
1432 /*
1433 * Indicates how deep in a filesystem stack this SB is
1434 */
1435 int s_stack_depth;
1436
1437 /* s_inode_list_lock protects s_inodes */
1438 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1439 struct list_head s_inodes; /* all inodes */
1440
1441 spinlock_t s_inode_wblist_lock;
1442 struct list_head s_inodes_wb; /* writeback inodes */
1443 } __randomize_layout;
1444
1445 /* Helper functions so that in most cases filesystems will
1446 * not need to deal directly with kuid_t and kgid_t and can
1447 * instead deal with the raw numeric values that are stored
1448 * in the filesystem.
1449 */
1450 static inline uid_t i_uid_read(const struct inode *inode)
1451 {
1452 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid);
1453 }
1454
1455 static inline gid_t i_gid_read(const struct inode *inode)
1456 {
1457 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid);
1458 }
1459
1460 static inline void i_uid_write(struct inode *inode, uid_t uid)
1461 {
1462 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid);
1463 }
1464
1465 static inline void i_gid_write(struct inode *inode, gid_t gid)
1466 {
1467 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid);
1468 }
1469
1470 extern struct timespec current_time(struct inode *inode);
1471
1472 /*
1473 * Snapshotting support.
1474 */
1475
1476 void __sb_end_write(struct super_block *sb, int level);
1477 int __sb_start_write(struct super_block *sb, int level, bool wait);
1478
1479 #define __sb_writers_acquired(sb, lev) \
1480 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1481 #define __sb_writers_release(sb, lev) \
1482 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1483
1484 /**
1485 * sb_end_write - drop write access to a superblock
1486 * @sb: the super we wrote to
1487 *
1488 * Decrement number of writers to the filesystem. Wake up possible waiters
1489 * wanting to freeze the filesystem.
1490 */
1491 static inline void sb_end_write(struct super_block *sb)
1492 {
1493 __sb_end_write(sb, SB_FREEZE_WRITE);
1494 }
1495
1496 /**
1497 * sb_end_pagefault - drop write access to a superblock from a page fault
1498 * @sb: the super we wrote to
1499 *
1500 * Decrement number of processes handling write page fault to the filesystem.
1501 * Wake up possible waiters wanting to freeze the filesystem.
1502 */
1503 static inline void sb_end_pagefault(struct super_block *sb)
1504 {
1505 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1506 }
1507
1508 /**
1509 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1510 * @sb: the super we wrote to
1511 *
1512 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1513 * waiters wanting to freeze the filesystem.
1514 */
1515 static inline void sb_end_intwrite(struct super_block *sb)
1516 {
1517 __sb_end_write(sb, SB_FREEZE_FS);
1518 }
1519
1520 /**
1521 * sb_start_write - get write access to a superblock
1522 * @sb: the super we write to
1523 *
1524 * When a process wants to write data or metadata to a file system (i.e. dirty
1525 * a page or an inode), it should embed the operation in a sb_start_write() -
1526 * sb_end_write() pair to get exclusion against file system freezing. This
1527 * function increments number of writers preventing freezing. If the file
1528 * system is already frozen, the function waits until the file system is
1529 * thawed.
1530 *
1531 * Since freeze protection behaves as a lock, users have to preserve
1532 * ordering of freeze protection and other filesystem locks. Generally,
1533 * freeze protection should be the outermost lock. In particular, we have:
1534 *
1535 * sb_start_write
1536 * -> i_mutex (write path, truncate, directory ops, ...)
1537 * -> s_umount (freeze_super, thaw_super)
1538 */
1539 static inline void sb_start_write(struct super_block *sb)
1540 {
1541 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1542 }
1543
1544 static inline int sb_start_write_trylock(struct super_block *sb)
1545 {
1546 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1547 }
1548
1549 /**
1550 * sb_start_pagefault - get write access to a superblock from a page fault
1551 * @sb: the super we write to
1552 *
1553 * When a process starts handling write page fault, it should embed the
1554 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1555 * exclusion against file system freezing. This is needed since the page fault
1556 * is going to dirty a page. This function increments number of running page
1557 * faults preventing freezing. If the file system is already frozen, the
1558 * function waits until the file system is thawed.
1559 *
1560 * Since page fault freeze protection behaves as a lock, users have to preserve
1561 * ordering of freeze protection and other filesystem locks. It is advised to
1562 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1563 * handling code implies lock dependency:
1564 *
1565 * mmap_sem
1566 * -> sb_start_pagefault
1567 */
1568 static inline void sb_start_pagefault(struct super_block *sb)
1569 {
1570 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1571 }
1572
1573 /*
1574 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1575 * @sb: the super we write to
1576 *
1577 * This is the third level of protection against filesystem freezing. It is
1578 * free for use by a filesystem. The only requirement is that it must rank
1579 * below sb_start_pagefault.
1580 *
1581 * For example filesystem can call sb_start_intwrite() when starting a
1582 * transaction which somewhat eases handling of freezing for internal sources
1583 * of filesystem changes (internal fs threads, discarding preallocation on file
1584 * close, etc.).
1585 */
1586 static inline void sb_start_intwrite(struct super_block *sb)
1587 {
1588 __sb_start_write(sb, SB_FREEZE_FS, true);
1589 }
1590
1591
1592 extern bool inode_owner_or_capable(const struct inode *inode);
1593
1594 /*
1595 * VFS helper functions..
1596 */
1597 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1598 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1599 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1600 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1601 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1602 extern int vfs_rmdir(struct inode *, struct dentry *);
1603 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1604 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1605 extern int vfs_whiteout(struct inode *, struct dentry *);
1606
1607 extern struct dentry *vfs_tmpfile(struct dentry *dentry, umode_t mode,
1608 int open_flag);
1609
1610 /*
1611 * VFS file helper functions.
1612 */
1613 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1614 umode_t mode);
1615 extern bool may_open_dev(const struct path *path);
1616 /*
1617 * VFS FS_IOC_FIEMAP helper definitions.
1618 */
1619 struct fiemap_extent_info {
1620 unsigned int fi_flags; /* Flags as passed from user */
1621 unsigned int fi_extents_mapped; /* Number of mapped extents */
1622 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1623 struct fiemap_extent __user *fi_extents_start; /* Start of
1624 fiemap_extent array */
1625 };
1626 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1627 u64 phys, u64 len, u32 flags);
1628 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1629
1630 /*
1631 * File types
1632 *
1633 * NOTE! These match bits 12..15 of stat.st_mode
1634 * (ie "(i_mode >> 12) & 15").
1635 */
1636 #define DT_UNKNOWN 0
1637 #define DT_FIFO 1
1638 #define DT_CHR 2
1639 #define DT_DIR 4
1640 #define DT_BLK 6
1641 #define DT_REG 8
1642 #define DT_LNK 10
1643 #define DT_SOCK 12
1644 #define DT_WHT 14
1645
1646 /*
1647 * This is the "filldir" function type, used by readdir() to let
1648 * the kernel specify what kind of dirent layout it wants to have.
1649 * This allows the kernel to read directories into kernel space or
1650 * to have different dirent layouts depending on the binary type.
1651 */
1652 struct dir_context;
1653 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1654 unsigned);
1655
1656 struct dir_context {
1657 const filldir_t actor;
1658 loff_t pos;
1659 };
1660
1661 struct block_device_operations;
1662
1663 /* These macros are for out of kernel modules to test that
1664 * the kernel supports the unlocked_ioctl and compat_ioctl
1665 * fields in struct file_operations. */
1666 #define HAVE_COMPAT_IOCTL 1
1667 #define HAVE_UNLOCKED_IOCTL 1
1668
1669 /*
1670 * These flags let !MMU mmap() govern direct device mapping vs immediate
1671 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1672 *
1673 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1674 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1675 * NOMMU_MAP_READ: Can be mapped for reading
1676 * NOMMU_MAP_WRITE: Can be mapped for writing
1677 * NOMMU_MAP_EXEC: Can be mapped for execution
1678 */
1679 #define NOMMU_MAP_COPY 0x00000001
1680 #define NOMMU_MAP_DIRECT 0x00000008
1681 #define NOMMU_MAP_READ VM_MAYREAD
1682 #define NOMMU_MAP_WRITE VM_MAYWRITE
1683 #define NOMMU_MAP_EXEC VM_MAYEXEC
1684
1685 #define NOMMU_VMFLAGS \
1686 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1687
1688
1689 struct iov_iter;
1690
1691 struct file_operations {
1692 struct module *owner;
1693 loff_t (*llseek) (struct file *, loff_t, int);
1694 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1695 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1696 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1697 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1698 int (*iterate) (struct file *, struct dir_context *);
1699 int (*iterate_shared) (struct file *, struct dir_context *);
1700 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1701 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1702 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1703 int (*mmap) (struct file *, struct vm_area_struct *);
1704 int (*open) (struct inode *, struct file *);
1705 int (*flush) (struct file *, fl_owner_t id);
1706 int (*release) (struct inode *, struct file *);
1707 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1708 int (*fasync) (int, struct file *, int);
1709 int (*lock) (struct file *, int, struct file_lock *);
1710 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1711 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1712 int (*check_flags)(int);
1713 int (*flock) (struct file *, int, struct file_lock *);
1714 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1715 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1716 int (*setlease)(struct file *, long, struct file_lock **, void **);
1717 long (*fallocate)(struct file *file, int mode, loff_t offset,
1718 loff_t len);
1719 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1720 #ifndef CONFIG_MMU
1721 unsigned (*mmap_capabilities)(struct file *);
1722 #endif
1723 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1724 loff_t, size_t, unsigned int);
1725 int (*clone_file_range)(struct file *, loff_t, struct file *, loff_t,
1726 u64);
1727 ssize_t (*dedupe_file_range)(struct file *, u64, u64, struct file *,
1728 u64);
1729 } __randomize_layout;
1730
1731 struct inode_operations {
1732 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1733 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
1734 int (*permission) (struct inode *, int);
1735 struct posix_acl * (*get_acl)(struct inode *, int);
1736
1737 int (*readlink) (struct dentry *, char __user *,int);
1738
1739 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1740 int (*link) (struct dentry *,struct inode *,struct dentry *);
1741 int (*unlink) (struct inode *,struct dentry *);
1742 int (*symlink) (struct inode *,struct dentry *,const char *);
1743 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1744 int (*rmdir) (struct inode *,struct dentry *);
1745 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1746 int (*rename) (struct inode *, struct dentry *,
1747 struct inode *, struct dentry *, unsigned int);
1748 int (*setattr) (struct dentry *, struct iattr *);
1749 int (*getattr) (const struct path *, struct kstat *, u32, unsigned int);
1750 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1751 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1752 u64 len);
1753 int (*update_time)(struct inode *, struct timespec *, int);
1754 int (*atomic_open)(struct inode *, struct dentry *,
1755 struct file *, unsigned open_flag,
1756 umode_t create_mode, int *opened);
1757 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1758 int (*set_acl)(struct inode *, struct posix_acl *, int);
1759 } ____cacheline_aligned;
1760
1761 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
1762 struct iov_iter *iter)
1763 {
1764 return file->f_op->read_iter(kio, iter);
1765 }
1766
1767 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
1768 struct iov_iter *iter)
1769 {
1770 return file->f_op->write_iter(kio, iter);
1771 }
1772
1773 static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
1774 {
1775 return file->f_op->mmap(file, vma);
1776 }
1777
1778 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1779 unsigned long nr_segs, unsigned long fast_segs,
1780 struct iovec *fast_pointer,
1781 struct iovec **ret_pointer);
1782
1783 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1784 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1785 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1786 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1787 unsigned long, loff_t *, rwf_t);
1788 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1789 loff_t, size_t, unsigned int);
1790 extern int vfs_clone_file_prep_inodes(struct inode *inode_in, loff_t pos_in,
1791 struct inode *inode_out, loff_t pos_out,
1792 u64 *len, bool is_dedupe);
1793 extern int vfs_clone_file_range(struct file *file_in, loff_t pos_in,
1794 struct file *file_out, loff_t pos_out, u64 len);
1795 extern int vfs_dedupe_file_range_compare(struct inode *src, loff_t srcoff,
1796 struct inode *dest, loff_t destoff,
1797 loff_t len, bool *is_same);
1798 extern int vfs_dedupe_file_range(struct file *file,
1799 struct file_dedupe_range *same);
1800
1801 struct super_operations {
1802 struct inode *(*alloc_inode)(struct super_block *sb);
1803 void (*destroy_inode)(struct inode *);
1804
1805 void (*dirty_inode) (struct inode *, int flags);
1806 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1807 int (*drop_inode) (struct inode *);
1808 void (*evict_inode) (struct inode *);
1809 void (*put_super) (struct super_block *);
1810 int (*sync_fs)(struct super_block *sb, int wait);
1811 int (*freeze_super) (struct super_block *);
1812 int (*freeze_fs) (struct super_block *);
1813 int (*thaw_super) (struct super_block *);
1814 int (*unfreeze_fs) (struct super_block *);
1815 int (*statfs) (struct dentry *, struct kstatfs *);
1816 int (*remount_fs) (struct super_block *, int *, char *);
1817 void (*umount_begin) (struct super_block *);
1818
1819 int (*show_options)(struct seq_file *, struct dentry *);
1820 int (*show_devname)(struct seq_file *, struct dentry *);
1821 int (*show_path)(struct seq_file *, struct dentry *);
1822 int (*show_stats)(struct seq_file *, struct dentry *);
1823 #ifdef CONFIG_QUOTA
1824 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1825 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1826 struct dquot **(*get_dquots)(struct inode *);
1827 #endif
1828 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1829 long (*nr_cached_objects)(struct super_block *,
1830 struct shrink_control *);
1831 long (*free_cached_objects)(struct super_block *,
1832 struct shrink_control *);
1833 };
1834
1835 /*
1836 * Inode flags - they have no relation to superblock flags now
1837 */
1838 #define S_SYNC 1 /* Writes are synced at once */
1839 #define S_NOATIME 2 /* Do not update access times */
1840 #define S_APPEND 4 /* Append-only file */
1841 #define S_IMMUTABLE 8 /* Immutable file */
1842 #define S_DEAD 16 /* removed, but still open directory */
1843 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1844 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1845 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1846 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1847 #define S_PRIVATE 512 /* Inode is fs-internal */
1848 #define S_IMA 1024 /* Inode has an associated IMA struct */
1849 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1850 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1851 #ifdef CONFIG_FS_DAX
1852 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1853 #else
1854 #define S_DAX 0 /* Make all the DAX code disappear */
1855 #endif
1856
1857 /*
1858 * Note that nosuid etc flags are inode-specific: setting some file-system
1859 * flags just means all the inodes inherit those flags by default. It might be
1860 * possible to override it selectively if you really wanted to with some
1861 * ioctl() that is not currently implemented.
1862 *
1863 * Exception: SB_RDONLY is always applied to the entire file system.
1864 *
1865 * Unfortunately, it is possible to change a filesystems flags with it mounted
1866 * with files in use. This means that all of the inodes will not have their
1867 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1868 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1869 */
1870 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1871
1872 static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & MS_RDONLY; }
1873 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
1874 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
1875 ((inode)->i_flags & S_SYNC))
1876 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
1877 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1878 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
1879 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
1880 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
1881
1882 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1883 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1884 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1885 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
1886
1887 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1888 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1889 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1890 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1891 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1892 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1893 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1894 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1895
1896 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1897 (inode)->i_rdev == WHITEOUT_DEV)
1898
1899 static inline bool HAS_UNMAPPED_ID(struct inode *inode)
1900 {
1901 return !uid_valid(inode->i_uid) || !gid_valid(inode->i_gid);
1902 }
1903
1904 static inline enum rw_hint file_write_hint(struct file *file)
1905 {
1906 if (file->f_write_hint != WRITE_LIFE_NOT_SET)
1907 return file->f_write_hint;
1908
1909 return file_inode(file)->i_write_hint;
1910 }
1911
1912 static inline int iocb_flags(struct file *file);
1913
1914 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
1915 {
1916 *kiocb = (struct kiocb) {
1917 .ki_filp = filp,
1918 .ki_flags = iocb_flags(filp),
1919 .ki_hint = file_write_hint(filp),
1920 };
1921 }
1922
1923 /*
1924 * Inode state bits. Protected by inode->i_lock
1925 *
1926 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1927 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1928 *
1929 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1930 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1931 * various stages of removing an inode.
1932 *
1933 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1934 *
1935 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1936 * fdatasync(). i_atime is the usual cause.
1937 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1938 * these changes separately from I_DIRTY_SYNC so that we
1939 * don't have to write inode on fdatasync() when only
1940 * mtime has changed in it.
1941 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1942 * I_NEW Serves as both a mutex and completion notification.
1943 * New inodes set I_NEW. If two processes both create
1944 * the same inode, one of them will release its inode and
1945 * wait for I_NEW to be released before returning.
1946 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1947 * also cause waiting on I_NEW, without I_NEW actually
1948 * being set. find_inode() uses this to prevent returning
1949 * nearly-dead inodes.
1950 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1951 * is zero. I_FREEING must be set when I_WILL_FREE is
1952 * cleared.
1953 * I_FREEING Set when inode is about to be freed but still has dirty
1954 * pages or buffers attached or the inode itself is still
1955 * dirty.
1956 * I_CLEAR Added by clear_inode(). In this state the inode is
1957 * clean and can be destroyed. Inode keeps I_FREEING.
1958 *
1959 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1960 * prohibited for many purposes. iget() must wait for
1961 * the inode to be completely released, then create it
1962 * anew. Other functions will just ignore such inodes,
1963 * if appropriate. I_NEW is used for waiting.
1964 *
1965 * I_SYNC Writeback of inode is running. The bit is set during
1966 * data writeback, and cleared with a wakeup on the bit
1967 * address once it is done. The bit is also used to pin
1968 * the inode in memory for flusher thread.
1969 *
1970 * I_REFERENCED Marks the inode as recently references on the LRU list.
1971 *
1972 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1973 *
1974 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
1975 * synchronize competing switching instances and to tell
1976 * wb stat updates to grab mapping->tree_lock. See
1977 * inode_switch_wb_work_fn() for details.
1978 *
1979 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
1980 * and work dirs among overlayfs mounts.
1981 *
1982 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1983 */
1984 #define I_DIRTY_SYNC (1 << 0)
1985 #define I_DIRTY_DATASYNC (1 << 1)
1986 #define I_DIRTY_PAGES (1 << 2)
1987 #define __I_NEW 3
1988 #define I_NEW (1 << __I_NEW)
1989 #define I_WILL_FREE (1 << 4)
1990 #define I_FREEING (1 << 5)
1991 #define I_CLEAR (1 << 6)
1992 #define __I_SYNC 7
1993 #define I_SYNC (1 << __I_SYNC)
1994 #define I_REFERENCED (1 << 8)
1995 #define __I_DIO_WAKEUP 9
1996 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
1997 #define I_LINKABLE (1 << 10)
1998 #define I_DIRTY_TIME (1 << 11)
1999 #define __I_DIRTY_TIME_EXPIRED 12
2000 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
2001 #define I_WB_SWITCH (1 << 13)
2002 #define I_OVL_INUSE (1 << 14)
2003
2004 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
2005 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2006
2007 extern void __mark_inode_dirty(struct inode *, int);
2008 static inline void mark_inode_dirty(struct inode *inode)
2009 {
2010 __mark_inode_dirty(inode, I_DIRTY);
2011 }
2012
2013 static inline void mark_inode_dirty_sync(struct inode *inode)
2014 {
2015 __mark_inode_dirty(inode, I_DIRTY_SYNC);
2016 }
2017
2018 extern void inc_nlink(struct inode *inode);
2019 extern void drop_nlink(struct inode *inode);
2020 extern void clear_nlink(struct inode *inode);
2021 extern void set_nlink(struct inode *inode, unsigned int nlink);
2022
2023 static inline void inode_inc_link_count(struct inode *inode)
2024 {
2025 inc_nlink(inode);
2026 mark_inode_dirty(inode);
2027 }
2028
2029 static inline void inode_dec_link_count(struct inode *inode)
2030 {
2031 drop_nlink(inode);
2032 mark_inode_dirty(inode);
2033 }
2034
2035 /**
2036 * inode_inc_iversion - increments i_version
2037 * @inode: inode that need to be updated
2038 *
2039 * Every time the inode is modified, the i_version field will be incremented.
2040 * The filesystem has to be mounted with i_version flag
2041 */
2042
2043 static inline void inode_inc_iversion(struct inode *inode)
2044 {
2045 spin_lock(&inode->i_lock);
2046 inode->i_version++;
2047 spin_unlock(&inode->i_lock);
2048 }
2049
2050 enum file_time_flags {
2051 S_ATIME = 1,
2052 S_MTIME = 2,
2053 S_CTIME = 4,
2054 S_VERSION = 8,
2055 };
2056
2057 extern void touch_atime(const struct path *);
2058 static inline void file_accessed(struct file *file)
2059 {
2060 if (!(file->f_flags & O_NOATIME))
2061 touch_atime(&file->f_path);
2062 }
2063
2064 int sync_inode(struct inode *inode, struct writeback_control *wbc);
2065 int sync_inode_metadata(struct inode *inode, int wait);
2066
2067 struct file_system_type {
2068 const char *name;
2069 int fs_flags;
2070 #define FS_REQUIRES_DEV 1
2071 #define FS_BINARY_MOUNTDATA 2
2072 #define FS_HAS_SUBTYPE 4
2073 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2074 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2075 struct dentry *(*mount) (struct file_system_type *, int,
2076 const char *, void *);
2077 void (*kill_sb) (struct super_block *);
2078 struct module *owner;
2079 struct file_system_type * next;
2080 struct hlist_head fs_supers;
2081
2082 struct lock_class_key s_lock_key;
2083 struct lock_class_key s_umount_key;
2084 struct lock_class_key s_vfs_rename_key;
2085 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2086
2087 struct lock_class_key i_lock_key;
2088 struct lock_class_key i_mutex_key;
2089 struct lock_class_key i_mutex_dir_key;
2090 };
2091
2092 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2093
2094 extern struct dentry *mount_ns(struct file_system_type *fs_type,
2095 int flags, void *data, void *ns, struct user_namespace *user_ns,
2096 int (*fill_super)(struct super_block *, void *, int));
2097 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2098 int flags, const char *dev_name, void *data,
2099 int (*fill_super)(struct super_block *, void *, int));
2100 extern struct dentry *mount_single(struct file_system_type *fs_type,
2101 int flags, void *data,
2102 int (*fill_super)(struct super_block *, void *, int));
2103 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2104 int flags, void *data,
2105 int (*fill_super)(struct super_block *, void *, int));
2106 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2107 void generic_shutdown_super(struct super_block *sb);
2108 void kill_block_super(struct super_block *sb);
2109 void kill_anon_super(struct super_block *sb);
2110 void kill_litter_super(struct super_block *sb);
2111 void deactivate_super(struct super_block *sb);
2112 void deactivate_locked_super(struct super_block *sb);
2113 int set_anon_super(struct super_block *s, void *data);
2114 int get_anon_bdev(dev_t *);
2115 void free_anon_bdev(dev_t);
2116 struct super_block *sget_userns(struct file_system_type *type,
2117 int (*test)(struct super_block *,void *),
2118 int (*set)(struct super_block *,void *),
2119 int flags, struct user_namespace *user_ns,
2120 void *data);
2121 struct super_block *sget(struct file_system_type *type,
2122 int (*test)(struct super_block *,void *),
2123 int (*set)(struct super_block *,void *),
2124 int flags, void *data);
2125 extern struct dentry *mount_pseudo_xattr(struct file_system_type *, char *,
2126 const struct super_operations *ops,
2127 const struct xattr_handler **xattr,
2128 const struct dentry_operations *dops,
2129 unsigned long);
2130
2131 static inline struct dentry *
2132 mount_pseudo(struct file_system_type *fs_type, char *name,
2133 const struct super_operations *ops,
2134 const struct dentry_operations *dops, unsigned long magic)
2135 {
2136 return mount_pseudo_xattr(fs_type, name, ops, NULL, dops, magic);
2137 }
2138
2139 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2140 #define fops_get(fops) \
2141 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2142 #define fops_put(fops) \
2143 do { if (fops) module_put((fops)->owner); } while(0)
2144 /*
2145 * This one is to be used *ONLY* from ->open() instances.
2146 * fops must be non-NULL, pinned down *and* module dependencies
2147 * should be sufficient to pin the caller down as well.
2148 */
2149 #define replace_fops(f, fops) \
2150 do { \
2151 struct file *__file = (f); \
2152 fops_put(__file->f_op); \
2153 BUG_ON(!(__file->f_op = (fops))); \
2154 } while(0)
2155
2156 extern int register_filesystem(struct file_system_type *);
2157 extern int unregister_filesystem(struct file_system_type *);
2158 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
2159 #define kern_mount(type) kern_mount_data(type, NULL)
2160 extern void kern_unmount(struct vfsmount *mnt);
2161 extern int may_umount_tree(struct vfsmount *);
2162 extern int may_umount(struct vfsmount *);
2163 extern long do_mount(const char *, const char __user *,
2164 const char *, unsigned long, void *);
2165 extern struct vfsmount *collect_mounts(const struct path *);
2166 extern void drop_collected_mounts(struct vfsmount *);
2167 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2168 struct vfsmount *);
2169 extern int vfs_statfs(const struct path *, struct kstatfs *);
2170 extern int user_statfs(const char __user *, struct kstatfs *);
2171 extern int fd_statfs(int, struct kstatfs *);
2172 extern int vfs_ustat(dev_t, struct kstatfs *);
2173 extern int freeze_super(struct super_block *super);
2174 extern int thaw_super(struct super_block *super);
2175 extern bool our_mnt(struct vfsmount *mnt);
2176 extern __printf(2, 3)
2177 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2178 extern int super_setup_bdi(struct super_block *sb);
2179
2180 extern int current_umask(void);
2181
2182 extern void ihold(struct inode * inode);
2183 extern void iput(struct inode *);
2184 extern int generic_update_time(struct inode *, struct timespec *, int);
2185
2186 /* /sys/fs */
2187 extern struct kobject *fs_kobj;
2188
2189 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2190
2191 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2192 extern int locks_mandatory_locked(struct file *);
2193 extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
2194
2195 /*
2196 * Candidates for mandatory locking have the setgid bit set
2197 * but no group execute bit - an otherwise meaningless combination.
2198 */
2199
2200 static inline int __mandatory_lock(struct inode *ino)
2201 {
2202 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2203 }
2204
2205 /*
2206 * ... and these candidates should be on SB_MANDLOCK mounted fs,
2207 * otherwise these will be advisory locks
2208 */
2209
2210 static inline int mandatory_lock(struct inode *ino)
2211 {
2212 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2213 }
2214
2215 static inline int locks_verify_locked(struct file *file)
2216 {
2217 if (mandatory_lock(locks_inode(file)))
2218 return locks_mandatory_locked(file);
2219 return 0;
2220 }
2221
2222 static inline int locks_verify_truncate(struct inode *inode,
2223 struct file *f,
2224 loff_t size)
2225 {
2226 if (!inode->i_flctx || !mandatory_lock(inode))
2227 return 0;
2228
2229 if (size < inode->i_size) {
2230 return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2231 F_WRLCK);
2232 } else {
2233 return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2234 F_WRLCK);
2235 }
2236 }
2237
2238 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2239
2240 static inline int locks_mandatory_locked(struct file *file)
2241 {
2242 return 0;
2243 }
2244
2245 static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2246 loff_t start, loff_t end, unsigned char type)
2247 {
2248 return 0;
2249 }
2250
2251 static inline int __mandatory_lock(struct inode *inode)
2252 {
2253 return 0;
2254 }
2255
2256 static inline int mandatory_lock(struct inode *inode)
2257 {
2258 return 0;
2259 }
2260
2261 static inline int locks_verify_locked(struct file *file)
2262 {
2263 return 0;
2264 }
2265
2266 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2267 size_t size)
2268 {
2269 return 0;
2270 }
2271
2272 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2273
2274
2275 #ifdef CONFIG_FILE_LOCKING
2276 static inline int break_lease(struct inode *inode, unsigned int mode)
2277 {
2278 /*
2279 * Since this check is lockless, we must ensure that any refcounts
2280 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2281 * could end up racing with tasks trying to set a new lease on this
2282 * file.
2283 */
2284 smp_mb();
2285 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2286 return __break_lease(inode, mode, FL_LEASE);
2287 return 0;
2288 }
2289
2290 static inline int break_deleg(struct inode *inode, unsigned int mode)
2291 {
2292 /*
2293 * Since this check is lockless, we must ensure that any refcounts
2294 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2295 * could end up racing with tasks trying to set a new lease on this
2296 * file.
2297 */
2298 smp_mb();
2299 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2300 return __break_lease(inode, mode, FL_DELEG);
2301 return 0;
2302 }
2303
2304 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2305 {
2306 int ret;
2307
2308 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2309 if (ret == -EWOULDBLOCK && delegated_inode) {
2310 *delegated_inode = inode;
2311 ihold(inode);
2312 }
2313 return ret;
2314 }
2315
2316 static inline int break_deleg_wait(struct inode **delegated_inode)
2317 {
2318 int ret;
2319
2320 ret = break_deleg(*delegated_inode, O_WRONLY);
2321 iput(*delegated_inode);
2322 *delegated_inode = NULL;
2323 return ret;
2324 }
2325
2326 static inline int break_layout(struct inode *inode, bool wait)
2327 {
2328 smp_mb();
2329 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2330 return __break_lease(inode,
2331 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2332 FL_LAYOUT);
2333 return 0;
2334 }
2335
2336 #else /* !CONFIG_FILE_LOCKING */
2337 static inline int break_lease(struct inode *inode, unsigned int mode)
2338 {
2339 return 0;
2340 }
2341
2342 static inline int break_deleg(struct inode *inode, unsigned int mode)
2343 {
2344 return 0;
2345 }
2346
2347 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2348 {
2349 return 0;
2350 }
2351
2352 static inline int break_deleg_wait(struct inode **delegated_inode)
2353 {
2354 BUG();
2355 return 0;
2356 }
2357
2358 static inline int break_layout(struct inode *inode, bool wait)
2359 {
2360 return 0;
2361 }
2362
2363 #endif /* CONFIG_FILE_LOCKING */
2364
2365 /* fs/open.c */
2366 struct audit_names;
2367 struct filename {
2368 const char *name; /* pointer to actual string */
2369 const __user char *uptr; /* original userland pointer */
2370 struct audit_names *aname;
2371 int refcnt;
2372 const char iname[];
2373 };
2374
2375 extern long vfs_truncate(const struct path *, loff_t);
2376 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2377 struct file *filp);
2378 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2379 loff_t len);
2380 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2381 umode_t mode);
2382 extern struct file *file_open_name(struct filename *, int, umode_t);
2383 extern struct file *filp_open(const char *, int, umode_t);
2384 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2385 const char *, int, umode_t);
2386 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2387 extern int filp_close(struct file *, fl_owner_t id);
2388
2389 extern struct filename *getname_flags(const char __user *, int, int *);
2390 extern struct filename *getname(const char __user *);
2391 extern struct filename *getname_kernel(const char *);
2392 extern void putname(struct filename *name);
2393
2394 enum {
2395 FILE_CREATED = 1,
2396 FILE_OPENED = 2
2397 };
2398 extern int finish_open(struct file *file, struct dentry *dentry,
2399 int (*open)(struct inode *, struct file *),
2400 int *opened);
2401 extern int finish_no_open(struct file *file, struct dentry *dentry);
2402
2403 /* fs/ioctl.c */
2404
2405 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2406
2407 /* fs/dcache.c */
2408 extern void __init vfs_caches_init_early(void);
2409 extern void __init vfs_caches_init(void);
2410
2411 extern struct kmem_cache *names_cachep;
2412
2413 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2414 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2415
2416 #ifdef CONFIG_BLOCK
2417 extern int register_blkdev(unsigned int, const char *);
2418 extern void unregister_blkdev(unsigned int, const char *);
2419 extern void bdev_unhash_inode(dev_t dev);
2420 extern struct block_device *bdget(dev_t);
2421 extern struct block_device *bdgrab(struct block_device *bdev);
2422 extern void bd_set_size(struct block_device *, loff_t size);
2423 extern void bd_forget(struct inode *inode);
2424 extern void bdput(struct block_device *);
2425 extern void invalidate_bdev(struct block_device *);
2426 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2427 extern int sync_blockdev(struct block_device *bdev);
2428 extern void kill_bdev(struct block_device *);
2429 extern struct super_block *freeze_bdev(struct block_device *);
2430 extern void emergency_thaw_all(void);
2431 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2432 extern int fsync_bdev(struct block_device *);
2433
2434 extern struct super_block *blockdev_superblock;
2435
2436 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2437 {
2438 return sb == blockdev_superblock;
2439 }
2440 #else
2441 static inline void bd_forget(struct inode *inode) {}
2442 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2443 static inline void kill_bdev(struct block_device *bdev) {}
2444 static inline void invalidate_bdev(struct block_device *bdev) {}
2445
2446 static inline struct super_block *freeze_bdev(struct block_device *sb)
2447 {
2448 return NULL;
2449 }
2450
2451 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2452 {
2453 return 0;
2454 }
2455
2456 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2457 {
2458 }
2459
2460 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2461 {
2462 return false;
2463 }
2464 #endif
2465 extern int sync_filesystem(struct super_block *);
2466 extern const struct file_operations def_blk_fops;
2467 extern const struct file_operations def_chr_fops;
2468 #ifdef CONFIG_BLOCK
2469 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2470 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2471 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2472 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2473 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2474 void *holder);
2475 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2476 void *holder);
2477 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2478 extern int __blkdev_reread_part(struct block_device *bdev);
2479 extern int blkdev_reread_part(struct block_device *bdev);
2480
2481 #ifdef CONFIG_SYSFS
2482 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2483 extern void bd_unlink_disk_holder(struct block_device *bdev,
2484 struct gendisk *disk);
2485 #else
2486 static inline int bd_link_disk_holder(struct block_device *bdev,
2487 struct gendisk *disk)
2488 {
2489 return 0;
2490 }
2491 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2492 struct gendisk *disk)
2493 {
2494 }
2495 #endif
2496 #endif
2497
2498 /* fs/char_dev.c */
2499 #define CHRDEV_MAJOR_MAX 512
2500 /* Marks the bottom of the first segment of free char majors */
2501 #define CHRDEV_MAJOR_DYN_END 234
2502 /* Marks the top and bottom of the second segment of free char majors */
2503 #define CHRDEV_MAJOR_DYN_EXT_START 511
2504 #define CHRDEV_MAJOR_DYN_EXT_END 384
2505
2506 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2507 extern int register_chrdev_region(dev_t, unsigned, const char *);
2508 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2509 unsigned int count, const char *name,
2510 const struct file_operations *fops);
2511 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2512 unsigned int count, const char *name);
2513 extern void unregister_chrdev_region(dev_t, unsigned);
2514 extern void chrdev_show(struct seq_file *,off_t);
2515
2516 static inline int register_chrdev(unsigned int major, const char *name,
2517 const struct file_operations *fops)
2518 {
2519 return __register_chrdev(major, 0, 256, name, fops);
2520 }
2521
2522 static inline void unregister_chrdev(unsigned int major, const char *name)
2523 {
2524 __unregister_chrdev(major, 0, 256, name);
2525 }
2526
2527 /* fs/block_dev.c */
2528 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2529 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2530
2531 #ifdef CONFIG_BLOCK
2532 #define BLKDEV_MAJOR_MAX 512
2533 extern const char *__bdevname(dev_t, char *buffer);
2534 extern const char *bdevname(struct block_device *bdev, char *buffer);
2535 extern struct block_device *lookup_bdev(const char *);
2536 extern void blkdev_show(struct seq_file *,off_t);
2537
2538 #else
2539 #define BLKDEV_MAJOR_MAX 0
2540 #endif
2541
2542 extern void init_special_inode(struct inode *, umode_t, dev_t);
2543
2544 /* Invalid inode operations -- fs/bad_inode.c */
2545 extern void make_bad_inode(struct inode *);
2546 extern bool is_bad_inode(struct inode *);
2547
2548 #ifdef CONFIG_BLOCK
2549 extern void check_disk_size_change(struct gendisk *disk,
2550 struct block_device *bdev);
2551 extern int revalidate_disk(struct gendisk *);
2552 extern int check_disk_change(struct block_device *);
2553 extern int __invalidate_device(struct block_device *, bool);
2554 extern int invalidate_partition(struct gendisk *, int);
2555 #endif
2556 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2557 pgoff_t start, pgoff_t end);
2558
2559 static inline void invalidate_remote_inode(struct inode *inode)
2560 {
2561 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2562 S_ISLNK(inode->i_mode))
2563 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2564 }
2565 extern int invalidate_inode_pages2(struct address_space *mapping);
2566 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2567 pgoff_t start, pgoff_t end);
2568 extern int write_inode_now(struct inode *, int);
2569 extern int filemap_fdatawrite(struct address_space *);
2570 extern int filemap_flush(struct address_space *);
2571 extern int filemap_fdatawait_keep_errors(struct address_space *mapping);
2572 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2573 loff_t lend);
2574
2575 static inline int filemap_fdatawait(struct address_space *mapping)
2576 {
2577 return filemap_fdatawait_range(mapping, 0, LLONG_MAX);
2578 }
2579
2580 extern bool filemap_range_has_page(struct address_space *, loff_t lstart,
2581 loff_t lend);
2582 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2583 loff_t lend);
2584 extern int filemap_write_and_wait(struct address_space *mapping);
2585 extern int filemap_write_and_wait_range(struct address_space *mapping,
2586 loff_t lstart, loff_t lend);
2587 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2588 loff_t start, loff_t end, int sync_mode);
2589 extern int filemap_fdatawrite_range(struct address_space *mapping,
2590 loff_t start, loff_t end);
2591 extern int filemap_check_errors(struct address_space *mapping);
2592 extern void __filemap_set_wb_err(struct address_space *mapping, int err);
2593
2594 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2595 loff_t lend);
2596 extern int __must_check file_check_and_advance_wb_err(struct file *file);
2597 extern int __must_check file_write_and_wait_range(struct file *file,
2598 loff_t start, loff_t end);
2599
2600 static inline int file_write_and_wait(struct file *file)
2601 {
2602 return file_write_and_wait_range(file, 0, LLONG_MAX);
2603 }
2604
2605 /**
2606 * filemap_set_wb_err - set a writeback error on an address_space
2607 * @mapping: mapping in which to set writeback error
2608 * @err: error to be set in mapping
2609 *
2610 * When writeback fails in some way, we must record that error so that
2611 * userspace can be informed when fsync and the like are called. We endeavor
2612 * to report errors on any file that was open at the time of the error. Some
2613 * internal callers also need to know when writeback errors have occurred.
2614 *
2615 * When a writeback error occurs, most filesystems will want to call
2616 * filemap_set_wb_err to record the error in the mapping so that it will be
2617 * automatically reported whenever fsync is called on the file.
2618 */
2619 static inline void filemap_set_wb_err(struct address_space *mapping, int err)
2620 {
2621 /* Fastpath for common case of no error */
2622 if (unlikely(err))
2623 __filemap_set_wb_err(mapping, err);
2624 }
2625
2626 /**
2627 * filemap_check_wb_error - has an error occurred since the mark was sampled?
2628 * @mapping: mapping to check for writeback errors
2629 * @since: previously-sampled errseq_t
2630 *
2631 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2632 * the given value was sampled.
2633 *
2634 * If it has then report the latest error set, otherwise return 0.
2635 */
2636 static inline int filemap_check_wb_err(struct address_space *mapping,
2637 errseq_t since)
2638 {
2639 return errseq_check(&mapping->wb_err, since);
2640 }
2641
2642 /**
2643 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2644 * @mapping: mapping to be sampled
2645 *
2646 * Writeback errors are always reported relative to a particular sample point
2647 * in the past. This function provides those sample points.
2648 */
2649 static inline errseq_t filemap_sample_wb_err(struct address_space *mapping)
2650 {
2651 return errseq_sample(&mapping->wb_err);
2652 }
2653
2654 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2655 int datasync);
2656 extern int vfs_fsync(struct file *file, int datasync);
2657
2658 /*
2659 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2660 * to already be updated for the write, and will return either the amount
2661 * of bytes passed in, or an error if syncing the file failed.
2662 */
2663 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2664 {
2665 if (iocb->ki_flags & IOCB_DSYNC) {
2666 int ret = vfs_fsync_range(iocb->ki_filp,
2667 iocb->ki_pos - count, iocb->ki_pos - 1,
2668 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2669 if (ret)
2670 return ret;
2671 }
2672
2673 return count;
2674 }
2675
2676 extern void emergency_sync(void);
2677 extern void emergency_remount(void);
2678 #ifdef CONFIG_BLOCK
2679 extern sector_t bmap(struct inode *, sector_t);
2680 #endif
2681 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2682 extern int inode_permission(struct inode *, int);
2683 extern int __inode_permission(struct inode *, int);
2684 extern int generic_permission(struct inode *, int);
2685 extern int __check_sticky(struct inode *dir, struct inode *inode);
2686
2687 static inline bool execute_ok(struct inode *inode)
2688 {
2689 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2690 }
2691
2692 static inline void file_start_write(struct file *file)
2693 {
2694 if (!S_ISREG(file_inode(file)->i_mode))
2695 return;
2696 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2697 }
2698
2699 static inline bool file_start_write_trylock(struct file *file)
2700 {
2701 if (!S_ISREG(file_inode(file)->i_mode))
2702 return true;
2703 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2704 }
2705
2706 static inline void file_end_write(struct file *file)
2707 {
2708 if (!S_ISREG(file_inode(file)->i_mode))
2709 return;
2710 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2711 }
2712
2713 static inline int do_clone_file_range(struct file *file_in, loff_t pos_in,
2714 struct file *file_out, loff_t pos_out,
2715 u64 len)
2716 {
2717 int ret;
2718
2719 file_start_write(file_out);
2720 ret = vfs_clone_file_range(file_in, pos_in, file_out, pos_out, len);
2721 file_end_write(file_out);
2722
2723 return ret;
2724 }
2725
2726 /*
2727 * get_write_access() gets write permission for a file.
2728 * put_write_access() releases this write permission.
2729 * This is used for regular files.
2730 * We cannot support write (and maybe mmap read-write shared) accesses and
2731 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2732 * can have the following values:
2733 * 0: no writers, no VM_DENYWRITE mappings
2734 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2735 * > 0: (i_writecount) users are writing to the file.
2736 *
2737 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2738 * except for the cases where we don't hold i_writecount yet. Then we need to
2739 * use {get,deny}_write_access() - these functions check the sign and refuse
2740 * to do the change if sign is wrong.
2741 */
2742 static inline int get_write_access(struct inode *inode)
2743 {
2744 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2745 }
2746 static inline int deny_write_access(struct file *file)
2747 {
2748 struct inode *inode = file_inode(file);
2749 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2750 }
2751 static inline void put_write_access(struct inode * inode)
2752 {
2753 atomic_dec(&inode->i_writecount);
2754 }
2755 static inline void allow_write_access(struct file *file)
2756 {
2757 if (file)
2758 atomic_inc(&file_inode(file)->i_writecount);
2759 }
2760 static inline bool inode_is_open_for_write(const struct inode *inode)
2761 {
2762 return atomic_read(&inode->i_writecount) > 0;
2763 }
2764
2765 #ifdef CONFIG_IMA
2766 static inline void i_readcount_dec(struct inode *inode)
2767 {
2768 BUG_ON(!atomic_read(&inode->i_readcount));
2769 atomic_dec(&inode->i_readcount);
2770 }
2771 static inline void i_readcount_inc(struct inode *inode)
2772 {
2773 atomic_inc(&inode->i_readcount);
2774 }
2775 #else
2776 static inline void i_readcount_dec(struct inode *inode)
2777 {
2778 return;
2779 }
2780 static inline void i_readcount_inc(struct inode *inode)
2781 {
2782 return;
2783 }
2784 #endif
2785 extern int do_pipe_flags(int *, int);
2786
2787 #define __kernel_read_file_id(id) \
2788 id(UNKNOWN, unknown) \
2789 id(FIRMWARE, firmware) \
2790 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
2791 id(MODULE, kernel-module) \
2792 id(KEXEC_IMAGE, kexec-image) \
2793 id(KEXEC_INITRAMFS, kexec-initramfs) \
2794 id(POLICY, security-policy) \
2795 id(MAX_ID, )
2796
2797 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2798 #define __fid_stringify(dummy, str) #str,
2799
2800 enum kernel_read_file_id {
2801 __kernel_read_file_id(__fid_enumify)
2802 };
2803
2804 static const char * const kernel_read_file_str[] = {
2805 __kernel_read_file_id(__fid_stringify)
2806 };
2807
2808 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id)
2809 {
2810 if ((unsigned)id >= READING_MAX_ID)
2811 return kernel_read_file_str[READING_UNKNOWN];
2812
2813 return kernel_read_file_str[id];
2814 }
2815
2816 extern int kernel_read_file(struct file *, void **, loff_t *, loff_t,
2817 enum kernel_read_file_id);
2818 extern int kernel_read_file_from_path(const char *, void **, loff_t *, loff_t,
2819 enum kernel_read_file_id);
2820 extern int kernel_read_file_from_fd(int, void **, loff_t *, loff_t,
2821 enum kernel_read_file_id);
2822 extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
2823 extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
2824 extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
2825 extern struct file * open_exec(const char *);
2826
2827 /* fs/dcache.c -- generic fs support functions */
2828 extern bool is_subdir(struct dentry *, struct dentry *);
2829 extern bool path_is_under(const struct path *, const struct path *);
2830
2831 extern char *file_path(struct file *, char *, int);
2832
2833 #include <linux/err.h>
2834
2835 /* needed for stackable file system support */
2836 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2837
2838 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2839
2840 extern int inode_init_always(struct super_block *, struct inode *);
2841 extern void inode_init_once(struct inode *);
2842 extern void address_space_init_once(struct address_space *mapping);
2843 extern struct inode * igrab(struct inode *);
2844 extern ino_t iunique(struct super_block *, ino_t);
2845 extern int inode_needs_sync(struct inode *inode);
2846 extern int generic_delete_inode(struct inode *inode);
2847 static inline int generic_drop_inode(struct inode *inode)
2848 {
2849 return !inode->i_nlink || inode_unhashed(inode);
2850 }
2851
2852 extern struct inode *ilookup5_nowait(struct super_block *sb,
2853 unsigned long hashval, int (*test)(struct inode *, void *),
2854 void *data);
2855 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2856 int (*test)(struct inode *, void *), void *data);
2857 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2858
2859 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2860 extern struct inode * iget_locked(struct super_block *, unsigned long);
2861 extern struct inode *find_inode_nowait(struct super_block *,
2862 unsigned long,
2863 int (*match)(struct inode *,
2864 unsigned long, void *),
2865 void *data);
2866 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2867 extern int insert_inode_locked(struct inode *);
2868 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2869 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2870 #else
2871 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2872 #endif
2873 extern void unlock_new_inode(struct inode *);
2874 extern unsigned int get_next_ino(void);
2875 extern void evict_inodes(struct super_block *sb);
2876
2877 extern void __iget(struct inode * inode);
2878 extern void iget_failed(struct inode *);
2879 extern void clear_inode(struct inode *);
2880 extern void __destroy_inode(struct inode *);
2881 extern struct inode *new_inode_pseudo(struct super_block *sb);
2882 extern struct inode *new_inode(struct super_block *sb);
2883 extern void free_inode_nonrcu(struct inode *inode);
2884 extern int should_remove_suid(struct dentry *);
2885 extern int file_remove_privs(struct file *);
2886
2887 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2888 static inline void insert_inode_hash(struct inode *inode)
2889 {
2890 __insert_inode_hash(inode, inode->i_ino);
2891 }
2892
2893 extern void __remove_inode_hash(struct inode *);
2894 static inline void remove_inode_hash(struct inode *inode)
2895 {
2896 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
2897 __remove_inode_hash(inode);
2898 }
2899
2900 extern void inode_sb_list_add(struct inode *inode);
2901
2902 #ifdef CONFIG_BLOCK
2903 extern int bdev_read_only(struct block_device *);
2904 #endif
2905 extern int set_blocksize(struct block_device *, int);
2906 extern int sb_set_blocksize(struct super_block *, int);
2907 extern int sb_min_blocksize(struct super_block *, int);
2908
2909 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2910 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2911 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2912 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2913 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2914 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2915 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
2916 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2917
2918 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
2919 rwf_t flags);
2920 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
2921 rwf_t flags);
2922
2923 /* fs/block_dev.c */
2924 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
2925 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2926 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2927 int datasync);
2928 extern void block_sync_page(struct page *page);
2929
2930 /* fs/splice.c */
2931 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2932 struct pipe_inode_info *, size_t, unsigned int);
2933 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2934 struct file *, loff_t *, size_t, unsigned int);
2935 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2936 struct file *out, loff_t *, size_t len, unsigned int flags);
2937 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2938 loff_t *opos, size_t len, unsigned int flags);
2939
2940
2941 extern void
2942 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2943 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2944 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2945 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
2946 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2947 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2948 int whence, loff_t maxsize, loff_t eof);
2949 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2950 int whence, loff_t size);
2951 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
2952 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
2953 extern int generic_file_open(struct inode * inode, struct file * filp);
2954 extern int nonseekable_open(struct inode * inode, struct file * filp);
2955
2956 #ifdef CONFIG_BLOCK
2957 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
2958 loff_t file_offset);
2959
2960 enum {
2961 /* need locking between buffered and direct access */
2962 DIO_LOCKING = 0x01,
2963
2964 /* filesystem does not support filling holes */
2965 DIO_SKIP_HOLES = 0x02,
2966
2967 /* filesystem can handle aio writes beyond i_size */
2968 DIO_ASYNC_EXTEND = 0x04,
2969
2970 /* inode/fs/bdev does not need truncate protection */
2971 DIO_SKIP_DIO_COUNT = 0x08,
2972 };
2973
2974 void dio_end_io(struct bio *bio);
2975
2976 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
2977 struct block_device *bdev, struct iov_iter *iter,
2978 get_block_t get_block,
2979 dio_iodone_t end_io, dio_submit_t submit_io,
2980 int flags);
2981
2982 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
2983 struct inode *inode,
2984 struct iov_iter *iter,
2985 get_block_t get_block)
2986 {
2987 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
2988 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
2989 }
2990 #endif
2991
2992 void inode_dio_wait(struct inode *inode);
2993
2994 /*
2995 * inode_dio_begin - signal start of a direct I/O requests
2996 * @inode: inode the direct I/O happens on
2997 *
2998 * This is called once we've finished processing a direct I/O request,
2999 * and is used to wake up callers waiting for direct I/O to be quiesced.
3000 */
3001 static inline void inode_dio_begin(struct inode *inode)
3002 {
3003 atomic_inc(&inode->i_dio_count);
3004 }
3005
3006 /*
3007 * inode_dio_end - signal finish of a direct I/O requests
3008 * @inode: inode the direct I/O happens on
3009 *
3010 * This is called once we've finished processing a direct I/O request,
3011 * and is used to wake up callers waiting for direct I/O to be quiesced.
3012 */
3013 static inline void inode_dio_end(struct inode *inode)
3014 {
3015 if (atomic_dec_and_test(&inode->i_dio_count))
3016 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
3017 }
3018
3019 extern void inode_set_flags(struct inode *inode, unsigned int flags,
3020 unsigned int mask);
3021
3022 extern const struct file_operations generic_ro_fops;
3023
3024 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3025
3026 extern int readlink_copy(char __user *, int, const char *);
3027 extern int page_readlink(struct dentry *, char __user *, int);
3028 extern const char *page_get_link(struct dentry *, struct inode *,
3029 struct delayed_call *);
3030 extern void page_put_link(void *);
3031 extern int __page_symlink(struct inode *inode, const char *symname, int len,
3032 int nofs);
3033 extern int page_symlink(struct inode *inode, const char *symname, int len);
3034 extern const struct inode_operations page_symlink_inode_operations;
3035 extern void kfree_link(void *);
3036 extern void generic_fillattr(struct inode *, struct kstat *);
3037 extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
3038 extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
3039 void __inode_add_bytes(struct inode *inode, loff_t bytes);
3040 void inode_add_bytes(struct inode *inode, loff_t bytes);
3041 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
3042 void inode_sub_bytes(struct inode *inode, loff_t bytes);
3043 static inline loff_t __inode_get_bytes(struct inode *inode)
3044 {
3045 return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes;
3046 }
3047 loff_t inode_get_bytes(struct inode *inode);
3048 void inode_set_bytes(struct inode *inode, loff_t bytes);
3049 const char *simple_get_link(struct dentry *, struct inode *,
3050 struct delayed_call *);
3051 extern const struct inode_operations simple_symlink_inode_operations;
3052
3053 extern int iterate_dir(struct file *, struct dir_context *);
3054
3055 extern int vfs_statx(int, const char __user *, int, struct kstat *, u32);
3056 extern int vfs_statx_fd(unsigned int, struct kstat *, u32, unsigned int);
3057
3058 static inline int vfs_stat(const char __user *filename, struct kstat *stat)
3059 {
3060 return vfs_statx(AT_FDCWD, filename, AT_NO_AUTOMOUNT,
3061 stat, STATX_BASIC_STATS);
3062 }
3063 static inline int vfs_lstat(const char __user *name, struct kstat *stat)
3064 {
3065 return vfs_statx(AT_FDCWD, name, AT_SYMLINK_NOFOLLOW | AT_NO_AUTOMOUNT,
3066 stat, STATX_BASIC_STATS);
3067 }
3068 static inline int vfs_fstatat(int dfd, const char __user *filename,
3069 struct kstat *stat, int flags)
3070 {
3071 return vfs_statx(dfd, filename, flags, stat, STATX_BASIC_STATS);
3072 }
3073 static inline int vfs_fstat(int fd, struct kstat *stat)
3074 {
3075 return vfs_statx_fd(fd, stat, STATX_BASIC_STATS, 0);
3076 }
3077
3078
3079 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
3080 extern int vfs_readlink(struct dentry *, char __user *, int);
3081
3082 extern int __generic_block_fiemap(struct inode *inode,
3083 struct fiemap_extent_info *fieinfo,
3084 loff_t start, loff_t len,
3085 get_block_t *get_block);
3086 extern int generic_block_fiemap(struct inode *inode,
3087 struct fiemap_extent_info *fieinfo, u64 start,
3088 u64 len, get_block_t *get_block);
3089
3090 extern struct file_system_type *get_filesystem(struct file_system_type *fs);
3091 extern void put_filesystem(struct file_system_type *fs);
3092 extern struct file_system_type *get_fs_type(const char *name);
3093 extern struct super_block *get_super(struct block_device *);
3094 extern struct super_block *get_super_thawed(struct block_device *);
3095 extern struct super_block *get_super_exclusive_thawed(struct block_device *bdev);
3096 extern struct super_block *get_active_super(struct block_device *bdev);
3097 extern void drop_super(struct super_block *sb);
3098 extern void drop_super_exclusive(struct super_block *sb);
3099 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
3100 extern void iterate_supers_type(struct file_system_type *,
3101 void (*)(struct super_block *, void *), void *);
3102
3103 extern int dcache_dir_open(struct inode *, struct file *);
3104 extern int dcache_dir_close(struct inode *, struct file *);
3105 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
3106 extern int dcache_readdir(struct file *, struct dir_context *);
3107 extern int simple_setattr(struct dentry *, struct iattr *);
3108 extern int simple_getattr(const struct path *, struct kstat *, u32, unsigned int);
3109 extern int simple_statfs(struct dentry *, struct kstatfs *);
3110 extern int simple_open(struct inode *inode, struct file *file);
3111 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3112 extern int simple_unlink(struct inode *, struct dentry *);
3113 extern int simple_rmdir(struct inode *, struct dentry *);
3114 extern int simple_rename(struct inode *, struct dentry *,
3115 struct inode *, struct dentry *, unsigned int);
3116 extern int noop_fsync(struct file *, loff_t, loff_t, int);
3117 extern int simple_empty(struct dentry *);
3118 extern int simple_readpage(struct file *file, struct page *page);
3119 extern int simple_write_begin(struct file *file, struct address_space *mapping,
3120 loff_t pos, unsigned len, unsigned flags,
3121 struct page **pagep, void **fsdata);
3122 extern int simple_write_end(struct file *file, struct address_space *mapping,
3123 loff_t pos, unsigned len, unsigned copied,
3124 struct page *page, void *fsdata);
3125 extern int always_delete_dentry(const struct dentry *);
3126 extern struct inode *alloc_anon_inode(struct super_block *);
3127 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
3128 extern const struct dentry_operations simple_dentry_operations;
3129
3130 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3131 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3132 extern const struct file_operations simple_dir_operations;
3133 extern const struct inode_operations simple_dir_inode_operations;
3134 extern void make_empty_dir_inode(struct inode *inode);
3135 extern bool is_empty_dir_inode(struct inode *inode);
3136 struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
3137 struct dentry *d_alloc_name(struct dentry *, const char *);
3138 extern int simple_fill_super(struct super_block *, unsigned long,
3139 const struct tree_descr *);
3140 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3141 extern void simple_release_fs(struct vfsmount **mount, int *count);
3142
3143 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3144 loff_t *ppos, const void *from, size_t available);
3145 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3146 const void __user *from, size_t count);
3147
3148 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3149 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3150
3151 extern int generic_check_addressable(unsigned, u64);
3152
3153 #ifdef CONFIG_MIGRATION
3154 extern int buffer_migrate_page(struct address_space *,
3155 struct page *, struct page *,
3156 enum migrate_mode);
3157 #else
3158 #define buffer_migrate_page NULL
3159 #endif
3160
3161 extern int setattr_prepare(struct dentry *, struct iattr *);
3162 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3163 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
3164
3165 extern int file_update_time(struct file *file);
3166
3167 static inline bool io_is_direct(struct file *filp)
3168 {
3169 return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host);
3170 }
3171
3172 static inline bool vma_is_dax(struct vm_area_struct *vma)
3173 {
3174 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3175 }
3176
3177 static inline int iocb_flags(struct file *file)
3178 {
3179 int res = 0;
3180 if (file->f_flags & O_APPEND)
3181 res |= IOCB_APPEND;
3182 if (io_is_direct(file))
3183 res |= IOCB_DIRECT;
3184 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3185 res |= IOCB_DSYNC;
3186 if (file->f_flags & __O_SYNC)
3187 res |= IOCB_SYNC;
3188 return res;
3189 }
3190
3191 static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
3192 {
3193 if (unlikely(flags & ~RWF_SUPPORTED))
3194 return -EOPNOTSUPP;
3195
3196 if (flags & RWF_NOWAIT) {
3197 if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
3198 return -EOPNOTSUPP;
3199 ki->ki_flags |= IOCB_NOWAIT;
3200 }
3201 if (flags & RWF_HIPRI)
3202 ki->ki_flags |= IOCB_HIPRI;
3203 if (flags & RWF_DSYNC)
3204 ki->ki_flags |= IOCB_DSYNC;
3205 if (flags & RWF_SYNC)
3206 ki->ki_flags |= (IOCB_DSYNC | IOCB_SYNC);
3207 return 0;
3208 }
3209
3210 static inline ino_t parent_ino(struct dentry *dentry)
3211 {
3212 ino_t res;
3213
3214 /*
3215 * Don't strictly need d_lock here? If the parent ino could change
3216 * then surely we'd have a deeper race in the caller?
3217 */
3218 spin_lock(&dentry->d_lock);
3219 res = dentry->d_parent->d_inode->i_ino;
3220 spin_unlock(&dentry->d_lock);
3221 return res;
3222 }
3223
3224 /* Transaction based IO helpers */
3225
3226 /*
3227 * An argresp is stored in an allocated page and holds the
3228 * size of the argument or response, along with its content
3229 */
3230 struct simple_transaction_argresp {
3231 ssize_t size;
3232 char data[0];
3233 };
3234
3235 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3236
3237 char *simple_transaction_get(struct file *file, const char __user *buf,
3238 size_t size);
3239 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3240 size_t size, loff_t *pos);
3241 int simple_transaction_release(struct inode *inode, struct file *file);
3242
3243 void simple_transaction_set(struct file *file, size_t n);
3244
3245 /*
3246 * simple attribute files
3247 *
3248 * These attributes behave similar to those in sysfs:
3249 *
3250 * Writing to an attribute immediately sets a value, an open file can be
3251 * written to multiple times.
3252 *
3253 * Reading from an attribute creates a buffer from the value that might get
3254 * read with multiple read calls. When the attribute has been read
3255 * completely, no further read calls are possible until the file is opened
3256 * again.
3257 *
3258 * All attributes contain a text representation of a numeric value
3259 * that are accessed with the get() and set() functions.
3260 */
3261 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3262 static int __fops ## _open(struct inode *inode, struct file *file) \
3263 { \
3264 __simple_attr_check_format(__fmt, 0ull); \
3265 return simple_attr_open(inode, file, __get, __set, __fmt); \
3266 } \
3267 static const struct file_operations __fops = { \
3268 .owner = THIS_MODULE, \
3269 .open = __fops ## _open, \
3270 .release = simple_attr_release, \
3271 .read = simple_attr_read, \
3272 .write = simple_attr_write, \
3273 .llseek = generic_file_llseek, \
3274 }
3275
3276 static inline __printf(1, 2)
3277 void __simple_attr_check_format(const char *fmt, ...)
3278 {
3279 /* don't do anything, just let the compiler check the arguments; */
3280 }
3281
3282 int simple_attr_open(struct inode *inode, struct file *file,
3283 int (*get)(void *, u64 *), int (*set)(void *, u64),
3284 const char *fmt);
3285 int simple_attr_release(struct inode *inode, struct file *file);
3286 ssize_t simple_attr_read(struct file *file, char __user *buf,
3287 size_t len, loff_t *ppos);
3288 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3289 size_t len, loff_t *ppos);
3290
3291 struct ctl_table;
3292 int proc_nr_files(struct ctl_table *table, int write,
3293 void __user *buffer, size_t *lenp, loff_t *ppos);
3294 int proc_nr_dentry(struct ctl_table *table, int write,
3295 void __user *buffer, size_t *lenp, loff_t *ppos);
3296 int proc_nr_inodes(struct ctl_table *table, int write,
3297 void __user *buffer, size_t *lenp, loff_t *ppos);
3298 int __init get_filesystem_list(char *buf);
3299
3300 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3301 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3302
3303 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3304 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3305 (flag & __FMODE_NONOTIFY)))
3306
3307 static inline bool is_sxid(umode_t mode)
3308 {
3309 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3310 }
3311
3312 static inline int check_sticky(struct inode *dir, struct inode *inode)
3313 {
3314 if (!(dir->i_mode & S_ISVTX))
3315 return 0;
3316
3317 return __check_sticky(dir, inode);
3318 }
3319
3320 static inline void inode_has_no_xattr(struct inode *inode)
3321 {
3322 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
3323 inode->i_flags |= S_NOSEC;
3324 }
3325
3326 static inline bool is_root_inode(struct inode *inode)
3327 {
3328 return inode == inode->i_sb->s_root->d_inode;
3329 }
3330
3331 static inline bool dir_emit(struct dir_context *ctx,
3332 const char *name, int namelen,
3333 u64 ino, unsigned type)
3334 {
3335 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3336 }
3337 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3338 {
3339 return ctx->actor(ctx, ".", 1, ctx->pos,
3340 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3341 }
3342 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3343 {
3344 return ctx->actor(ctx, "..", 2, ctx->pos,
3345 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3346 }
3347 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3348 {
3349 if (ctx->pos == 0) {
3350 if (!dir_emit_dot(file, ctx))
3351 return false;
3352 ctx->pos = 1;
3353 }
3354 if (ctx->pos == 1) {
3355 if (!dir_emit_dotdot(file, ctx))
3356 return false;
3357 ctx->pos = 2;
3358 }
3359 return true;
3360 }
3361 static inline bool dir_relax(struct inode *inode)
3362 {
3363 inode_unlock(inode);
3364 inode_lock(inode);
3365 return !IS_DEADDIR(inode);
3366 }
3367
3368 static inline bool dir_relax_shared(struct inode *inode)
3369 {
3370 inode_unlock_shared(inode);
3371 inode_lock_shared(inode);
3372 return !IS_DEADDIR(inode);
3373 }
3374
3375 extern bool path_noexec(const struct path *path);
3376 extern void inode_nohighmem(struct inode *inode);
3377
3378 #endif /* _LINUX_FS_H */