1	/* SPDX-License-Identifier: GPL-2.0 */
2	#ifndef _LINUX_FS_H
3	#define _LINUX_FS_H
4
5	#include <linux/vfsdebug.h>
6	#include <linux/linkage.h>
7	#include <linux/wait_bit.h>
8	#include <linux/kdev_t.h>
9	#include <linux/dcache.h>
10	#include <linux/path.h>
11	#include <linux/stat.h>
12	#include <linux/cache.h>
13	#include <linux/list.h>
14	#include <linux/list_lru.h>
15	#include <linux/llist.h>
16	#include <linux/radix-tree.h>
17	#include <linux/xarray.h>
18	#include <linux/rbtree.h>
19	#include <linux/init.h>
20	#include <linux/pid.h>
21	#include <linux/bug.h>
22	#include <linux/mutex.h>
23	#include <linux/rwsem.h>
24	#include <linux/mm_types.h>
25	#include <linux/capability.h>
26	#include <linux/semaphore.h>
27	#include <linux/fcntl.h>
28	#include <linux/rculist_bl.h>
29	#include <linux/atomic.h>
30	#include <linux/shrinker.h>
31	#include <linux/migrate_mode.h>
32	#include <linux/uidgid.h>
33	#include <linux/lockdep.h>
34	#include <linux/percpu-rwsem.h>
35	#include <linux/workqueue.h>
36	#include <linux/delayed_call.h>
37	#include <linux/uuid.h>
38	#include <linux/errseq.h>
39	#include <linux/ioprio.h>
40	#include <linux/fs_types.h>
41	#include <linux/build_bug.h>
42	#include <linux/stddef.h>
43	#include <linux/mount.h>
44	#include <linux/cred.h>
45	#include <linux/mnt_idmapping.h>
46	#include <linux/slab.h>
47	#include <linux/maple_tree.h>
48	#include <linux/rw_hint.h>
49	#include <linux/file_ref.h>
50	#include <linux/unicode.h>
51
52	#include <asm/byteorder.h>
53	#include <uapi/linux/fs.h>
54
55	struct backing_dev_info;
56	struct bdi_writeback;
57	struct bio;
58	struct io_comp_batch;
59	struct export_operations;
60	struct fiemap_extent_info;
61	struct hd_geometry;
62	struct iovec;
63	struct kiocb;
64	struct kobject;
65	struct pipe_inode_info;
66	struct poll_table_struct;
67	struct kstatfs;
68	struct vm_area_struct;
69	struct vfsmount;
70	struct cred;
71	struct swap_info_struct;
72	struct seq_file;
73	struct workqueue_struct;
74	struct iov_iter;
75	struct fscrypt_inode_info;
76	struct fscrypt_operations;
77	struct fsverity_info;
78	struct fsverity_operations;
79	struct fsnotify_mark_connector;
80	struct fsnotify_sb_info;
81	struct fs_context;
82	struct fs_parameter_spec;
83	struct fileattr;
84	struct iomap_ops;
85
86	extern void __init inode_init(void);
87	extern void __init inode_init_early(void);
88	extern void __init files_init(void);
89	extern void __init files_maxfiles_init(void);
90
91	extern unsigned long get_max_files(void);
92	extern unsigned int sysctl_nr_open;
93
94	typedef __kernel_rwf_t rwf_t;
95
96	struct buffer_head;
97	typedef int (get_block_t)(struct inode *inode, sector_t iblock,
98	struct buffer_head *bh_result, int create);
99	typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
100	ssize_t bytes, void *private);
101
102	#define MAY_EXEC 0x00000001
103	#define MAY_WRITE 0x00000002
104	#define MAY_READ 0x00000004
105	#define MAY_APPEND 0x00000008
106	#define MAY_ACCESS 0x00000010
107	#define MAY_OPEN 0x00000020
108	#define MAY_CHDIR 0x00000040
109	/* called from RCU mode, don't block */
110	#define MAY_NOT_BLOCK 0x00000080
111
112	/*
113	* flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
114	* to O_WRONLY and O_RDWR via the strange trick in do_dentry_open()
115	*/
116
117	/* file is open for reading */
118	#define FMODE_READ ((__force fmode_t)(1 << 0))
119	/* file is open for writing */
120	#define FMODE_WRITE ((__force fmode_t)(1 << 1))
121	/* file is seekable */
122	#define FMODE_LSEEK ((__force fmode_t)(1 << 2))
123	/* file can be accessed using pread */
124	#define FMODE_PREAD ((__force fmode_t)(1 << 3))
125	/* file can be accessed using pwrite */
126	#define FMODE_PWRITE ((__force fmode_t)(1 << 4))
127	/* File is opened for execution with sys_execve / sys_uselib */
128	#define FMODE_EXEC ((__force fmode_t)(1 << 5))
129	/* File writes are restricted (block device specific) */
130	#define FMODE_WRITE_RESTRICTED ((__force fmode_t)(1 << 6))
131	/* File supports atomic writes */
132	#define FMODE_CAN_ATOMIC_WRITE ((__force fmode_t)(1 << 7))
133
134	/* FMODE_* bit 8 */
135
136	/* 32bit hashes as llseek() offset (for directories) */
137	#define FMODE_32BITHASH ((__force fmode_t)(1 << 9))
138	/* 64bit hashes as llseek() offset (for directories) */
139	#define FMODE_64BITHASH ((__force fmode_t)(1 << 10))
140
141	/*
142	* Don't update ctime and mtime.
143	*
144	* Currently a special hack for the XFS open_by_handle ioctl, but we'll
145	* hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
146	*/
147	#define FMODE_NOCMTIME ((__force fmode_t)(1 << 11))
148
149	/* Expect random access pattern */
150	#define FMODE_RANDOM ((__force fmode_t)(1 << 12))
151
152	/* FMODE_* bit 13 */
153
154	/* File is opened with O_PATH; almost nothing can be done with it */
155	#define FMODE_PATH ((__force fmode_t)(1 << 14))
156
157	/* File needs atomic accesses to f_pos */
158	#define FMODE_ATOMIC_POS ((__force fmode_t)(1 << 15))
159	/* Write access to underlying fs */
160	#define FMODE_WRITER ((__force fmode_t)(1 << 16))
161	/* Has read method(s) */
162	#define FMODE_CAN_READ ((__force fmode_t)(1 << 17))
163	/* Has write method(s) */
164	#define FMODE_CAN_WRITE ((__force fmode_t)(1 << 18))
165
166	#define FMODE_OPENED ((__force fmode_t)(1 << 19))
167	#define FMODE_CREATED ((__force fmode_t)(1 << 20))
168
169	/* File is stream-like */
170	#define FMODE_STREAM ((__force fmode_t)(1 << 21))
171
172	/* File supports DIRECT IO */
173	#define FMODE_CAN_ODIRECT ((__force fmode_t)(1 << 22))
174
175	#define FMODE_NOREUSE ((__force fmode_t)(1 << 23))
176
177	/* File is embedded in backing_file object */
178	#define FMODE_BACKING ((__force fmode_t)(1 << 24))
179
180	/*
181	* Together with FMODE_NONOTIFY_PERM defines which fsnotify events shouldn't be
182	* generated (see below)
183	*/
184	#define FMODE_NONOTIFY ((__force fmode_t)(1 << 25))
185
186	/*
187	* Together with FMODE_NONOTIFY defines which fsnotify events shouldn't be
188	* generated (see below)
189	*/
190	#define FMODE_NONOTIFY_PERM ((__force fmode_t)(1 << 26))
191
192	/* File is capable of returning -EAGAIN if I/O will block */
193	#define FMODE_NOWAIT ((__force fmode_t)(1 << 27))
194
195	/* File represents mount that needs unmounting */
196	#define FMODE_NEED_UNMOUNT ((__force fmode_t)(1 << 28))
197
198	/* File does not contribute to nr_files count */
199	#define FMODE_NOACCOUNT ((__force fmode_t)(1 << 29))
200
201	/*
202	* The two FMODE_NONOTIFY* define which fsnotify events should not be generated
203	* for a file. These are the possible values of (f->f_mode &
204	* FMODE_FSNOTIFY_MASK) and their meaning:
205	*
206	* FMODE_NONOTIFY - suppress all (incl. non-permission) events.
207	* FMODE_NONOTIFY_PERM - suppress permission (incl. pre-content) events.
208	* FMODE_NONOTIFY | FMODE_NONOTIFY_PERM - suppress only pre-content events.
209	*/
210	#define FMODE_FSNOTIFY_MASK \
211	(FMODE_NONOTIFY | FMODE_NONOTIFY_PERM)
212
213	#define FMODE_FSNOTIFY_NONE(mode) \
214	((mode & FMODE_FSNOTIFY_MASK) == FMODE_NONOTIFY)
215	#ifdef CONFIG_FANOTIFY_ACCESS_PERMISSIONS
216	#define FMODE_FSNOTIFY_PERM(mode) \
217	((mode & FMODE_FSNOTIFY_MASK) == 0 || \
218	(mode & FMODE_FSNOTIFY_MASK) == (FMODE_NONOTIFY | FMODE_NONOTIFY_PERM))
219	#define FMODE_FSNOTIFY_HSM(mode) \
220	((mode & FMODE_FSNOTIFY_MASK) == 0)
221	#else
222	#define FMODE_FSNOTIFY_PERM(mode) 0
223	#define FMODE_FSNOTIFY_HSM(mode) 0
224	#endif
225
226	/*
227	* Attribute flags. These should be or-ed together to figure out what
228	* has been changed!
229	*/
230	#define ATTR_MODE (1 << 0)
231	#define ATTR_UID (1 << 1)
232	#define ATTR_GID (1 << 2)
233	#define ATTR_SIZE (1 << 3)
234	#define ATTR_ATIME (1 << 4)
235	#define ATTR_MTIME (1 << 5)
236	#define ATTR_CTIME (1 << 6)
237	#define ATTR_ATIME_SET (1 << 7)
238	#define ATTR_MTIME_SET (1 << 8)
239	#define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
240	#define ATTR_KILL_SUID (1 << 11)
241	#define ATTR_KILL_SGID (1 << 12)
242	#define ATTR_FILE (1 << 13)
243	#define ATTR_KILL_PRIV (1 << 14)
244	#define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
245	#define ATTR_TIMES_SET (1 << 16)
246	#define ATTR_TOUCH (1 << 17)
247	#define ATTR_DELEG (1 << 18) /* Delegated attrs. Don't break write delegations */
248
249	/*
250	* Whiteout is represented by a char device. The following constants define the
251	* mode and device number to use.
252	*/
253	#define WHITEOUT_MODE 0
254	#define WHITEOUT_DEV 0
255
256	/*
257	* This is the Inode Attributes structure, used for notify_change(). It
258	* uses the above definitions as flags, to know which values have changed.
259	* Also, in this manner, a Filesystem can look at only the values it cares
260	* about. Basically, these are the attributes that the VFS layer can
261	* request to change from the FS layer.
262	*
263	* Derek Atkins <warlord@MIT.EDU> 94-10-20
264	*/
265	struct iattr {
266	unsigned int ia_valid;
267	umode_t ia_mode;
268	/*
269	* The two anonymous unions wrap structures with the same member.
270	*
271	* Filesystems raising FS_ALLOW_IDMAP need to use ia_vfs{g,u}id which
272	* are a dedicated type requiring the filesystem to use the dedicated
273	* helpers. Other filesystem can continue to use ia_{g,u}id until they
274	* have been ported.
275	*
276	* They always contain the same value. In other words FS_ALLOW_IDMAP
277	* pass down the same value on idmapped mounts as they would on regular
278	* mounts.
279	*/
280	union {
281	kuid_t ia_uid;
282	vfsuid_t ia_vfsuid;
283	};
284	union {
285	kgid_t ia_gid;
286	vfsgid_t ia_vfsgid;
287	};
288	loff_t ia_size;
289	struct timespec64 ia_atime;
290	struct timespec64 ia_mtime;
291	struct timespec64 ia_ctime;
292
293	/*
294	* Not an attribute, but an auxiliary info for filesystems wanting to
295	* implement an ftruncate() like method. NOTE: filesystem should
296	* check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
297	*/
298	struct file *ia_file;
299	};
300
301	/*
302	* Includes for diskquotas.
303	*/
304	#include <linux/quota.h>
305
306	/*
307	* Maximum number of layers of fs stack. Needs to be limited to
308	* prevent kernel stack overflow
309	*/
310	#define FILESYSTEM_MAX_STACK_DEPTH 2
311
312	/**
313	* enum positive_aop_returns - aop return codes with specific semantics
314	*
315	* @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
316	* completed, that the page is still locked, and
317	* should be considered active. The VM uses this hint
318	* to return the page to the active list -- it won't
319	* be a candidate for writeback again in the near
320	* future. Other callers must be careful to unlock
321	* the page if they get this return. Returned by
322	* writepage();
323	*
324	* @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
325	* unlocked it and the page might have been truncated.
326	* The caller should back up to acquiring a new page and
327	* trying again. The aop will be taking reasonable
328	* precautions not to livelock. If the caller held a page
329	* reference, it should drop it before retrying. Returned
330	* by read_folio().
331	*
332	* address_space_operation functions return these large constants to indicate
333	* special semantics to the caller. These are much larger than the bytes in a
334	* page to allow for functions that return the number of bytes operated on in a
335	* given page.
336	*/
337
338	enum positive_aop_returns {
339	AOP_WRITEPAGE_ACTIVATE = 0x80000,
340	AOP_TRUNCATED_PAGE = 0x80001,
341	};
342
343	/*
344	* oh the beauties of C type declarations.
345	*/
346	struct page;
347	struct address_space;
348	struct writeback_control;
349	struct readahead_control;
350
351	/* Match RWF_* bits to IOCB bits */
352	#define IOCB_HIPRI (__force int) RWF_HIPRI
353	#define IOCB_DSYNC (__force int) RWF_DSYNC
354	#define IOCB_SYNC (__force int) RWF_SYNC
355	#define IOCB_NOWAIT (__force int) RWF_NOWAIT
356	#define IOCB_APPEND (__force int) RWF_APPEND
357	#define IOCB_ATOMIC (__force int) RWF_ATOMIC
358	#define IOCB_DONTCACHE (__force int) RWF_DONTCACHE
359
360	/* non-RWF related bits - start at 16 */
361	#define IOCB_EVENTFD (1 << 16)
362	#define IOCB_DIRECT (1 << 17)
363	#define IOCB_WRITE (1 << 18)
364	/* iocb->ki_waitq is valid */
365	#define IOCB_WAITQ (1 << 19)
366	#define IOCB_NOIO (1 << 20)
367	/* can use bio alloc cache */
368	#define IOCB_ALLOC_CACHE (1 << 21)
369	/*
370	* IOCB_DIO_CALLER_COMP can be set by the iocb owner, to indicate that the
371	* iocb completion can be passed back to the owner for execution from a safe
372	* context rather than needing to be punted through a workqueue. If this
373	* flag is set, the bio completion handling may set iocb->dio_complete to a
374	* handler function and iocb->private to context information for that handler.
375	* The issuer should call the handler with that context information from task
376	* context to complete the processing of the iocb. Note that while this
377	* provides a task context for the dio_complete() callback, it should only be
378	* used on the completion side for non-IO generating completions. It's fine to
379	* call blocking functions from this callback, but they should not wait for
380	* unrelated IO (like cache flushing, new IO generation, etc).
381	*/
382	#define IOCB_DIO_CALLER_COMP (1 << 22)
383	/* kiocb is a read or write operation submitted by fs/aio.c. */
384	#define IOCB_AIO_RW (1 << 23)
385	#define IOCB_HAS_METADATA (1 << 24)
386
387	/* for use in trace events */
388	#define TRACE_IOCB_STRINGS \
389	{ IOCB_HIPRI, "HIPRI" }, \
390	{ IOCB_DSYNC, "DSYNC" }, \
391	{ IOCB_SYNC, "SYNC" }, \
392	{ IOCB_NOWAIT, "NOWAIT" }, \
393	{ IOCB_APPEND, "APPEND" }, \
394	{ IOCB_ATOMIC, "ATOMIC" }, \
395	{ IOCB_DONTCACHE, "DONTCACHE" }, \
396	{ IOCB_EVENTFD, "EVENTFD"}, \
397	{ IOCB_DIRECT, "DIRECT" }, \
398	{ IOCB_WRITE, "WRITE" }, \
399	{ IOCB_WAITQ, "WAITQ" }, \
400	{ IOCB_NOIO, "NOIO" }, \
401	{ IOCB_ALLOC_CACHE, "ALLOC_CACHE" }, \
402	{ IOCB_DIO_CALLER_COMP, "CALLER_COMP" }
403
404	struct kiocb {
405	struct file *ki_filp;
406	loff_t ki_pos;
407	void (*ki_complete)(struct kiocb *iocb, long ret);
408	void *private;
409	int ki_flags;
410	u16 ki_ioprio; /* See linux/ioprio.h */
411	u8 ki_write_stream;
412	union {
413	/*
414	* Only used for async buffered reads, where it denotes the
415	* page waitqueue associated with completing the read. Valid
416	* IFF IOCB_WAITQ is set.
417	*/
418	struct wait_page_queue *ki_waitq;
419	/*
420	* Can be used for O_DIRECT IO, where the completion handling
421	* is punted back to the issuer of the IO. May only be set
422	* if IOCB_DIO_CALLER_COMP is set by the issuer, and the issuer
423	* must then check for presence of this handler when ki_complete
424	* is invoked. The data passed in to this handler must be
425	* assigned to ->private when dio_complete is assigned.
426	*/
427	ssize_t (*dio_complete)(void *data);
428	};
429	};
430
431	static inline bool is_sync_kiocb(struct kiocb *kiocb)
432	{
433	return kiocb->ki_complete == NULL;
434	}
435
436	struct address_space_operations {
437	int (*read_folio)(struct file *, struct folio *);
438
439	/* Write back some dirty pages from this mapping. */
440	int (*writepages)(struct address_space *, struct writeback_control *);
441
442	/* Mark a folio dirty. Return true if this dirtied it */
443	bool (*dirty_folio)(struct address_space *, struct folio *);
444
445	void (*readahead)(struct readahead_control *);
446
447	int (*write_begin)(struct file *, struct address_space *mapping,
448	loff_t pos, unsigned len,
449	struct folio **foliop, void **fsdata);
450	int (*write_end)(struct file *, struct address_space *mapping,
451	loff_t pos, unsigned len, unsigned copied,
452	struct folio *folio, void *fsdata);
453
454	/* Unfortunately this kludge is needed for FIBMAP. Don't use it */
455	sector_t (*bmap)(struct address_space *, sector_t);
456	void (*invalidate_folio) (struct folio *, size_t offset, size_t len);
457	bool (*release_folio)(struct folio *, gfp_t);
458	void (*free_folio)(struct folio *folio);
459	ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
460	/*
461	* migrate the contents of a folio to the specified target. If
462	* migrate_mode is MIGRATE_ASYNC, it must not block.
463	*/
464	int (*migrate_folio)(struct address_space *, struct folio *dst,
465	struct folio *src, enum migrate_mode);
466	int (*launder_folio)(struct folio *);
467	bool (*is_partially_uptodate) (struct folio *, size_t from,
468	size_t count);
469	void (*is_dirty_writeback) (struct folio *, bool *dirty, bool *wb);
470	int (*error_remove_folio)(struct address_space *, struct folio *);
471
472	/* swapfile support */
473	int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
474	sector_t *span);
475	void (*swap_deactivate)(struct file *file);
476	int (*swap_rw)(struct kiocb *iocb, struct iov_iter *iter);
477	};
478
479	extern const struct address_space_operations empty_aops;
480
481	/**
482	* struct address_space - Contents of a cacheable, mappable object.
483	* @host: Owner, either the inode or the block_device.
484	* @i_pages: Cached pages.
485	* @invalidate_lock: Guards coherency between page cache contents and
486	* file offset->disk block mappings in the filesystem during invalidates.
487	* It is also used to block modification of page cache contents through
488	* memory mappings.
489	* @gfp_mask: Memory allocation flags to use for allocating pages.
490	* @i_mmap_writable: Number of VM_SHARED, VM_MAYWRITE mappings.
491	* @nr_thps: Number of THPs in the pagecache (non-shmem only).
492	* @i_mmap: Tree of private and shared mappings.
493	* @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable.
494	* @nrpages: Number of page entries, protected by the i_pages lock.
495	* @writeback_index: Writeback starts here.
496	* @a_ops: Methods.
497	* @flags: Error bits and flags (AS_*).
498	* @wb_err: The most recent error which has occurred.
499	* @i_private_lock: For use by the owner of the address_space.
500	* @i_private_list: For use by the owner of the address_space.
501	* @i_private_data: For use by the owner of the address_space.
502	*/
503	struct address_space {
504	struct inode *host;
505	struct xarray i_pages;
506	struct rw_semaphore invalidate_lock;
507	gfp_t gfp_mask;
508	atomic_t i_mmap_writable;
509	#ifdef CONFIG_READ_ONLY_THP_FOR_FS
510	/* number of thp, only for non-shmem files */
511	atomic_t nr_thps;
512	#endif
513	struct rb_root_cached i_mmap;
514	unsigned long nrpages;
515	pgoff_t writeback_index;
516	const struct address_space_operations *a_ops;
517	unsigned long flags;
518	errseq_t wb_err;
519	spinlock_t i_private_lock;
520	struct list_head i_private_list;
521	struct rw_semaphore i_mmap_rwsem;
522	void * i_private_data;
523	} __attribute__((aligned(sizeof(long)))) __randomize_layout;
524	/*
525	* On most architectures that alignment is already the case; but
526	* must be enforced here for CRIS, to let the least significant bit
527	* of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
528	*/
529
530	/* XArray tags, for tagging dirty and writeback pages in the pagecache. */
531	#define PAGECACHE_TAG_DIRTY XA_MARK_0
532	#define PAGECACHE_TAG_WRITEBACK XA_MARK_1
533	#define PAGECACHE_TAG_TOWRITE XA_MARK_2
534
535	/*
536	* Returns true if any of the pages in the mapping are marked with the tag.
537	*/
538	static inline bool mapping_tagged(struct address_space *mapping, xa_mark_t tag)
539	{
540	return xa_marked(xa: &mapping->i_pages, mark: tag);
541	}
542
543	static inline void i_mmap_lock_write(struct address_space *mapping)
544	{
545	down_write(sem: &mapping->i_mmap_rwsem);
546	}
547
548	static inline int i_mmap_trylock_write(struct address_space *mapping)
549	{
550	return down_write_trylock(sem: &mapping->i_mmap_rwsem);
551	}
552
553	static inline void i_mmap_unlock_write(struct address_space *mapping)
554	{
555	up_write(sem: &mapping->i_mmap_rwsem);
556	}
557
558	static inline int i_mmap_trylock_read(struct address_space *mapping)
559	{
560	return down_read_trylock(sem: &mapping->i_mmap_rwsem);
561	}
562
563	static inline void i_mmap_lock_read(struct address_space *mapping)
564	{
565	down_read(sem: &mapping->i_mmap_rwsem);
566	}
567
568	static inline void i_mmap_unlock_read(struct address_space *mapping)
569	{
570	up_read(sem: &mapping->i_mmap_rwsem);
571	}
572
573	static inline void i_mmap_assert_locked(struct address_space *mapping)
574	{
575	lockdep_assert_held(&mapping->i_mmap_rwsem);
576	}
577
578	static inline void i_mmap_assert_write_locked(struct address_space *mapping)
579	{
580	lockdep_assert_held_write(&mapping->i_mmap_rwsem);
581	}
582
583	/*
584	* Might pages of this file be mapped into userspace?
585	*/
586	static inline int mapping_mapped(struct address_space *mapping)
587	{
588	return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root);
589	}
590
591	/*
592	* Might pages of this file have been modified in userspace?
593	* Note that i_mmap_writable counts all VM_SHARED, VM_MAYWRITE vmas: do_mmap
594	* marks vma as VM_SHARED if it is shared, and the file was opened for
595	* writing i.e. vma may be mprotected writable even if now readonly.
596	*
597	* If i_mmap_writable is negative, no new writable mappings are allowed. You
598	* can only deny writable mappings, if none exists right now.
599	*/
600	static inline int mapping_writably_mapped(struct address_space *mapping)
601	{
602	return atomic_read(v: &mapping->i_mmap_writable) > 0;
603	}
604
605	static inline int mapping_map_writable(struct address_space *mapping)
606	{
607	return atomic_inc_unless_negative(v: &mapping->i_mmap_writable) ?
608	0 : -EPERM;
609	}
610
611	static inline void mapping_unmap_writable(struct address_space *mapping)
612	{
613	atomic_dec(v: &mapping->i_mmap_writable);
614	}
615
616	static inline int mapping_deny_writable(struct address_space *mapping)
617	{
618	return atomic_dec_unless_positive(v: &mapping->i_mmap_writable) ?
619	0 : -EBUSY;
620	}
621
622	static inline void mapping_allow_writable(struct address_space *mapping)
623	{
624	atomic_inc(v: &mapping->i_mmap_writable);
625	}
626
627	/*
628	* Use sequence counter to get consistent i_size on 32-bit processors.
629	*/
630	#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
631	#include <linux/seqlock.h>
632	#define __NEED_I_SIZE_ORDERED
633	#define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
634	#else
635	#define i_size_ordered_init(inode) do { } while (0)
636	#endif
637
638	struct posix_acl;
639	#define ACL_NOT_CACHED ((void *)(-1))
640	/*
641	* ACL_DONT_CACHE is for stacked filesystems, that rely on underlying fs to
642	* cache the ACL. This also means that ->get_inode_acl() can be called in RCU
643	* mode with the LOOKUP_RCU flag.
644	*/
645	#define ACL_DONT_CACHE ((void *)(-3))
646
647	static inline struct posix_acl *
648	uncached_acl_sentinel(struct task_struct *task)
649	{
650	return (void *)task + 1;
651	}
652
653	static inline bool
654	is_uncached_acl(struct posix_acl *acl)
655	{
656	return (long)acl & 1;
657	}
658
659	#define IOP_FASTPERM 0x0001
660	#define IOP_LOOKUP 0x0002
661	#define IOP_NOFOLLOW 0x0004
662	#define IOP_XATTR 0x0008
663	#define IOP_DEFAULT_READLINK 0x0010
664	#define IOP_MGTIME 0x0020
665	#define IOP_CACHED_LINK 0x0040
666
667	/*
668	* Keep mostly read-only and often accessed (especially for
669	* the RCU path lookup and 'stat' data) fields at the beginning
670	* of the 'struct inode'
671	*/
672	struct inode {
673	umode_t i_mode;
674	unsigned short i_opflags;
675	kuid_t i_uid;
676	kgid_t i_gid;
677	unsigned int i_flags;
678
679	#ifdef CONFIG_FS_POSIX_ACL
680	struct posix_acl *i_acl;
681	struct posix_acl *i_default_acl;
682	#endif
683
684	const struct inode_operations *i_op;
685	struct super_block *i_sb;
686	struct address_space *i_mapping;
687
688	#ifdef CONFIG_SECURITY
689	void *i_security;
690	#endif
691
692	/* Stat data, not accessed from path walking */
693	unsigned long i_ino;
694	/*
695	* Filesystems may only read i_nlink directly. They shall use the
696	* following functions for modification:
697	*
698	* (set|clear|inc|drop)_nlink
699	* inode_(inc|dec)_link_count
700	*/
701	union {
702	const unsigned int i_nlink;
703	unsigned int __i_nlink;
704	};
705	dev_t i_rdev;
706	loff_t i_size;
707	time64_t i_atime_sec;
708	time64_t i_mtime_sec;
709	time64_t i_ctime_sec;
710	u32 i_atime_nsec;
711	u32 i_mtime_nsec;
712	u32 i_ctime_nsec;
713	u32 i_generation;
714	spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
715	unsigned short i_bytes;
716	u8 i_blkbits;
717	enum rw_hint i_write_hint;
718	blkcnt_t i_blocks;
719
720	#ifdef __NEED_I_SIZE_ORDERED
721	seqcount_t i_size_seqcount;
722	#endif
723
724	/* Misc */
725	u32 i_state;
726	/* 32-bit hole */
727	struct rw_semaphore i_rwsem;
728
729	unsigned long dirtied_when; /* jiffies of first dirtying */
730	unsigned long dirtied_time_when;
731
732	struct hlist_node i_hash;
733	struct list_head i_io_list; /* backing dev IO list */
734	#ifdef CONFIG_CGROUP_WRITEBACK
735	struct bdi_writeback *i_wb; /* the associated cgroup wb */
736
737	/* foreign inode detection, see wbc_detach_inode() */
738	int i_wb_frn_winner;
739	u16 i_wb_frn_avg_time;
740	u16 i_wb_frn_history;
741	#endif
742	struct list_head i_lru; /* inode LRU list */
743	struct list_head i_sb_list;
744	struct list_head i_wb_list; /* backing dev writeback list */
745	union {
746	struct hlist_head i_dentry;
747	struct rcu_head i_rcu;
748	};
749	atomic64_t i_version;
750	atomic64_t i_sequence; /* see futex */
751	atomic_t i_count;
752	atomic_t i_dio_count;
753	atomic_t i_writecount;
754	#if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
755	atomic_t i_readcount; /* struct files open RO */
756	#endif
757	union {
758	const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
759	void (*free_inode)(struct inode *);
760	};
761	struct file_lock_context *i_flctx;
762	struct address_space i_data;
763	union {
764	struct list_head i_devices;
765	int i_linklen;
766	};
767	union {
768	struct pipe_inode_info *i_pipe;
769	struct cdev *i_cdev;
770	char *i_link;
771	unsigned i_dir_seq;
772	};
773
774
775	#ifdef CONFIG_FSNOTIFY
776	__u32 i_fsnotify_mask; /* all events this inode cares about */
777	/* 32-bit hole reserved for expanding i_fsnotify_mask */
778	struct fsnotify_mark_connector __rcu *i_fsnotify_marks;
779	#endif
780
781	#ifdef CONFIG_FS_ENCRYPTION
782	struct fscrypt_inode_info *i_crypt_info;
783	#endif
784
785	#ifdef CONFIG_FS_VERITY
786	struct fsverity_info *i_verity_info;
787	#endif
788
789	void *i_private; /* fs or device private pointer */
790	} __randomize_layout;
791
792	static inline void inode_set_cached_link(struct inode *inode, char *link, int linklen)
793	{
794	VFS_WARN_ON_INODE(strlen(link) != linklen, inode);
795	VFS_WARN_ON_INODE(inode->i_opflags & IOP_CACHED_LINK, inode);
796	inode->i_link = link;
797	inode->i_linklen = linklen;
798	inode->i_opflags |= IOP_CACHED_LINK;
799	}
800
801	/*
802	* Get bit address from inode->i_state to use with wait_var_event()
803	* infrastructre.
804	*/
805	#define inode_state_wait_address(inode, bit) ((char *)&(inode)->i_state + (bit))
806
807	struct wait_queue_head *inode_bit_waitqueue(struct wait_bit_queue_entry *wqe,
808	struct inode *inode, u32 bit);
809
810	static inline void inode_wake_up_bit(struct inode *inode, u32 bit)
811	{
812	/* Caller is responsible for correct memory barriers. */
813	wake_up_var(inode_state_wait_address(inode, bit));
814	}
815
816	struct timespec64 timestamp_truncate(struct timespec64 t, struct inode *inode);
817
818	static inline unsigned int i_blocksize(const struct inode *node)
819	{
820	return (1 << node->i_blkbits);
821	}
822
823	static inline int inode_unhashed(struct inode *inode)
824	{
825	return hlist_unhashed(h: &inode->i_hash);
826	}
827
828	/*
829	* __mark_inode_dirty expects inodes to be hashed. Since we don't
830	* want special inodes in the fileset inode space, we make them
831	* appear hashed, but do not put on any lists. hlist_del()
832	* will work fine and require no locking.
833	*/
834	static inline void inode_fake_hash(struct inode *inode)
835	{
836	hlist_add_fake(n: &inode->i_hash);
837	}
838
839	/*
840	* inode->i_mutex nesting subclasses for the lock validator:
841	*
842	* 0: the object of the current VFS operation
843	* 1: parent
844	* 2: child/target
845	* 3: xattr
846	* 4: second non-directory
847	* 5: second parent (when locking independent directories in rename)
848	*
849	* I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
850	* non-directories at once.
851	*
852	* The locking order between these classes is
853	* parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
854	*/
855	enum inode_i_mutex_lock_class
856	{
857	I_MUTEX_NORMAL,
858	I_MUTEX_PARENT,
859	I_MUTEX_CHILD,
860	I_MUTEX_XATTR,
861	I_MUTEX_NONDIR2,
862	I_MUTEX_PARENT2,
863	};
864
865	static inline void inode_lock(struct inode *inode)
866	{
867	down_write(sem: &inode->i_rwsem);
868	}
869
870	static inline __must_check int inode_lock_killable(struct inode *inode)
871	{
872	return down_write_killable(sem: &inode->i_rwsem);
873	}
874
875	static inline void inode_unlock(struct inode *inode)
876	{
877	up_write(sem: &inode->i_rwsem);
878	}
879
880	static inline void inode_lock_shared(struct inode *inode)
881	{
882	down_read(sem: &inode->i_rwsem);
883	}
884
885	static inline __must_check int inode_lock_shared_killable(struct inode *inode)
886	{
887	return down_read_killable(sem: &inode->i_rwsem);
888	}
889
890	static inline void inode_unlock_shared(struct inode *inode)
891	{
892	up_read(sem: &inode->i_rwsem);
893	}
894
895	static inline int inode_trylock(struct inode *inode)
896	{
897	return down_write_trylock(sem: &inode->i_rwsem);
898	}
899
900	static inline int inode_trylock_shared(struct inode *inode)
901	{
902	return down_read_trylock(sem: &inode->i_rwsem);
903	}
904
905	static inline int inode_is_locked(struct inode *inode)
906	{
907	return rwsem_is_locked(sem: &inode->i_rwsem);
908	}
909
910	static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
911	{
912	down_write_nested(sem: &inode->i_rwsem, subclass);
913	}
914
915	static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass)
916	{
917	down_read_nested(sem: &inode->i_rwsem, subclass);
918	}
919
920	static inline void filemap_invalidate_lock(struct address_space *mapping)
921	{
922	down_write(sem: &mapping->invalidate_lock);
923	}
924
925	static inline void filemap_invalidate_unlock(struct address_space *mapping)
926	{
927	up_write(sem: &mapping->invalidate_lock);
928	}
929
930	static inline void filemap_invalidate_lock_shared(struct address_space *mapping)
931	{
932	down_read(sem: &mapping->invalidate_lock);
933	}
934
935	static inline int filemap_invalidate_trylock_shared(
936	struct address_space *mapping)
937	{
938	return down_read_trylock(sem: &mapping->invalidate_lock);
939	}
940
941	static inline void filemap_invalidate_unlock_shared(
942	struct address_space *mapping)
943	{
944	up_read(sem: &mapping->invalidate_lock);
945	}
946
947	void lock_two_nondirectories(struct inode *, struct inode*);
948	void unlock_two_nondirectories(struct inode *, struct inode*);
949
950	void filemap_invalidate_lock_two(struct address_space *mapping1,
951	struct address_space *mapping2);
952	void filemap_invalidate_unlock_two(struct address_space *mapping1,
953	struct address_space *mapping2);
954
955
956	/*
957	* NOTE: in a 32bit arch with a preemptable kernel and
958	* an UP compile the i_size_read/write must be atomic
959	* with respect to the local cpu (unlike with preempt disabled),
960	* but they don't need to be atomic with respect to other cpus like in
961	* true SMP (so they need either to either locally disable irq around
962	* the read or for example on x86 they can be still implemented as a
963	* cmpxchg8b without the need of the lock prefix). For SMP compiles
964	* and 64bit archs it makes no difference if preempt is enabled or not.
965	*/
966	static inline loff_t i_size_read(const struct inode *inode)
967	{
968	#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
969	loff_t i_size;
970	unsigned int seq;
971
972	do {
973	seq = read_seqcount_begin(&inode->i_size_seqcount);
974	i_size = inode->i_size;
975	} while (read_seqcount_retry(&inode->i_size_seqcount, seq));
976	return i_size;
977	#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
978	loff_t i_size;
979
980	preempt_disable();
981	i_size = inode->i_size;
982	preempt_enable();
983	return i_size;
984	#else
985	/* Pairs with smp_store_release() in i_size_write() */
986	return smp_load_acquire(&inode->i_size);
987	#endif
988	}
989
990	/*
991	* NOTE: unlike i_size_read(), i_size_write() does need locking around it
992	* (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
993	* can be lost, resulting in subsequent i_size_read() calls spinning forever.
994	*/
995	static inline void i_size_write(struct inode *inode, loff_t i_size)
996	{
997	#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
998	preempt_disable();
999	write_seqcount_begin(&inode->i_size_seqcount);
1000	inode->i_size = i_size;
1001	write_seqcount_end(&inode->i_size_seqcount);
1002	preempt_enable();
1003	#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
1004	preempt_disable();
1005	inode->i_size = i_size;
1006	preempt_enable();
1007	#else
1008	/*
1009	* Pairs with smp_load_acquire() in i_size_read() to ensure
1010	* changes related to inode size (such as page contents) are
1011	* visible before we see the changed inode size.
1012	*/
1013	smp_store_release(&inode->i_size, i_size);
1014	#endif
1015	}
1016
1017	static inline unsigned iminor(const struct inode *inode)
1018	{
1019	return MINOR(inode->i_rdev);
1020	}
1021
1022	static inline unsigned imajor(const struct inode *inode)
1023	{
1024	return MAJOR(inode->i_rdev);
1025	}
1026
1027	struct fown_struct {
1028	struct file *file; /* backpointer for security modules */
1029	rwlock_t lock; /* protects pid, uid, euid fields */
1030	struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
1031	enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
1032	kuid_t uid, euid; /* uid/euid of process setting the owner */
1033	int signum; /* posix.1b rt signal to be delivered on IO */
1034	};
1035
1036	/**
1037	* struct file_ra_state - Track a file's readahead state.
1038	* @start: Where the most recent readahead started.
1039	* @size: Number of pages read in the most recent readahead.
1040	* @async_size: Numer of pages that were/are not needed immediately
1041	* and so were/are genuinely "ahead". Start next readahead when
1042	* the first of these pages is accessed.
1043	* @ra_pages: Maximum size of a readahead request, copied from the bdi.
1044	* @mmap_miss: How many mmap accesses missed in the page cache.
1045	* @prev_pos: The last byte in the most recent read request.
1046	*
1047	* When this structure is passed to ->readahead(), the "most recent"
1048	* readahead means the current readahead.
1049	*/
1050	struct file_ra_state {
1051	pgoff_t start;
1052	unsigned int size;
1053	unsigned int async_size;
1054	unsigned int ra_pages;
1055	unsigned int mmap_miss;
1056	loff_t prev_pos;
1057	};
1058
1059	/*
1060	* Check if @index falls in the readahead windows.
1061	*/
1062	static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
1063	{
1064	return (index >= ra->start &&
1065	index < ra->start + ra->size);
1066	}
1067
1068	/**
1069	* struct file - Represents a file
1070	* @f_lock: Protects f_ep, f_flags. Must not be taken from IRQ context.
1071	* @f_mode: FMODE_* flags often used in hotpaths
1072	* @f_op: file operations
1073	* @f_mapping: Contents of a cacheable, mappable object.
1074	* @private_data: filesystem or driver specific data
1075	* @f_inode: cached inode
1076	* @f_flags: file flags
1077	* @f_iocb_flags: iocb flags
1078	* @f_cred: stashed credentials of creator/opener
1079	* @f_owner: file owner
1080	* @f_path: path of the file
1081	* @f_pos_lock: lock protecting file position
1082	* @f_pipe: specific to pipes
1083	* @f_pos: file position
1084	* @f_security: LSM security context of this file
1085	* @f_wb_err: writeback error
1086	* @f_sb_err: per sb writeback errors
1087	* @f_ep: link of all epoll hooks for this file
1088	* @f_task_work: task work entry point
1089	* @f_llist: work queue entrypoint
1090	* @f_ra: file's readahead state
1091	* @f_freeptr: Pointer used by SLAB_TYPESAFE_BY_RCU file cache (don't touch.)
1092	* @f_ref: reference count
1093	*/
1094	struct file {
1095	spinlock_t f_lock;
1096	fmode_t f_mode;
1097	const struct file_operations *f_op;
1098	struct address_space *f_mapping;
1099	void *private_data;
1100	struct inode *f_inode;
1101	unsigned int f_flags;
1102	unsigned int f_iocb_flags;
1103	const struct cred *f_cred;
1104	struct fown_struct *f_owner;
1105	/* --- cacheline 1 boundary (64 bytes) --- */
1106	struct path f_path;
1107	union {
1108	/* regular files (with FMODE_ATOMIC_POS) and directories */
1109	struct mutex f_pos_lock;
1110	/* pipes */
1111	u64 f_pipe;
1112	};
1113	loff_t f_pos;
1114	#ifdef CONFIG_SECURITY
1115	void *f_security;
1116	#endif
1117	/* --- cacheline 2 boundary (128 bytes) --- */
1118	errseq_t f_wb_err;
1119	errseq_t f_sb_err;
1120	#ifdef CONFIG_EPOLL
1121	struct hlist_head *f_ep;
1122	#endif
1123	union {
1124	struct callback_head f_task_work;
1125	struct llist_node f_llist;
1126	struct file_ra_state f_ra;
1127	freeptr_t f_freeptr;
1128	};
1129	file_ref_t f_ref;
1130	/* --- cacheline 3 boundary (192 bytes) --- */
1131	} __randomize_layout
1132	__attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
1133
1134	struct file_handle {
1135	__u32 handle_bytes;
1136	int handle_type;
1137	/* file identifier */
1138	unsigned char f_handle[] __counted_by(handle_bytes);
1139	};
1140
1141	static inline struct file *get_file(struct file *f)
1142	{
1143	file_ref_inc(ref: &f->f_ref);
1144	return f;
1145	}
1146
1147	struct file *get_file_rcu(struct file __rcu **f);
1148	struct file *get_file_active(struct file **f);
1149
1150	#define file_count(f) file_ref_read(&(f)->f_ref)
1151
1152	#define MAX_NON_LFS ((1UL<<31) - 1)
1153
1154	/* Page cache limit. The filesystems should put that into their s_maxbytes
1155	limits, otherwise bad things can happen in VM. */
1156	#if BITS_PER_LONG==32
1157	#define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
1158	#elif BITS_PER_LONG==64
1159	#define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
1160	#endif
1161
1162	/* legacy typedef, should eventually be removed */
1163	typedef void *fl_owner_t;
1164
1165	struct file_lock;
1166	struct file_lease;
1167
1168	/* The following constant reflects the upper bound of the file/locking space */
1169	#ifndef OFFSET_MAX
1170	#define OFFSET_MAX type_max(loff_t)
1171	#define OFFT_OFFSET_MAX type_max(off_t)
1172	#endif
1173
1174	int file_f_owner_allocate(struct file *file);
1175	static inline struct fown_struct *file_f_owner(const struct file *file)
1176	{
1177	return READ_ONCE(file->f_owner);
1178	}
1179
1180	extern void send_sigio(struct fown_struct *fown, int fd, int band);
1181
1182	static inline struct inode *file_inode(const struct file *f)
1183	{
1184	return f->f_inode;
1185	}
1186
1187	/*
1188	* file_dentry() is a relic from the days that overlayfs was using files with a
1189	* "fake" path, meaning, f_path on overlayfs and f_inode on underlying fs.
1190	* In those days, file_dentry() was needed to get the underlying fs dentry that
1191	* matches f_inode.
1192	* Files with "fake" path should not exist nowadays, so use an assertion to make
1193	* sure that file_dentry() was not papering over filesystem bugs.
1194	*/
1195	static inline struct dentry *file_dentry(const struct file *file)
1196	{
1197	struct dentry *dentry = file->f_path.dentry;
1198
1199	WARN_ON_ONCE(d_inode(dentry) != file_inode(file));
1200	return dentry;
1201	}
1202
1203	struct fasync_struct {
1204	rwlock_t fa_lock;
1205	int magic;
1206	int fa_fd;
1207	struct fasync_struct *fa_next; /* singly linked list */
1208	struct file *fa_file;
1209	struct rcu_head fa_rcu;
1210	};
1211
1212	#define FASYNC_MAGIC 0x4601
1213
1214	/* SMP safe fasync helpers: */
1215	extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1216	extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1217	extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1218	extern struct fasync_struct *fasync_alloc(void);
1219	extern void fasync_free(struct fasync_struct *);
1220
1221	/* can be called from interrupts */
1222	extern void kill_fasync(struct fasync_struct **, int, int);
1223
1224	extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1225	extern int f_setown(struct file *filp, int who, int force);
1226	extern void f_delown(struct file *filp);
1227	extern pid_t f_getown(struct file *filp);
1228	extern int send_sigurg(struct file *file);
1229
1230	/*
1231	* sb->s_flags. Note that these mirror the equivalent MS_* flags where
1232	* represented in both.
1233	*/
1234	#define SB_RDONLY BIT(0) /* Mount read-only */
1235	#define SB_NOSUID BIT(1) /* Ignore suid and sgid bits */
1236	#define SB_NODEV BIT(2) /* Disallow access to device special files */
1237	#define SB_NOEXEC BIT(3) /* Disallow program execution */
1238	#define SB_SYNCHRONOUS BIT(4) /* Writes are synced at once */
1239	#define SB_MANDLOCK BIT(6) /* Allow mandatory locks on an FS */
1240	#define SB_DIRSYNC BIT(7) /* Directory modifications are synchronous */
1241	#define SB_NOATIME BIT(10) /* Do not update access times. */
1242	#define SB_NODIRATIME BIT(11) /* Do not update directory access times */
1243	#define SB_SILENT BIT(15)
1244	#define SB_POSIXACL BIT(16) /* Supports POSIX ACLs */
1245	#define SB_INLINECRYPT BIT(17) /* Use blk-crypto for encrypted files */
1246	#define SB_KERNMOUNT BIT(22) /* this is a kern_mount call */
1247	#define SB_I_VERSION BIT(23) /* Update inode I_version field */
1248	#define SB_LAZYTIME BIT(25) /* Update the on-disk [acm]times lazily */
1249
1250	/* These sb flags are internal to the kernel */
1251	#define SB_DEAD BIT(21)
1252	#define SB_DYING BIT(24)
1253	#define SB_FORCE BIT(27)
1254	#define SB_NOSEC BIT(28)
1255	#define SB_BORN BIT(29)
1256	#define SB_ACTIVE BIT(30)
1257	#define SB_NOUSER BIT(31)
1258
1259	/* These flags relate to encoding and casefolding */
1260	#define SB_ENC_STRICT_MODE_FL (1 << 0)
1261	#define SB_ENC_NO_COMPAT_FALLBACK_FL (1 << 1)
1262
1263	#define sb_has_strict_encoding(sb) \
1264	(sb->s_encoding_flags & SB_ENC_STRICT_MODE_FL)
1265
1266	#if IS_ENABLED(CONFIG_UNICODE)
1267	#define sb_no_casefold_compat_fallback(sb) \
1268	(sb->s_encoding_flags & SB_ENC_NO_COMPAT_FALLBACK_FL)
1269	#else
1270	#define sb_no_casefold_compat_fallback(sb) (1)
1271	#endif
1272
1273	/*
1274	* Umount options
1275	*/
1276
1277	#define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1278	#define MNT_DETACH 0x00000002 /* Just detach from the tree */
1279	#define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1280	#define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1281	#define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1282
1283	/* sb->s_iflags */
1284	#define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1285	#define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1286	#define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1287	#define SB_I_STABLE_WRITES 0x00000008 /* don't modify blks until WB is done */
1288
1289	/* sb->s_iflags to limit user namespace mounts */
1290	#define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1291	#define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1292	#define SB_I_UNTRUSTED_MOUNTER 0x00000040
1293	#define SB_I_EVM_HMAC_UNSUPPORTED 0x00000080
1294
1295	#define SB_I_SKIP_SYNC 0x00000100 /* Skip superblock at global sync */
1296	#define SB_I_PERSB_BDI 0x00000200 /* has a per-sb bdi */
1297	#define SB_I_TS_EXPIRY_WARNED 0x00000400 /* warned about timestamp range expiry */
1298	#define SB_I_RETIRED 0x00000800 /* superblock shouldn't be reused */
1299	#define SB_I_NOUMASK 0x00001000 /* VFS does not apply umask */
1300	#define SB_I_NOIDMAP 0x00002000 /* No idmapped mounts on this superblock */
1301	#define SB_I_ALLOW_HSM 0x00004000 /* Allow HSM events on this superblock */
1302
1303	/* Possible states of 'frozen' field */
1304	enum {
1305	SB_UNFROZEN = 0, /* FS is unfrozen */
1306	SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1307	SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1308	SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1309	* internal threads if needed) */
1310	SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1311	};
1312
1313	#define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1314
1315	struct sb_writers {
1316	unsigned short frozen; /* Is sb frozen? */
1317	int freeze_kcount; /* How many kernel freeze requests? */
1318	int freeze_ucount; /* How many userspace freeze requests? */
1319	const void *freeze_owner; /* Owner of the freeze */
1320	struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1321	};
1322
1323	struct super_block {
1324	struct list_head s_list; /* Keep this first */
1325	dev_t s_dev; /* search index; _not_ kdev_t */
1326	unsigned char s_blocksize_bits;
1327	unsigned long s_blocksize;
1328	loff_t s_maxbytes; /* Max file size */
1329	struct file_system_type *s_type;
1330	const struct super_operations *s_op;
1331	const struct dquot_operations *dq_op;
1332	const struct quotactl_ops *s_qcop;
1333	const struct export_operations *s_export_op;
1334	unsigned long s_flags;
1335	unsigned long s_iflags; /* internal SB_I_* flags */
1336	unsigned long s_magic;
1337	struct dentry *s_root;
1338	struct rw_semaphore s_umount;
1339	int s_count;
1340	atomic_t s_active;
1341	#ifdef CONFIG_SECURITY
1342	void *s_security;
1343	#endif
1344	const struct xattr_handler * const *s_xattr;
1345	#ifdef CONFIG_FS_ENCRYPTION
1346	const struct fscrypt_operations *s_cop;
1347	struct fscrypt_keyring *s_master_keys; /* master crypto keys in use */
1348	#endif
1349	#ifdef CONFIG_FS_VERITY
1350	const struct fsverity_operations *s_vop;
1351	#endif
1352	#if IS_ENABLED(CONFIG_UNICODE)
1353	struct unicode_map *s_encoding;
1354	__u16 s_encoding_flags;
1355	#endif
1356	struct hlist_bl_head s_roots; /* alternate root dentries for NFS */
1357	struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1358	struct block_device *s_bdev; /* can go away once we use an accessor for @s_bdev_file */
1359	struct file *s_bdev_file;
1360	struct backing_dev_info *s_bdi;
1361	struct mtd_info *s_mtd;
1362	struct hlist_node s_instances;
1363	unsigned int s_quota_types; /* Bitmask of supported quota types */
1364	struct quota_info s_dquot; /* Diskquota specific options */
1365
1366	struct sb_writers s_writers;
1367
1368	/*
1369	* Keep s_fs_info, s_time_gran, s_fsnotify_mask, and
1370	* s_fsnotify_info together for cache efficiency. They are frequently
1371	* accessed and rarely modified.
1372	*/
1373	void *s_fs_info; /* Filesystem private info */
1374
1375	/* Granularity of c/m/atime in ns (cannot be worse than a second) */
1376	u32 s_time_gran;
1377	/* Time limits for c/m/atime in seconds */
1378	time64_t s_time_min;
1379	time64_t s_time_max;
1380	#ifdef CONFIG_FSNOTIFY
1381	u32 s_fsnotify_mask;
1382	struct fsnotify_sb_info *s_fsnotify_info;
1383	#endif
1384
1385	/*
1386	* q: why are s_id and s_sysfs_name not the same? both are human
1387	* readable strings that identify the filesystem
1388	* a: s_id is allowed to change at runtime; it's used in log messages,
1389	* and we want to when a device starts out as single device (s_id is dev
1390	* name) but then a device is hot added and we have to switch to
1391	* identifying it by UUID
1392	* but s_sysfs_name is a handle for programmatic access, and can't
1393	* change at runtime
1394	*/
1395	char s_id[32]; /* Informational name */
1396	uuid_t s_uuid; /* UUID */
1397	u8 s_uuid_len; /* Default 16, possibly smaller for weird filesystems */
1398
1399	/* if set, fs shows up under sysfs at /sys/fs/$FSTYP/s_sysfs_name */
1400	char s_sysfs_name[UUID_STRING_LEN + 1];
1401
1402	unsigned int s_max_links;
1403
1404	/*
1405	* The next field is for VFS *only*. No filesystems have any business
1406	* even looking at it. You had been warned.
1407	*/
1408	struct mutex s_vfs_rename_mutex; /* Kludge */
1409
1410	/*
1411	* Filesystem subtype. If non-empty the filesystem type field
1412	* in /proc/mounts will be "type.subtype"
1413	*/
1414	const char *s_subtype;
1415
1416	const struct dentry_operations *s_d_op; /* default d_op for dentries */
1417
1418	struct shrinker *s_shrink; /* per-sb shrinker handle */
1419
1420	/* Number of inodes with nlink == 0 but still referenced */
1421	atomic_long_t s_remove_count;
1422
1423	/* Read-only state of the superblock is being changed */
1424	int s_readonly_remount;
1425
1426	/* per-sb errseq_t for reporting writeback errors via syncfs */
1427	errseq_t s_wb_err;
1428
1429	/* AIO completions deferred from interrupt context */
1430	struct workqueue_struct *s_dio_done_wq;
1431	struct hlist_head s_pins;
1432
1433	/*
1434	* Owning user namespace and default context in which to
1435	* interpret filesystem uids, gids, quotas, device nodes,
1436	* xattrs and security labels.
1437	*/
1438	struct user_namespace *s_user_ns;
1439
1440	/*
1441	* The list_lru structure is essentially just a pointer to a table
1442	* of per-node lru lists, each of which has its own spinlock.
1443	* There is no need to put them into separate cachelines.
1444	*/
1445	struct list_lru s_dentry_lru;
1446	struct list_lru s_inode_lru;
1447	struct rcu_head rcu;
1448	struct work_struct destroy_work;
1449
1450	struct mutex s_sync_lock; /* sync serialisation lock */
1451
1452	/*
1453	* Indicates how deep in a filesystem stack this SB is
1454	*/
1455	int s_stack_depth;
1456
1457	/* s_inode_list_lock protects s_inodes */
1458	spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1459	struct list_head s_inodes; /* all inodes */
1460
1461	spinlock_t s_inode_wblist_lock;
1462	struct list_head s_inodes_wb; /* writeback inodes */
1463	} __randomize_layout;
1464
1465	static inline struct user_namespace *i_user_ns(const struct inode *inode)
1466	{
1467	return inode->i_sb->s_user_ns;
1468	}
1469
1470	/* Helper functions so that in most cases filesystems will
1471	* not need to deal directly with kuid_t and kgid_t and can
1472	* instead deal with the raw numeric values that are stored
1473	* in the filesystem.
1474	*/
1475	static inline uid_t i_uid_read(const struct inode *inode)
1476	{
1477	return from_kuid(to: i_user_ns(inode), uid: inode->i_uid);
1478	}
1479
1480	static inline gid_t i_gid_read(const struct inode *inode)
1481	{
1482	return from_kgid(to: i_user_ns(inode), gid: inode->i_gid);
1483	}
1484
1485	static inline void i_uid_write(struct inode *inode, uid_t uid)
1486	{
1487	inode->i_uid = make_kuid(from: i_user_ns(inode), uid);
1488	}
1489
1490	static inline void i_gid_write(struct inode *inode, gid_t gid)
1491	{
1492	inode->i_gid = make_kgid(from: i_user_ns(inode), gid);
1493	}
1494
1495	/**
1496	* i_uid_into_vfsuid - map an inode's i_uid down according to an idmapping
1497	* @idmap: idmap of the mount the inode was found from
1498	* @inode: inode to map
1499	*
1500	* Return: whe inode's i_uid mapped down according to @idmap.
1501	* If the inode's i_uid has no mapping INVALID_VFSUID is returned.
1502	*/
1503	static inline vfsuid_t i_uid_into_vfsuid(struct mnt_idmap *idmap,
1504	const struct inode *inode)
1505	{
1506	return make_vfsuid(idmap, fs_userns: i_user_ns(inode), kuid: inode->i_uid);
1507	}
1508
1509	/**
1510	* i_uid_needs_update - check whether inode's i_uid needs to be updated
1511	* @idmap: idmap of the mount the inode was found from
1512	* @attr: the new attributes of @inode
1513	* @inode: the inode to update
1514	*
1515	* Check whether the $inode's i_uid field needs to be updated taking idmapped
1516	* mounts into account if the filesystem supports it.
1517	*
1518	* Return: true if @inode's i_uid field needs to be updated, false if not.
1519	*/
1520	static inline bool i_uid_needs_update(struct mnt_idmap *idmap,
1521	const struct iattr *attr,
1522	const struct inode *inode)
1523	{
1524	return ((attr->ia_valid & ATTR_UID) &&
1525	!vfsuid_eq(left: attr->ia_vfsuid,
1526	right: i_uid_into_vfsuid(idmap, inode)));
1527	}
1528
1529	/**
1530	* i_uid_update - update @inode's i_uid field
1531	* @idmap: idmap of the mount the inode was found from
1532	* @attr: the new attributes of @inode
1533	* @inode: the inode to update
1534	*
1535	* Safely update @inode's i_uid field translating the vfsuid of any idmapped
1536	* mount into the filesystem kuid.
1537	*/
1538	static inline void i_uid_update(struct mnt_idmap *idmap,
1539	const struct iattr *attr,
1540	struct inode *inode)
1541	{
1542	if (attr->ia_valid & ATTR_UID)
1543	inode->i_uid = from_vfsuid(idmap, fs_userns: i_user_ns(inode),
1544	vfsuid: attr->ia_vfsuid);
1545	}
1546
1547	/**
1548	* i_gid_into_vfsgid - map an inode's i_gid down according to an idmapping
1549	* @idmap: idmap of the mount the inode was found from
1550	* @inode: inode to map
1551	*
1552	* Return: the inode's i_gid mapped down according to @idmap.
1553	* If the inode's i_gid has no mapping INVALID_VFSGID is returned.
1554	*/
1555	static inline vfsgid_t i_gid_into_vfsgid(struct mnt_idmap *idmap,
1556	const struct inode *inode)
1557	{
1558	return make_vfsgid(idmap, fs_userns: i_user_ns(inode), kgid: inode->i_gid);
1559	}
1560
1561	/**
1562	* i_gid_needs_update - check whether inode's i_gid needs to be updated
1563	* @idmap: idmap of the mount the inode was found from
1564	* @attr: the new attributes of @inode
1565	* @inode: the inode to update
1566	*
1567	* Check whether the $inode's i_gid field needs to be updated taking idmapped
1568	* mounts into account if the filesystem supports it.
1569	*
1570	* Return: true if @inode's i_gid field needs to be updated, false if not.
1571	*/
1572	static inline bool i_gid_needs_update(struct mnt_idmap *idmap,
1573	const struct iattr *attr,
1574	const struct inode *inode)
1575	{
1576	return ((attr->ia_valid & ATTR_GID) &&
1577	!vfsgid_eq(left: attr->ia_vfsgid,
1578	right: i_gid_into_vfsgid(idmap, inode)));
1579	}
1580
1581	/**
1582	* i_gid_update - update @inode's i_gid field
1583	* @idmap: idmap of the mount the inode was found from
1584	* @attr: the new attributes of @inode
1585	* @inode: the inode to update
1586	*
1587	* Safely update @inode's i_gid field translating the vfsgid of any idmapped
1588	* mount into the filesystem kgid.
1589	*/
1590	static inline void i_gid_update(struct mnt_idmap *idmap,
1591	const struct iattr *attr,
1592	struct inode *inode)
1593	{
1594	if (attr->ia_valid & ATTR_GID)
1595	inode->i_gid = from_vfsgid(idmap, fs_userns: i_user_ns(inode),
1596	vfsgid: attr->ia_vfsgid);
1597	}
1598
1599	/**
1600	* inode_fsuid_set - initialize inode's i_uid field with callers fsuid
1601	* @inode: inode to initialize
1602	* @idmap: idmap of the mount the inode was found from
1603	*
1604	* Initialize the i_uid field of @inode. If the inode was found/created via
1605	* an idmapped mount map the caller's fsuid according to @idmap.
1606	*/
1607	static inline void inode_fsuid_set(struct inode *inode,
1608	struct mnt_idmap *idmap)
1609	{
1610	inode->i_uid = mapped_fsuid(idmap, fs_userns: i_user_ns(inode));
1611	}
1612
1613	/**
1614	* inode_fsgid_set - initialize inode's i_gid field with callers fsgid
1615	* @inode: inode to initialize
1616	* @idmap: idmap of the mount the inode was found from
1617	*
1618	* Initialize the i_gid field of @inode. If the inode was found/created via
1619	* an idmapped mount map the caller's fsgid according to @idmap.
1620	*/
1621	static inline void inode_fsgid_set(struct inode *inode,
1622	struct mnt_idmap *idmap)
1623	{
1624	inode->i_gid = mapped_fsgid(idmap, fs_userns: i_user_ns(inode));
1625	}
1626
1627	/**
1628	* fsuidgid_has_mapping() - check whether caller's fsuid/fsgid is mapped
1629	* @sb: the superblock we want a mapping in
1630	* @idmap: idmap of the relevant mount
1631	*
1632	* Check whether the caller's fsuid and fsgid have a valid mapping in the
1633	* s_user_ns of the superblock @sb. If the caller is on an idmapped mount map
1634	* the caller's fsuid and fsgid according to the @idmap first.
1635	*
1636	* Return: true if fsuid and fsgid is mapped, false if not.
1637	*/
1638	static inline bool fsuidgid_has_mapping(struct super_block *sb,
1639	struct mnt_idmap *idmap)
1640	{
1641	struct user_namespace *fs_userns = sb->s_user_ns;
1642	kuid_t kuid;
1643	kgid_t kgid;
1644
1645	kuid = mapped_fsuid(idmap, fs_userns);
1646	if (!uid_valid(uid: kuid))
1647	return false;
1648	kgid = mapped_fsgid(idmap, fs_userns);
1649	if (!gid_valid(gid: kgid))
1650	return false;
1651	return kuid_has_mapping(ns: fs_userns, uid: kuid) &&
1652	kgid_has_mapping(ns: fs_userns, gid: kgid);
1653	}
1654
1655	struct timespec64 current_time(struct inode *inode);
1656	struct timespec64 inode_set_ctime_current(struct inode *inode);
1657	struct timespec64 inode_set_ctime_deleg(struct inode *inode,
1658	struct timespec64 update);
1659
1660	static inline time64_t inode_get_atime_sec(const struct inode *inode)
1661	{
1662	return inode->i_atime_sec;
1663	}
1664
1665	static inline long inode_get_atime_nsec(const struct inode *inode)
1666	{
1667	return inode->i_atime_nsec;
1668	}
1669
1670	static inline struct timespec64 inode_get_atime(const struct inode *inode)
1671	{
1672	struct timespec64 ts = { .tv_sec = inode_get_atime_sec(inode),
1673	.tv_nsec = inode_get_atime_nsec(inode) };
1674
1675	return ts;
1676	}
1677
1678	static inline struct timespec64 inode_set_atime_to_ts(struct inode *inode,
1679	struct timespec64 ts)
1680	{
1681	inode->i_atime_sec = ts.tv_sec;
1682	inode->i_atime_nsec = ts.tv_nsec;
1683	return ts;
1684	}
1685
1686	static inline struct timespec64 inode_set_atime(struct inode *inode,
1687	time64_t sec, long nsec)
1688	{
1689	struct timespec64 ts = { .tv_sec = sec,
1690	.tv_nsec = nsec };
1691
1692	return inode_set_atime_to_ts(inode, ts);
1693	}
1694
1695	static inline time64_t inode_get_mtime_sec(const struct inode *inode)
1696	{
1697	return inode->i_mtime_sec;
1698	}
1699
1700	static inline long inode_get_mtime_nsec(const struct inode *inode)
1701	{
1702	return inode->i_mtime_nsec;
1703	}
1704
1705	static inline struct timespec64 inode_get_mtime(const struct inode *inode)
1706	{
1707	struct timespec64 ts = { .tv_sec = inode_get_mtime_sec(inode),
1708	.tv_nsec = inode_get_mtime_nsec(inode) };
1709	return ts;
1710	}
1711
1712	static inline struct timespec64 inode_set_mtime_to_ts(struct inode *inode,
1713	struct timespec64 ts)
1714	{
1715	inode->i_mtime_sec = ts.tv_sec;
1716	inode->i_mtime_nsec = ts.tv_nsec;
1717	return ts;
1718	}
1719
1720	static inline struct timespec64 inode_set_mtime(struct inode *inode,
1721	time64_t sec, long nsec)
1722	{
1723	struct timespec64 ts = { .tv_sec = sec,
1724	.tv_nsec = nsec };
1725	return inode_set_mtime_to_ts(inode, ts);
1726	}
1727
1728	/*
1729	* Multigrain timestamps
1730	*
1731	* Conditionally use fine-grained ctime and mtime timestamps when there
1732	* are users actively observing them via getattr. The primary use-case
1733	* for this is NFS clients that use the ctime to distinguish between
1734	* different states of the file, and that are often fooled by multiple
1735	* operations that occur in the same coarse-grained timer tick.
1736	*/
1737	#define I_CTIME_QUERIED ((u32)BIT(31))
1738
1739	static inline time64_t inode_get_ctime_sec(const struct inode *inode)
1740	{
1741	return inode->i_ctime_sec;
1742	}
1743
1744	static inline long inode_get_ctime_nsec(const struct inode *inode)
1745	{
1746	return inode->i_ctime_nsec & ~I_CTIME_QUERIED;
1747	}
1748
1749	static inline struct timespec64 inode_get_ctime(const struct inode *inode)
1750	{
1751	struct timespec64 ts = { .tv_sec = inode_get_ctime_sec(inode),
1752	.tv_nsec = inode_get_ctime_nsec(inode) };
1753
1754	return ts;
1755	}
1756
1757	struct timespec64 inode_set_ctime_to_ts(struct inode *inode, struct timespec64 ts);
1758
1759	/**
1760	* inode_set_ctime - set the ctime in the inode
1761	* @inode: inode in which to set the ctime
1762	* @sec: tv_sec value to set
1763	* @nsec: tv_nsec value to set
1764	*
1765	* Set the ctime in @inode to { @sec, @nsec }
1766	*/
1767	static inline struct timespec64 inode_set_ctime(struct inode *inode,
1768	time64_t sec, long nsec)
1769	{
1770	struct timespec64 ts = { .tv_sec = sec,
1771	.tv_nsec = nsec };
1772
1773	return inode_set_ctime_to_ts(inode, ts);
1774	}
1775
1776	struct timespec64 simple_inode_init_ts(struct inode *inode);
1777
1778	/*
1779	* Snapshotting support.
1780	*/
1781
1782	/*
1783	* These are internal functions, please use sb_start_{write,pagefault,intwrite}
1784	* instead.
1785	*/
1786	static inline void __sb_end_write(struct super_block *sb, int level)
1787	{
1788	percpu_up_read(sem: sb->s_writers.rw_sem + level-1);
1789	}
1790
1791	static inline void __sb_start_write(struct super_block *sb, int level)
1792	{
1793	percpu_down_read_freezable(sem: sb->s_writers.rw_sem + level - 1, freeze: true);
1794	}
1795
1796	static inline bool __sb_start_write_trylock(struct super_block *sb, int level)
1797	{
1798	return percpu_down_read_trylock(sem: sb->s_writers.rw_sem + level - 1);
1799	}
1800
1801	#define __sb_writers_acquired(sb, lev) \
1802	percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1803	#define __sb_writers_release(sb, lev) \
1804	percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], _THIS_IP_)
1805
1806	/**
1807	* __sb_write_started - check if sb freeze level is held
1808	* @sb: the super we write to
1809	* @level: the freeze level
1810	*
1811	* * > 0 - sb freeze level is held
1812	* * 0 - sb freeze level is not held
1813	* * < 0 - !CONFIG_LOCKDEP/LOCK_STATE_UNKNOWN
1814	*/
1815	static inline int __sb_write_started(const struct super_block *sb, int level)
1816	{
1817	return lockdep_is_held_type(sb->s_writers.rw_sem + level - 1, 1);
1818	}
1819
1820	/**
1821	* sb_write_started - check if SB_FREEZE_WRITE is held
1822	* @sb: the super we write to
1823	*
1824	* May be false positive with !CONFIG_LOCKDEP/LOCK_STATE_UNKNOWN.
1825	*/
1826	static inline bool sb_write_started(const struct super_block *sb)
1827	{
1828	return __sb_write_started(sb, level: SB_FREEZE_WRITE);
1829	}
1830
1831	/**
1832	* sb_write_not_started - check if SB_FREEZE_WRITE is not held
1833	* @sb: the super we write to
1834	*
1835	* May be false positive with !CONFIG_LOCKDEP/LOCK_STATE_UNKNOWN.
1836	*/
1837	static inline bool sb_write_not_started(const struct super_block *sb)
1838	{
1839	return __sb_write_started(sb, level: SB_FREEZE_WRITE) <= 0;
1840	}
1841
1842	/**
1843	* file_write_started - check if SB_FREEZE_WRITE is held
1844	* @file: the file we write to
1845	*
1846	* May be false positive with !CONFIG_LOCKDEP/LOCK_STATE_UNKNOWN.
1847	* May be false positive with !S_ISREG, because file_start_write() has
1848	* no effect on !S_ISREG.
1849	*/
1850	static inline bool file_write_started(const struct file *file)
1851	{
1852	if (!S_ISREG(file_inode(file)->i_mode))
1853	return true;
1854	return sb_write_started(sb: file_inode(f: file)->i_sb);
1855	}
1856
1857	/**
1858	* file_write_not_started - check if SB_FREEZE_WRITE is not held
1859	* @file: the file we write to
1860	*
1861	* May be false positive with !CONFIG_LOCKDEP/LOCK_STATE_UNKNOWN.
1862	* May be false positive with !S_ISREG, because file_start_write() has
1863	* no effect on !S_ISREG.
1864	*/
1865	static inline bool file_write_not_started(const struct file *file)
1866	{
1867	if (!S_ISREG(file_inode(file)->i_mode))
1868	return true;
1869	return sb_write_not_started(sb: file_inode(f: file)->i_sb);
1870	}
1871
1872	/**
1873	* sb_end_write - drop write access to a superblock
1874	* @sb: the super we wrote to
1875	*
1876	* Decrement number of writers to the filesystem. Wake up possible waiters
1877	* wanting to freeze the filesystem.
1878	*/
1879	static inline void sb_end_write(struct super_block *sb)
1880	{
1881	__sb_end_write(sb, level: SB_FREEZE_WRITE);
1882	}
1883
1884	/**
1885	* sb_end_pagefault - drop write access to a superblock from a page fault
1886	* @sb: the super we wrote to
1887	*
1888	* Decrement number of processes handling write page fault to the filesystem.
1889	* Wake up possible waiters wanting to freeze the filesystem.
1890	*/
1891	static inline void sb_end_pagefault(struct super_block *sb)
1892	{
1893	__sb_end_write(sb, level: SB_FREEZE_PAGEFAULT);
1894	}
1895
1896	/**
1897	* sb_end_intwrite - drop write access to a superblock for internal fs purposes
1898	* @sb: the super we wrote to
1899	*
1900	* Decrement fs-internal number of writers to the filesystem. Wake up possible
1901	* waiters wanting to freeze the filesystem.
1902	*/
1903	static inline void sb_end_intwrite(struct super_block *sb)
1904	{
1905	__sb_end_write(sb, level: SB_FREEZE_FS);
1906	}
1907
1908	/**
1909	* sb_start_write - get write access to a superblock
1910	* @sb: the super we write to
1911	*
1912	* When a process wants to write data or metadata to a file system (i.e. dirty
1913	* a page or an inode), it should embed the operation in a sb_start_write() -
1914	* sb_end_write() pair to get exclusion against file system freezing. This
1915	* function increments number of writers preventing freezing. If the file
1916	* system is already frozen, the function waits until the file system is
1917	* thawed.
1918	*
1919	* Since freeze protection behaves as a lock, users have to preserve
1920	* ordering of freeze protection and other filesystem locks. Generally,
1921	* freeze protection should be the outermost lock. In particular, we have:
1922	*
1923	* sb_start_write
1924	* -> i_mutex (write path, truncate, directory ops, ...)
1925	* -> s_umount (freeze_super, thaw_super)
1926	*/
1927	static inline void sb_start_write(struct super_block *sb)
1928	{
1929	__sb_start_write(sb, level: SB_FREEZE_WRITE);
1930	}
1931
1932	static inline bool sb_start_write_trylock(struct super_block *sb)
1933	{
1934	return __sb_start_write_trylock(sb, level: SB_FREEZE_WRITE);
1935	}
1936
1937	/**
1938	* sb_start_pagefault - get write access to a superblock from a page fault
1939	* @sb: the super we write to
1940	*
1941	* When a process starts handling write page fault, it should embed the
1942	* operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1943	* exclusion against file system freezing. This is needed since the page fault
1944	* is going to dirty a page. This function increments number of running page
1945	* faults preventing freezing. If the file system is already frozen, the
1946	* function waits until the file system is thawed.
1947	*
1948	* Since page fault freeze protection behaves as a lock, users have to preserve
1949	* ordering of freeze protection and other filesystem locks. It is advised to
1950	* put sb_start_pagefault() close to mmap_lock in lock ordering. Page fault
1951	* handling code implies lock dependency:
1952	*
1953	* mmap_lock
1954	* -> sb_start_pagefault
1955	*/
1956	static inline void sb_start_pagefault(struct super_block *sb)
1957	{
1958	__sb_start_write(sb, level: SB_FREEZE_PAGEFAULT);
1959	}
1960
1961	/**
1962	* sb_start_intwrite - get write access to a superblock for internal fs purposes
1963	* @sb: the super we write to
1964	*
1965	* This is the third level of protection against filesystem freezing. It is
1966	* free for use by a filesystem. The only requirement is that it must rank
1967	* below sb_start_pagefault.
1968	*
1969	* For example filesystem can call sb_start_intwrite() when starting a
1970	* transaction which somewhat eases handling of freezing for internal sources
1971	* of filesystem changes (internal fs threads, discarding preallocation on file
1972	* close, etc.).
1973	*/
1974	static inline void sb_start_intwrite(struct super_block *sb)
1975	{
1976	__sb_start_write(sb, level: SB_FREEZE_FS);
1977	}
1978
1979	static inline bool sb_start_intwrite_trylock(struct super_block *sb)
1980	{
1981	return __sb_start_write_trylock(sb, level: SB_FREEZE_FS);
1982	}
1983
1984	bool inode_owner_or_capable(struct mnt_idmap *idmap,
1985	const struct inode *inode);
1986
1987	/*
1988	* VFS helper functions..
1989	*/
1990	int vfs_create(struct mnt_idmap *, struct inode *,
1991	struct dentry *, umode_t, bool);
1992	struct dentry *vfs_mkdir(struct mnt_idmap *, struct inode *,
1993	struct dentry *, umode_t);
1994	int vfs_mknod(struct mnt_idmap *, struct inode *, struct dentry *,
1995	umode_t, dev_t);
1996	int vfs_symlink(struct mnt_idmap *, struct inode *,
1997	struct dentry *, const char *);
1998	int vfs_link(struct dentry *, struct mnt_idmap *, struct inode *,
1999	struct dentry *, struct inode **);
2000	int vfs_rmdir(struct mnt_idmap *, struct inode *, struct dentry *);
2001	int vfs_unlink(struct mnt_idmap *, struct inode *, struct dentry *,
2002	struct inode **);
2003
2004	/**
2005	* struct renamedata - contains all information required for renaming
2006	* @old_mnt_idmap: idmap of the old mount the inode was found from
2007	* @old_dir: parent of source
2008	* @old_dentry: source
2009	* @new_mnt_idmap: idmap of the new mount the inode was found from
2010	* @new_dir: parent of destination
2011	* @new_dentry: destination
2012	* @delegated_inode: returns an inode needing a delegation break
2013	* @flags: rename flags
2014	*/
2015	struct renamedata {
2016	struct mnt_idmap *old_mnt_idmap;
2017	struct inode *old_dir;
2018	struct dentry *old_dentry;
2019	struct mnt_idmap *new_mnt_idmap;
2020	struct inode *new_dir;
2021	struct dentry *new_dentry;
2022	struct inode **delegated_inode;
2023	unsigned int flags;
2024	} __randomize_layout;
2025
2026	int vfs_rename(struct renamedata *);
2027
2028	static inline int vfs_whiteout(struct mnt_idmap *idmap,
2029	struct inode *dir, struct dentry *dentry)
2030	{
2031	return vfs_mknod(idmap, dir, dentry, S_IFCHR | WHITEOUT_MODE,
2032	WHITEOUT_DEV);
2033	}
2034
2035	struct file *kernel_tmpfile_open(struct mnt_idmap *idmap,
2036	const struct path *parentpath,
2037	umode_t mode, int open_flag,
2038	const struct cred *cred);
2039	struct file *kernel_file_open(const struct path *path, int flags,
2040	const struct cred *cred);
2041
2042	int vfs_mkobj(struct dentry *, umode_t,
2043	int (*f)(struct dentry *, umode_t, void *),
2044	void *);
2045
2046	int vfs_fchown(struct file *file, uid_t user, gid_t group);
2047	int vfs_fchmod(struct file *file, umode_t mode);
2048	int vfs_utimes(const struct path *path, struct timespec64 *times);
2049
2050	int vfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
2051
2052	#ifdef CONFIG_COMPAT
2053	extern long compat_ptr_ioctl(struct file *file, unsigned int cmd,
2054	unsigned long arg);
2055	#else
2056	#define compat_ptr_ioctl NULL
2057	#endif
2058
2059	/*
2060	* VFS file helper functions.
2061	*/
2062	void inode_init_owner(struct mnt_idmap *idmap, struct inode *inode,
2063	const struct inode *dir, umode_t mode);
2064	extern bool may_open_dev(const struct path *path);
2065	umode_t mode_strip_sgid(struct mnt_idmap *idmap,
2066	const struct inode *dir, umode_t mode);
2067	bool in_group_or_capable(struct mnt_idmap *idmap,
2068	const struct inode *inode, vfsgid_t vfsgid);
2069
2070	/*
2071	* This is the "filldir" function type, used by readdir() to let
2072	* the kernel specify what kind of dirent layout it wants to have.
2073	* This allows the kernel to read directories into kernel space or
2074	* to have different dirent layouts depending on the binary type.
2075	* Return 'true' to keep going and 'false' if there are no more entries.
2076	*/
2077	struct dir_context;
2078	typedef bool (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
2079	unsigned);
2080
2081	struct dir_context {
2082	filldir_t actor;
2083	loff_t pos;
2084	/*
2085	* Filesystems MUST NOT MODIFY count, but may use as a hint:
2086	* 0 unknown
2087	* > 0 space in buffer (assume at least one entry)
2088	* INT_MAX unlimited
2089	*/
2090	int count;
2091	};
2092
2093	/* If OR-ed with d_type, pending signals are not checked */
2094	#define FILLDIR_FLAG_NOINTR 0x1000
2095
2096	/*
2097	* These flags let !MMU mmap() govern direct device mapping vs immediate
2098	* copying more easily for MAP_PRIVATE, especially for ROM filesystems.
2099	*
2100	* NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
2101	* NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
2102	* NOMMU_MAP_READ: Can be mapped for reading
2103	* NOMMU_MAP_WRITE: Can be mapped for writing
2104	* NOMMU_MAP_EXEC: Can be mapped for execution
2105	*/
2106	#define NOMMU_MAP_COPY 0x00000001
2107	#define NOMMU_MAP_DIRECT 0x00000008
2108	#define NOMMU_MAP_READ VM_MAYREAD
2109	#define NOMMU_MAP_WRITE VM_MAYWRITE
2110	#define NOMMU_MAP_EXEC VM_MAYEXEC
2111
2112	#define NOMMU_VMFLAGS \
2113	(NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
2114
2115	/*
2116	* These flags control the behavior of the remap_file_range function pointer.
2117	* If it is called with len == 0 that means "remap to end of source file".
2118	* See Documentation/filesystems/vfs.rst for more details about this call.
2119	*
2120	* REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate)
2121	* REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request
2122	*/
2123	#define REMAP_FILE_DEDUP (1 << 0)
2124	#define REMAP_FILE_CAN_SHORTEN (1 << 1)
2125
2126	/*
2127	* These flags signal that the caller is ok with altering various aspects of
2128	* the behavior of the remap operation. The changes must be made by the
2129	* implementation; the vfs remap helper functions can take advantage of them.
2130	* Flags in this category exist to preserve the quirky behavior of the hoisted
2131	* btrfs clone/dedupe ioctls.
2132	*/
2133	#define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN)
2134
2135	/*
2136	* These flags control the behavior of vfs_copy_file_range().
2137	* They are not available to the user via syscall.
2138	*
2139	* COPY_FILE_SPLICE: call splice direct instead of fs clone/copy ops
2140	*/
2141	#define COPY_FILE_SPLICE (1 << 0)
2142
2143	struct iov_iter;
2144	struct io_uring_cmd;
2145	struct offset_ctx;
2146
2147	typedef unsigned int __bitwise fop_flags_t;
2148
2149	struct file_operations {
2150	struct module *owner;
2151	fop_flags_t fop_flags;
2152	loff_t (*llseek) (struct file *, loff_t, int);
2153	ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
2154	ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
2155	ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
2156	ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
2157	int (*iopoll)(struct kiocb *kiocb, struct io_comp_batch *,
2158	unsigned int flags);
2159	int (*iterate_shared) (struct file *, struct dir_context *);
2160	__poll_t (*poll) (struct file *, struct poll_table_struct *);
2161	long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
2162	long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
2163	int (*mmap) (struct file *, struct vm_area_struct *);
2164	int (*open) (struct inode *, struct file *);
2165	int (*flush) (struct file *, fl_owner_t id);
2166	int (*release) (struct inode *, struct file *);
2167	int (*fsync) (struct file *, loff_t, loff_t, int datasync);
2168	int (*fasync) (int, struct file *, int);
2169	int (*lock) (struct file *, int, struct file_lock *);
2170	unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
2171	int (*check_flags)(int);
2172	int (*flock) (struct file *, int, struct file_lock *);
2173	ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
2174	ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
2175	void (*splice_eof)(struct file *file);
2176	int (*setlease)(struct file *, int, struct file_lease **, void **);
2177	long (*fallocate)(struct file *file, int mode, loff_t offset,
2178	loff_t len);
2179	void (*show_fdinfo)(struct seq_file *m, struct file *f);
2180	#ifndef CONFIG_MMU
2181	unsigned (*mmap_capabilities)(struct file *);
2182	#endif
2183	ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
2184	loff_t, size_t, unsigned int);
2185	loff_t (*remap_file_range)(struct file *file_in, loff_t pos_in,
2186	struct file *file_out, loff_t pos_out,
2187	loff_t len, unsigned int remap_flags);
2188	int (*fadvise)(struct file *, loff_t, loff_t, int);
2189	int (*uring_cmd)(struct io_uring_cmd *ioucmd, unsigned int issue_flags);
2190	int (*uring_cmd_iopoll)(struct io_uring_cmd *, struct io_comp_batch *,
2191	unsigned int poll_flags);
2192	int (*mmap_prepare)(struct vm_area_desc *);
2193	} __randomize_layout;
2194
2195	/* Supports async buffered reads */
2196	#define FOP_BUFFER_RASYNC ((__force fop_flags_t)(1 << 0))
2197	/* Supports async buffered writes */
2198	#define FOP_BUFFER_WASYNC ((__force fop_flags_t)(1 << 1))
2199	/* Supports synchronous page faults for mappings */
2200	#define FOP_MMAP_SYNC ((__force fop_flags_t)(1 << 2))
2201	/* Supports non-exclusive O_DIRECT writes from multiple threads */
2202	#define FOP_DIO_PARALLEL_WRITE ((__force fop_flags_t)(1 << 3))
2203	/* Contains huge pages */
2204	#define FOP_HUGE_PAGES ((__force fop_flags_t)(1 << 4))
2205	/* Treat loff_t as unsigned (e.g., /dev/mem) */
2206	#define FOP_UNSIGNED_OFFSET ((__force fop_flags_t)(1 << 5))
2207	/* Supports asynchronous lock callbacks */
2208	#define FOP_ASYNC_LOCK ((__force fop_flags_t)(1 << 6))
2209	/* File system supports uncached read/write buffered IO */
2210	#define FOP_DONTCACHE ((__force fop_flags_t)(1 << 7))
2211
2212	/* Wrap a directory iterator that needs exclusive inode access */
2213	int wrap_directory_iterator(struct file *, struct dir_context *,
2214	int (*) (struct file *, struct dir_context *));
2215	#define WRAP_DIR_ITER(x) \
2216	static int shared_##x(struct file *file , struct dir_context *ctx) \
2217	{ return wrap_directory_iterator(file, ctx, x); }
2218
2219	struct inode_operations {
2220	struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
2221	const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
2222	int (*permission) (struct mnt_idmap *, struct inode *, int);
2223	struct posix_acl * (*get_inode_acl)(struct inode *, int, bool);
2224
2225	int (*readlink) (struct dentry *, char __user *,int);
2226
2227	int (*create) (struct mnt_idmap *, struct inode *,struct dentry *,
2228	umode_t, bool);
2229	int (*link) (struct dentry *,struct inode *,struct dentry *);
2230	int (*unlink) (struct inode *,struct dentry *);
2231	int (*symlink) (struct mnt_idmap *, struct inode *,struct dentry *,
2232	const char *);
2233	struct dentry *(*mkdir) (struct mnt_idmap *, struct inode *,
2234	struct dentry *, umode_t);
2235	int (*rmdir) (struct inode *,struct dentry *);
2236	int (*mknod) (struct mnt_idmap *, struct inode *,struct dentry *,
2237	umode_t,dev_t);
2238	int (*rename) (struct mnt_idmap *, struct inode *, struct dentry *,
2239	struct inode *, struct dentry *, unsigned int);
2240	int (*setattr) (struct mnt_idmap *, struct dentry *, struct iattr *);
2241	int (*getattr) (struct mnt_idmap *, const struct path *,
2242	struct kstat *, u32, unsigned int);
2243	ssize_t (*listxattr) (struct dentry *, char *, size_t);
2244	int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
2245	u64 len);
2246	int (*update_time)(struct inode *, int);
2247	int (*atomic_open)(struct inode *, struct dentry *,
2248	struct file *, unsigned open_flag,
2249	umode_t create_mode);
2250	int (*tmpfile) (struct mnt_idmap *, struct inode *,
2251	struct file *, umode_t);
2252	struct posix_acl *(*get_acl)(struct mnt_idmap *, struct dentry *,
2253	int);
2254	int (*set_acl)(struct mnt_idmap *, struct dentry *,
2255	struct posix_acl *, int);
2256	int (*fileattr_set)(struct mnt_idmap *idmap,
2257	struct dentry *dentry, struct fileattr *fa);
2258	int (*fileattr_get)(struct dentry *dentry, struct fileattr *fa);
2259	struct offset_ctx *(*get_offset_ctx)(struct inode *inode);
2260	} ____cacheline_aligned;
2261
2262	/* Did the driver provide valid mmap hook configuration? */
2263	static inline bool file_has_valid_mmap_hooks(struct file *file)
2264	{
2265	bool has_mmap = file->f_op->mmap;
2266	bool has_mmap_prepare = file->f_op->mmap_prepare;
2267
2268	/* Hooks are mutually exclusive. */
2269	if (WARN_ON_ONCE(has_mmap && has_mmap_prepare))
2270	return false;
2271	if (!has_mmap && !has_mmap_prepare)
2272	return false;
2273
2274	return true;
2275	}
2276
2277	static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
2278	{
2279	if (WARN_ON_ONCE(file->f_op->mmap_prepare))
2280	return -EINVAL;
2281
2282	return file->f_op->mmap(file, vma);
2283	}
2284
2285	static inline int __call_mmap_prepare(struct file *file,
2286	struct vm_area_desc *desc)
2287	{
2288	return file->f_op->mmap_prepare(desc);
2289	}
2290
2291	extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
2292	extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
2293	extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
2294	loff_t, size_t, unsigned int);
2295	int remap_verify_area(struct file *file, loff_t pos, loff_t len, bool write);
2296	int __generic_remap_file_range_prep(struct file *file_in, loff_t pos_in,
2297	struct file *file_out, loff_t pos_out,
2298	loff_t *len, unsigned int remap_flags,
2299	const struct iomap_ops *dax_read_ops);
2300	int generic_remap_file_range_prep(struct file *file_in, loff_t pos_in,
2301	struct file *file_out, loff_t pos_out,
2302	loff_t *count, unsigned int remap_flags);
2303	extern loff_t vfs_clone_file_range(struct file *file_in, loff_t pos_in,
2304	struct file *file_out, loff_t pos_out,
2305	loff_t len, unsigned int remap_flags);
2306	extern int vfs_dedupe_file_range(struct file *file,
2307	struct file_dedupe_range *same);
2308	extern loff_t vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos,
2309	struct file *dst_file, loff_t dst_pos,
2310	loff_t len, unsigned int remap_flags);
2311
2312	/**
2313	* enum freeze_holder - holder of the freeze
2314	* @FREEZE_HOLDER_KERNEL: kernel wants to freeze or thaw filesystem
2315	* @FREEZE_HOLDER_USERSPACE: userspace wants to freeze or thaw filesystem
2316	* @FREEZE_MAY_NEST: whether nesting freeze and thaw requests is allowed
2317	* @FREEZE_EXCL: a freeze that can only be undone by the owner
2318	*
2319	* Indicate who the owner of the freeze or thaw request is and whether
2320	* the freeze needs to be exclusive or can nest.
2321	* Without @FREEZE_MAY_NEST, multiple freeze and thaw requests from the
2322	* same holder aren't allowed. It is however allowed to hold a single
2323	* @FREEZE_HOLDER_USERSPACE and a single @FREEZE_HOLDER_KERNEL freeze at
2324	* the same time. This is relied upon by some filesystems during online
2325	* repair or similar.
2326	*/
2327	enum freeze_holder {
2328	FREEZE_HOLDER_KERNEL = (1U << 0),
2329	FREEZE_HOLDER_USERSPACE = (1U << 1),
2330	FREEZE_MAY_NEST = (1U << 2),
2331	FREEZE_EXCL = (1U << 3),
2332	};
2333
2334	struct super_operations {
2335	struct inode *(*alloc_inode)(struct super_block *sb);
2336	void (*destroy_inode)(struct inode *);
2337	void (*free_inode)(struct inode *);
2338
2339	void (*dirty_inode) (struct inode *, int flags);
2340	int (*write_inode) (struct inode *, struct writeback_control *wbc);
2341	int (*drop_inode) (struct inode *);
2342	void (*evict_inode) (struct inode *);
2343	void (*put_super) (struct super_block *);
2344	int (*sync_fs)(struct super_block *sb, int wait);
2345	int (*freeze_super) (struct super_block *, enum freeze_holder who, const void *owner);
2346	int (*freeze_fs) (struct super_block *);
2347	int (*thaw_super) (struct super_block *, enum freeze_holder who, const void *owner);
2348	int (*unfreeze_fs) (struct super_block *);
2349	int (*statfs) (struct dentry *, struct kstatfs *);
2350	int (*remount_fs) (struct super_block *, int *, char *);
2351	void (*umount_begin) (struct super_block *);
2352
2353	int (*show_options)(struct seq_file *, struct dentry *);
2354	int (*show_devname)(struct seq_file *, struct dentry *);
2355	int (*show_path)(struct seq_file *, struct dentry *);
2356	int (*show_stats)(struct seq_file *, struct dentry *);
2357	#ifdef CONFIG_QUOTA
2358	ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
2359	ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
2360	struct dquot __rcu **(*get_dquots)(struct inode *);
2361	#endif
2362	long (*nr_cached_objects)(struct super_block *,
2363	struct shrink_control *);
2364	long (*free_cached_objects)(struct super_block *,
2365	struct shrink_control *);
2366	void (*shutdown)(struct super_block *sb);
2367	};
2368
2369	/*
2370	* Inode flags - they have no relation to superblock flags now
2371	*/
2372	#define S_SYNC (1 << 0) /* Writes are synced at once */
2373	#define S_NOATIME (1 << 1) /* Do not update access times */
2374	#define S_APPEND (1 << 2) /* Append-only file */
2375	#define S_IMMUTABLE (1 << 3) /* Immutable file */
2376	#define S_DEAD (1 << 4) /* removed, but still open directory */
2377	#define S_NOQUOTA (1 << 5) /* Inode is not counted to quota */
2378	#define S_DIRSYNC (1 << 6) /* Directory modifications are synchronous */
2379	#define S_NOCMTIME (1 << 7) /* Do not update file c/mtime */
2380	#define S_SWAPFILE (1 << 8) /* Do not truncate: swapon got its bmaps */
2381	#define S_PRIVATE (1 << 9) /* Inode is fs-internal */
2382	#define S_IMA (1 << 10) /* Inode has an associated IMA struct */
2383	#define S_AUTOMOUNT (1 << 11) /* Automount/referral quasi-directory */
2384	#define S_NOSEC (1 << 12) /* no suid or xattr security attributes */
2385	#ifdef CONFIG_FS_DAX
2386	#define S_DAX (1 << 13) /* Direct Access, avoiding the page cache */
2387	#else
2388	#define S_DAX 0 /* Make all the DAX code disappear */
2389	#endif
2390	#define S_ENCRYPTED (1 << 14) /* Encrypted file (using fs/crypto/) */
2391	#define S_CASEFOLD (1 << 15) /* Casefolded file */
2392	#define S_VERITY (1 << 16) /* Verity file (using fs/verity/) */
2393	#define S_KERNEL_FILE (1 << 17) /* File is in use by the kernel (eg. fs/cachefiles) */
2394	#define S_ANON_INODE (1 << 19) /* Inode is an anonymous inode */
2395
2396	/*
2397	* Note that nosuid etc flags are inode-specific: setting some file-system
2398	* flags just means all the inodes inherit those flags by default. It might be
2399	* possible to override it selectively if you really wanted to with some
2400	* ioctl() that is not currently implemented.
2401	*
2402	* Exception: SB_RDONLY is always applied to the entire file system.
2403	*
2404	* Unfortunately, it is possible to change a filesystems flags with it mounted
2405	* with files in use. This means that all of the inodes will not have their
2406	* i_flags updated. Hence, i_flags no longer inherit the superblock mount
2407	* flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
2408	*/
2409	#define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
2410
2411	static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; }
2412	#define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
2413	#define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
2414	((inode)->i_flags & S_SYNC))
2415	#define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
2416	((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
2417	#define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
2418	#define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
2419	#define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
2420
2421	#define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
2422	#define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
2423	#define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
2424
2425	#ifdef CONFIG_FS_POSIX_ACL
2426	#define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
2427	#else
2428	#define IS_POSIXACL(inode) 0
2429	#endif
2430
2431	#define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
2432	#define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
2433
2434	#ifdef CONFIG_SWAP
2435	#define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
2436	#else
2437	#define IS_SWAPFILE(inode) ((void)(inode), 0U)
2438	#endif
2439
2440	#define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
2441	#define IS_IMA(inode) ((inode)->i_flags & S_IMA)
2442	#define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
2443	#define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
2444	#define IS_DAX(inode) ((inode)->i_flags & S_DAX)
2445	#define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
2446	#define IS_CASEFOLDED(inode) ((inode)->i_flags & S_CASEFOLD)
2447	#define IS_VERITY(inode) ((inode)->i_flags & S_VERITY)
2448
2449	#define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
2450	(inode)->i_rdev == WHITEOUT_DEV)
2451	#define IS_ANON_FILE(inode) ((inode)->i_flags & S_ANON_INODE)
2452
2453	static inline bool HAS_UNMAPPED_ID(struct mnt_idmap *idmap,
2454	struct inode *inode)
2455	{
2456	return !vfsuid_valid(uid: i_uid_into_vfsuid(idmap, inode)) ||
2457	!vfsgid_valid(gid: i_gid_into_vfsgid(idmap, inode));
2458	}
2459
2460	static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
2461	{
2462	*kiocb = (struct kiocb) {
2463	.ki_filp = filp,
2464	.ki_flags = filp->f_iocb_flags,
2465	.ki_ioprio = get_current_ioprio(),
2466	};
2467	}
2468
2469	static inline void kiocb_clone(struct kiocb *kiocb, struct kiocb *kiocb_src,
2470	struct file *filp)
2471	{
2472	*kiocb = (struct kiocb) {
2473	.ki_filp = filp,
2474	.ki_flags = kiocb_src->ki_flags,
2475	.ki_ioprio = kiocb_src->ki_ioprio,
2476	.ki_pos = kiocb_src->ki_pos,
2477	};
2478	}
2479
2480	/*
2481	* Inode state bits. Protected by inode->i_lock
2482	*
2483	* Four bits determine the dirty state of the inode: I_DIRTY_SYNC,
2484	* I_DIRTY_DATASYNC, I_DIRTY_PAGES, and I_DIRTY_TIME.
2485	*
2486	* Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
2487	* until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
2488	* various stages of removing an inode.
2489	*
2490	* Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2491	*
2492	* I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
2493	* fdatasync() (unless I_DIRTY_DATASYNC is also set).
2494	* Timestamp updates are the usual cause.
2495	* I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
2496	* these changes separately from I_DIRTY_SYNC so that we
2497	* don't have to write inode on fdatasync() when only
2498	* e.g. the timestamps have changed.
2499	* I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
2500	* I_DIRTY_TIME The inode itself has dirty timestamps, and the
2501	* lazytime mount option is enabled. We keep track of this
2502	* separately from I_DIRTY_SYNC in order to implement
2503	* lazytime. This gets cleared if I_DIRTY_INODE
2504	* (I_DIRTY_SYNC and/or I_DIRTY_DATASYNC) gets set. But
2505	* I_DIRTY_TIME can still be set if I_DIRTY_SYNC is already
2506	* in place because writeback might already be in progress
2507	* and we don't want to lose the time update
2508	* I_NEW Serves as both a mutex and completion notification.
2509	* New inodes set I_NEW. If two processes both create
2510	* the same inode, one of them will release its inode and
2511	* wait for I_NEW to be released before returning.
2512	* Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2513	* also cause waiting on I_NEW, without I_NEW actually
2514	* being set. find_inode() uses this to prevent returning
2515	* nearly-dead inodes.
2516	* I_WILL_FREE Must be set when calling write_inode_now() if i_count
2517	* is zero. I_FREEING must be set when I_WILL_FREE is
2518	* cleared.
2519	* I_FREEING Set when inode is about to be freed but still has dirty
2520	* pages or buffers attached or the inode itself is still
2521	* dirty.
2522	* I_CLEAR Added by clear_inode(). In this state the inode is
2523	* clean and can be destroyed. Inode keeps I_FREEING.
2524	*
2525	* Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2526	* prohibited for many purposes. iget() must wait for
2527	* the inode to be completely released, then create it
2528	* anew. Other functions will just ignore such inodes,
2529	* if appropriate. I_NEW is used for waiting.
2530	*
2531	* I_SYNC Writeback of inode is running. The bit is set during
2532	* data writeback, and cleared with a wakeup on the bit
2533	* address once it is done. The bit is also used to pin
2534	* the inode in memory for flusher thread.
2535	*
2536	* I_REFERENCED Marks the inode as recently references on the LRU list.
2537	*
2538	* I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2539	* synchronize competing switching instances and to tell
2540	* wb stat updates to grab the i_pages lock. See
2541	* inode_switch_wbs_work_fn() for details.
2542	*
2543	* I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2544	* and work dirs among overlayfs mounts.
2545	*
2546	* I_CREATING New object's inode in the middle of setting up.
2547	*
2548	* I_DONTCACHE Evict inode as soon as it is not used anymore.
2549	*
2550	* I_SYNC_QUEUED Inode is queued in b_io or b_more_io writeback lists.
2551	* Used to detect that mark_inode_dirty() should not move
2552	* inode between dirty lists.
2553	*
2554	* I_PINNING_FSCACHE_WB Inode is pinning an fscache object for writeback.
2555	*
2556	* I_LRU_ISOLATING Inode is pinned being isolated from LRU without holding
2557	* i_count.
2558	*
2559	* Q: What is the difference between I_WILL_FREE and I_FREEING?
2560	*
2561	* __I_{SYNC,NEW,LRU_ISOLATING} are used to derive unique addresses to wait
2562	* upon. There's one free address left.
2563	*/
2564	#define __I_NEW 0
2565	#define I_NEW (1 << __I_NEW)
2566	#define __I_SYNC 1
2567	#define I_SYNC (1 << __I_SYNC)
2568	#define __I_LRU_ISOLATING 2
2569	#define I_LRU_ISOLATING (1 << __I_LRU_ISOLATING)
2570
2571	#define I_DIRTY_SYNC (1 << 3)
2572	#define I_DIRTY_DATASYNC (1 << 4)
2573	#define I_DIRTY_PAGES (1 << 5)
2574	#define I_WILL_FREE (1 << 6)
2575	#define I_FREEING (1 << 7)
2576	#define I_CLEAR (1 << 8)
2577	#define I_REFERENCED (1 << 9)
2578	#define I_LINKABLE (1 << 10)
2579	#define I_DIRTY_TIME (1 << 11)
2580	#define I_WB_SWITCH (1 << 12)
2581	#define I_OVL_INUSE (1 << 13)
2582	#define I_CREATING (1 << 14)
2583	#define I_DONTCACHE (1 << 15)
2584	#define I_SYNC_QUEUED (1 << 16)
2585	#define I_PINNING_NETFS_WB (1 << 17)
2586
2587	#define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2588	#define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2589	#define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2590
2591	extern void __mark_inode_dirty(struct inode *, int);
2592	static inline void mark_inode_dirty(struct inode *inode)
2593	{
2594	__mark_inode_dirty(inode, I_DIRTY);
2595	}
2596
2597	static inline void mark_inode_dirty_sync(struct inode *inode)
2598	{
2599	__mark_inode_dirty(inode, I_DIRTY_SYNC);
2600	}
2601
2602	/*
2603	* Returns true if the given inode itself only has dirty timestamps (its pages
2604	* may still be dirty) and isn't currently being allocated or freed.
2605	* Filesystems should call this if when writing an inode when lazytime is
2606	* enabled, they want to opportunistically write the timestamps of other inodes
2607	* located very nearby on-disk, e.g. in the same inode block. This returns true
2608	* if the given inode is in need of such an opportunistic update. Requires
2609	* i_lock, or at least later re-checking under i_lock.
2610	*/
2611	static inline bool inode_is_dirtytime_only(struct inode *inode)
2612	{
2613	return (inode->i_state & (I_DIRTY_TIME | I_NEW |
2614	I_FREEING | I_WILL_FREE)) == I_DIRTY_TIME;
2615	}
2616
2617	extern void inc_nlink(struct inode *inode);
2618	extern void drop_nlink(struct inode *inode);
2619	extern void clear_nlink(struct inode *inode);
2620	extern void set_nlink(struct inode *inode, unsigned int nlink);
2621
2622	static inline void inode_inc_link_count(struct inode *inode)
2623	{
2624	inc_nlink(inode);
2625	mark_inode_dirty(inode);
2626	}
2627
2628	static inline void inode_dec_link_count(struct inode *inode)
2629	{
2630	drop_nlink(inode);
2631	mark_inode_dirty(inode);
2632	}
2633
2634	enum file_time_flags {
2635	S_ATIME = 1,
2636	S_MTIME = 2,
2637	S_CTIME = 4,
2638	S_VERSION = 8,
2639	};
2640
2641	extern bool atime_needs_update(const struct path *, struct inode *);
2642	extern void touch_atime(const struct path *);
2643	int inode_update_time(struct inode *inode, int flags);
2644
2645	static inline void file_accessed(struct file *file)
2646	{
2647	if (!(file->f_flags & O_NOATIME))
2648	touch_atime(&file->f_path);
2649	}
2650
2651	extern int file_modified(struct file *file);
2652	int kiocb_modified(struct kiocb *iocb);
2653
2654	int sync_inode_metadata(struct inode *inode, int wait);
2655
2656	struct file_system_type {
2657	const char *name;
2658	int fs_flags;
2659	#define FS_REQUIRES_DEV 1
2660	#define FS_BINARY_MOUNTDATA 2
2661	#define FS_HAS_SUBTYPE 4
2662	#define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2663	#define FS_DISALLOW_NOTIFY_PERM 16 /* Disable fanotify permission events */
2664	#define FS_ALLOW_IDMAP 32 /* FS has been updated to handle vfs idmappings. */
2665	#define FS_MGTIME 64 /* FS uses multigrain timestamps */
2666	#define FS_LBS 128 /* FS supports LBS */
2667	#define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2668	int (*init_fs_context)(struct fs_context *);
2669	const struct fs_parameter_spec *parameters;
2670	struct dentry *(*mount) (struct file_system_type *, int,
2671	const char *, void *);
2672	void (*kill_sb) (struct super_block *);
2673	struct module *owner;
2674	struct file_system_type * next;
2675	struct hlist_head fs_supers;
2676
2677	struct lock_class_key s_lock_key;
2678	struct lock_class_key s_umount_key;
2679	struct lock_class_key s_vfs_rename_key;
2680	struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2681
2682	struct lock_class_key i_lock_key;
2683	struct lock_class_key i_mutex_key;
2684	struct lock_class_key invalidate_lock_key;
2685	struct lock_class_key i_mutex_dir_key;
2686	};
2687
2688	#define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2689
2690	/**
2691	* is_mgtime: is this inode using multigrain timestamps
2692	* @inode: inode to test for multigrain timestamps
2693	*
2694	* Return true if the inode uses multigrain timestamps, false otherwise.
2695	*/
2696	static inline bool is_mgtime(const struct inode *inode)
2697	{
2698	return inode->i_opflags & IOP_MGTIME;
2699	}
2700
2701	extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2702	int flags, const char *dev_name, void *data,
2703	int (*fill_super)(struct super_block *, void *, int));
2704	extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2705	int flags, void *data,
2706	int (*fill_super)(struct super_block *, void *, int));
2707	extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2708	void retire_super(struct super_block *sb);
2709	void generic_shutdown_super(struct super_block *sb);
2710	void kill_block_super(struct super_block *sb);
2711	void kill_anon_super(struct super_block *sb);
2712	void kill_litter_super(struct super_block *sb);
2713	void deactivate_super(struct super_block *sb);
2714	void deactivate_locked_super(struct super_block *sb);
2715	int set_anon_super(struct super_block *s, void *data);
2716	int set_anon_super_fc(struct super_block *s, struct fs_context *fc);
2717	int get_anon_bdev(dev_t *);
2718	void free_anon_bdev(dev_t);
2719	struct super_block *sget_fc(struct fs_context *fc,
2720	int (*test)(struct super_block *, struct fs_context *),
2721	int (*set)(struct super_block *, struct fs_context *));
2722	struct super_block *sget(struct file_system_type *type,
2723	int (*test)(struct super_block *,void *),
2724	int (*set)(struct super_block *,void *),
2725	int flags, void *data);
2726	struct super_block *sget_dev(struct fs_context *fc, dev_t dev);
2727
2728	/* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2729	#define fops_get(fops) ({ \
2730	const struct file_operations *_fops = (fops); \
2731	(((_fops) && try_module_get((_fops)->owner) ? (_fops) : NULL)); \
2732	})
2733
2734	#define fops_put(fops) ({ \
2735	const struct file_operations *_fops = (fops); \
2736	if (_fops) \
2737	module_put((_fops)->owner); \
2738	})
2739
2740	/*
2741	* This one is to be used *ONLY* from ->open() instances.
2742	* fops must be non-NULL, pinned down *and* module dependencies
2743	* should be sufficient to pin the caller down as well.
2744	*/
2745	#define replace_fops(f, fops) \
2746	do { \
2747	struct file *__file = (f); \
2748	fops_put(__file->f_op); \
2749	BUG_ON(!(__file->f_op = (fops))); \
2750	} while(0)
2751
2752	extern int register_filesystem(struct file_system_type *);
2753	extern int unregister_filesystem(struct file_system_type *);
2754	extern int vfs_statfs(const struct path *, struct kstatfs *);
2755	extern int user_statfs(const char __user *, struct kstatfs *);
2756	extern int fd_statfs(int, struct kstatfs *);
2757	int freeze_super(struct super_block *super, enum freeze_holder who,
2758	const void *freeze_owner);
2759	int thaw_super(struct super_block *super, enum freeze_holder who,
2760	const void *freeze_owner);
2761	extern __printf(2, 3)
2762	int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2763	extern int super_setup_bdi(struct super_block *sb);
2764
2765	static inline void super_set_uuid(struct super_block *sb, const u8 *uuid, unsigned len)
2766	{
2767	if (WARN_ON(len > sizeof(sb->s_uuid)))
2768	len = sizeof(sb->s_uuid);
2769	sb->s_uuid_len = len;
2770	memcpy(&sb->s_uuid, uuid, len);
2771	}
2772
2773	/* set sb sysfs name based on sb->s_bdev */
2774	static inline void super_set_sysfs_name_bdev(struct super_block *sb)
2775	{
2776	snprintf(buf: sb->s_sysfs_name, size: sizeof(sb->s_sysfs_name), fmt: "%pg", sb->s_bdev);
2777	}
2778
2779	/* set sb sysfs name based on sb->s_uuid */
2780	static inline void super_set_sysfs_name_uuid(struct super_block *sb)
2781	{
2782	WARN_ON(sb->s_uuid_len != sizeof(sb->s_uuid));
2783	snprintf(buf: sb->s_sysfs_name, size: sizeof(sb->s_sysfs_name), fmt: "%pU", sb->s_uuid.b);
2784	}
2785
2786	/* set sb sysfs name based on sb->s_id */
2787	static inline void super_set_sysfs_name_id(struct super_block *sb)
2788	{
2789	strscpy(sb->s_sysfs_name, sb->s_id, sizeof(sb->s_sysfs_name));
2790	}
2791
2792	/* try to use something standard before you use this */
2793	__printf(2, 3)
2794	static inline void super_set_sysfs_name_generic(struct super_block *sb, const char *fmt, ...)
2795	{
2796	va_list args;
2797
2798	va_start(args, fmt);
2799	vsnprintf(buf: sb->s_sysfs_name, size: sizeof(sb->s_sysfs_name), fmt, args);
2800	va_end(args);
2801	}
2802
2803	extern int current_umask(void);
2804
2805	extern void ihold(struct inode * inode);
2806	extern void iput(struct inode *);
2807	int inode_update_timestamps(struct inode *inode, int flags);
2808	int generic_update_time(struct inode *, int);
2809
2810	/* /sys/fs */
2811	extern struct kobject *fs_kobj;
2812
2813	#define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2814
2815	/* fs/open.c */
2816	struct audit_names;
2817	struct filename {
2818	const char *name; /* pointer to actual string */
2819	const __user char *uptr; /* original userland pointer */
2820	atomic_t refcnt;
2821	struct audit_names *aname;
2822	const char iname[];
2823	};
2824	static_assert(offsetof(struct filename, iname) % sizeof(long) == 0);
2825
2826	static inline struct mnt_idmap *file_mnt_idmap(const struct file *file)
2827	{
2828	return mnt_idmap(mnt: file->f_path.mnt);
2829	}
2830
2831	/**
2832	* is_idmapped_mnt - check whether a mount is mapped
2833	* @mnt: the mount to check
2834	*
2835	* If @mnt has an non @nop_mnt_idmap attached to it then @mnt is mapped.
2836	*
2837	* Return: true if mount is mapped, false if not.
2838	*/
2839	static inline bool is_idmapped_mnt(const struct vfsmount *mnt)
2840	{
2841	return mnt_idmap(mnt) != &nop_mnt_idmap;
2842	}
2843
2844	int vfs_truncate(const struct path *, loff_t);
2845	int do_truncate(struct mnt_idmap *, struct dentry *, loff_t start,
2846	unsigned int time_attrs, struct file *filp);
2847	extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2848	loff_t len);
2849	int do_sys_open(int dfd, const char __user *filename, int flags,
2850	umode_t mode);
2851	extern struct file *file_open_name(struct filename *, int, umode_t);
2852	extern struct file *filp_open(const char *, int, umode_t);
2853	extern struct file *file_open_root(const struct path *,
2854	const char *, int, umode_t);
2855	static inline struct file *file_open_root_mnt(struct vfsmount *mnt,
2856	const char *name, int flags, umode_t mode)
2857	{
2858	return file_open_root(&(struct path){.mnt = mnt, .dentry = mnt->mnt_root},
2859	name, flags, mode);
2860	}
2861	struct file *dentry_open(const struct path *path, int flags,
2862	const struct cred *creds);
2863	struct file *dentry_open_nonotify(const struct path *path, int flags,
2864	const struct cred *cred);
2865	struct file *dentry_create(const struct path *path, int flags, umode_t mode,
2866	const struct cred *cred);
2867	struct path *backing_file_user_path(struct file *f);
2868
2869	/*
2870	* When mmapping a file on a stackable filesystem (e.g., overlayfs), the file
2871	* stored in ->vm_file is a backing file whose f_inode is on the underlying
2872	* filesystem. When the mapped file path and inode number are displayed to
2873	* user (e.g. via /proc/<pid>/maps), these helpers should be used to get the
2874	* path and inode number to display to the user, which is the path of the fd
2875	* that user has requested to map and the inode number that would be returned
2876	* by fstat() on that same fd.
2877	*/
2878	/* Get the path to display in /proc/<pid>/maps */
2879	static inline const struct path *file_user_path(struct file *f)
2880	{
2881	if (unlikely(f->f_mode & FMODE_BACKING))
2882	return backing_file_user_path(f);
2883	return &f->f_path;
2884	}
2885	/* Get the inode whose inode number to display in /proc/<pid>/maps */
2886	static inline const struct inode *file_user_inode(struct file *f)
2887	{
2888	if (unlikely(f->f_mode & FMODE_BACKING))
2889	return d_inode(dentry: backing_file_user_path(f)->dentry);
2890	return file_inode(f);
2891	}
2892
2893	static inline struct file *file_clone_open(struct file *file)
2894	{
2895	return dentry_open(path: &file->f_path, flags: file->f_flags, creds: file->f_cred);
2896	}
2897	extern int filp_close(struct file *, fl_owner_t id);
2898
2899	extern struct filename *getname_flags(const char __user *, int);
2900	extern struct filename *getname_uflags(const char __user *, int);
2901	static inline struct filename *getname(const char __user *name)
2902	{
2903	return getname_flags(name, 0);
2904	}
2905	extern struct filename *getname_kernel(const char *);
2906	extern struct filename *__getname_maybe_null(const char __user *);
2907	static inline struct filename *getname_maybe_null(const char __user *name, int flags)
2908	{
2909	if (!(flags & AT_EMPTY_PATH))
2910	return getname(name);
2911
2912	if (!name)
2913	return NULL;
2914	return __getname_maybe_null(name);
2915	}
2916	extern void putname(struct filename *name);
2917	DEFINE_FREE(putname, struct filename *, if (!IS_ERR_OR_NULL(_T)) putname(_T))
2918
2919	static inline struct filename *refname(struct filename *name)
2920	{
2921	atomic_inc(v: &name->refcnt);
2922	return name;
2923	}
2924
2925	extern int finish_open(struct file *file, struct dentry *dentry,
2926	int (*open)(struct inode *, struct file *));
2927	extern int finish_no_open(struct file *file, struct dentry *dentry);
2928
2929	/* Helper for the simple case when original dentry is used */
2930	static inline int finish_open_simple(struct file *file, int error)
2931	{
2932	if (error)
2933	return error;
2934
2935	return finish_open(file, dentry: file->f_path.dentry, NULL);
2936	}
2937
2938	/* fs/dcache.c */
2939	extern void __init vfs_caches_init_early(void);
2940	extern void __init vfs_caches_init(void);
2941
2942	extern struct kmem_cache *names_cachep;
2943
2944	#define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2945	#define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2946
2947	extern struct super_block *blockdev_superblock;
2948	static inline bool sb_is_blkdev_sb(struct super_block *sb)
2949	{
2950	return IS_ENABLED(CONFIG_BLOCK) && sb == blockdev_superblock;
2951	}
2952
2953	void emergency_thaw_all(void);
2954	extern int sync_filesystem(struct super_block *);
2955	extern const struct file_operations def_blk_fops;
2956	extern const struct file_operations def_chr_fops;
2957
2958	/* fs/char_dev.c */
2959	#define CHRDEV_MAJOR_MAX 512
2960	/* Marks the bottom of the first segment of free char majors */
2961	#define CHRDEV_MAJOR_DYN_END 234
2962	/* Marks the top and bottom of the second segment of free char majors */
2963	#define CHRDEV_MAJOR_DYN_EXT_START 511
2964	#define CHRDEV_MAJOR_DYN_EXT_END 384
2965
2966	extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2967	extern int register_chrdev_region(dev_t, unsigned, const char *);
2968	extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2969	unsigned int count, const char *name,
2970	const struct file_operations *fops);
2971	extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2972	unsigned int count, const char *name);
2973	extern void unregister_chrdev_region(dev_t, unsigned);
2974	extern void chrdev_show(struct seq_file *,off_t);
2975
2976	static inline int register_chrdev(unsigned int major, const char *name,
2977	const struct file_operations *fops)
2978	{
2979	return __register_chrdev(major, baseminor: 0, count: 256, name, fops);
2980	}
2981
2982	static inline void unregister_chrdev(unsigned int major, const char *name)
2983	{
2984	__unregister_chrdev(major, baseminor: 0, count: 256, name);
2985	}
2986
2987	extern void init_special_inode(struct inode *, umode_t, dev_t);
2988
2989	/* Invalid inode operations -- fs/bad_inode.c */
2990	extern void make_bad_inode(struct inode *);
2991	extern bool is_bad_inode(struct inode *);
2992
2993	extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2994	loff_t lend);
2995	extern int __must_check file_check_and_advance_wb_err(struct file *file);
2996	extern int __must_check file_write_and_wait_range(struct file *file,
2997	loff_t start, loff_t end);
2998	int filemap_fdatawrite_range_kick(struct address_space *mapping, loff_t start,
2999	loff_t end);
3000
3001	static inline int file_write_and_wait(struct file *file)
3002	{
3003	return file_write_and_wait_range(file, start: 0, LLONG_MAX);
3004	}
3005
3006	extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
3007	int datasync);
3008	extern int vfs_fsync(struct file *file, int datasync);
3009
3010	extern int sync_file_range(struct file *file, loff_t offset, loff_t nbytes,
3011	unsigned int flags);
3012
3013	static inline bool iocb_is_dsync(const struct kiocb *iocb)
3014	{
3015	return (iocb->ki_flags & IOCB_DSYNC) ||
3016	IS_SYNC(iocb->ki_filp->f_mapping->host);
3017	}
3018
3019	/*
3020	* Sync the bytes written if this was a synchronous write. Expect ki_pos
3021	* to already be updated for the write, and will return either the amount
3022	* of bytes passed in, or an error if syncing the file failed.
3023	*/
3024	static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
3025	{
3026	if (iocb_is_dsync(iocb)) {
3027	int ret = vfs_fsync_range(file: iocb->ki_filp,
3028	start: iocb->ki_pos - count, end: iocb->ki_pos - 1,
3029	datasync: (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
3030	if (ret)
3031	return ret;
3032	} else if (iocb->ki_flags & IOCB_DONTCACHE) {
3033	struct address_space *mapping = iocb->ki_filp->f_mapping;
3034
3035	filemap_fdatawrite_range_kick(mapping, start: iocb->ki_pos - count,
3036	end: iocb->ki_pos - 1);
3037	}
3038
3039	return count;
3040	}
3041
3042	extern void emergency_sync(void);
3043	extern void emergency_remount(void);
3044
3045	#ifdef CONFIG_BLOCK
3046	extern int bmap(struct inode *inode, sector_t *block);
3047	#else
3048	static inline int bmap(struct inode *inode, sector_t *block)
3049	{
3050	return -EINVAL;
3051	}
3052	#endif
3053
3054	int notify_change(struct mnt_idmap *, struct dentry *,
3055	struct iattr *, struct inode **);
3056	int inode_permission(struct mnt_idmap *, struct inode *, int);
3057	int generic_permission(struct mnt_idmap *, struct inode *, int);
3058	static inline int file_permission(struct file *file, int mask)
3059	{
3060	return inode_permission(file_mnt_idmap(file),
3061	file_inode(f: file), mask);
3062	}
3063	static inline int path_permission(const struct path *path, int mask)
3064	{
3065	return inode_permission(mnt_idmap(mnt: path->mnt),
3066	d_inode(dentry: path->dentry), mask);
3067	}
3068	int __check_sticky(struct mnt_idmap *idmap, struct inode *dir,
3069	struct inode *inode);
3070
3071	static inline bool execute_ok(struct inode *inode)
3072	{
3073	return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
3074	}
3075
3076	static inline bool inode_wrong_type(const struct inode *inode, umode_t mode)
3077	{
3078	return (inode->i_mode ^ mode) & S_IFMT;
3079	}
3080
3081	/**
3082	* file_start_write - get write access to a superblock for regular file io
3083	* @file: the file we want to write to
3084	*
3085	* This is a variant of sb_start_write() which is a noop on non-regualr file.
3086	* Should be matched with a call to file_end_write().
3087	*/
3088	static inline void file_start_write(struct file *file)
3089	{
3090	if (!S_ISREG(file_inode(file)->i_mode))
3091	return;
3092	sb_start_write(sb: file_inode(f: file)->i_sb);
3093	}
3094
3095	static inline bool file_start_write_trylock(struct file *file)
3096	{
3097	if (!S_ISREG(file_inode(file)->i_mode))
3098	return true;
3099	return sb_start_write_trylock(sb: file_inode(f: file)->i_sb);
3100	}
3101
3102	/**
3103	* file_end_write - drop write access to a superblock of a regular file
3104	* @file: the file we wrote to
3105	*
3106	* Should be matched with a call to file_start_write().
3107	*/
3108	static inline void file_end_write(struct file *file)
3109	{
3110	if (!S_ISREG(file_inode(file)->i_mode))
3111	return;
3112	sb_end_write(sb: file_inode(f: file)->i_sb);
3113	}
3114
3115	/**
3116	* kiocb_start_write - get write access to a superblock for async file io
3117	* @iocb: the io context we want to submit the write with
3118	*
3119	* This is a variant of sb_start_write() for async io submission.
3120	* Should be matched with a call to kiocb_end_write().
3121	*/
3122	static inline void kiocb_start_write(struct kiocb *iocb)
3123	{
3124	struct inode *inode = file_inode(f: iocb->ki_filp);
3125
3126	sb_start_write(sb: inode->i_sb);
3127	/*
3128	* Fool lockdep by telling it the lock got released so that it
3129	* doesn't complain about the held lock when we return to userspace.
3130	*/
3131	__sb_writers_release(inode->i_sb, SB_FREEZE_WRITE);
3132	}
3133
3134	/**
3135	* kiocb_end_write - drop write access to a superblock after async file io
3136	* @iocb: the io context we sumbitted the write with
3137	*
3138	* Should be matched with a call to kiocb_start_write().
3139	*/
3140	static inline void kiocb_end_write(struct kiocb *iocb)
3141	{
3142	struct inode *inode = file_inode(f: iocb->ki_filp);
3143
3144	/*
3145	* Tell lockdep we inherited freeze protection from submission thread.
3146	*/
3147	__sb_writers_acquired(inode->i_sb, SB_FREEZE_WRITE);
3148	sb_end_write(sb: inode->i_sb);
3149	}
3150
3151	/*
3152	* This is used for regular files where some users -- especially the
3153	* currently executed binary in a process, previously handled via
3154	* VM_DENYWRITE -- cannot handle concurrent write (and maybe mmap
3155	* read-write shared) accesses.
3156	*
3157	* get_write_access() gets write permission for a file.
3158	* put_write_access() releases this write permission.
3159	* deny_write_access() denies write access to a file.
3160	* allow_write_access() re-enables write access to a file.
3161	*
3162	* The i_writecount field of an inode can have the following values:
3163	* 0: no write access, no denied write access
3164	* < 0: (-i_writecount) users that denied write access to the file.
3165	* > 0: (i_writecount) users that have write access to the file.
3166	*
3167	* Normally we operate on that counter with atomic_{inc,dec} and it's safe
3168	* except for the cases where we don't hold i_writecount yet. Then we need to
3169	* use {get,deny}_write_access() - these functions check the sign and refuse
3170	* to do the change if sign is wrong.
3171	*/
3172	static inline int get_write_access(struct inode *inode)
3173	{
3174	return atomic_inc_unless_negative(v: &inode->i_writecount) ? 0 : -ETXTBSY;
3175	}
3176	static inline int deny_write_access(struct file *file)
3177	{
3178	struct inode *inode = file_inode(f: file);
3179	return atomic_dec_unless_positive(v: &inode->i_writecount) ? 0 : -ETXTBSY;
3180	}
3181	static inline void put_write_access(struct inode * inode)
3182	{
3183	atomic_dec(v: &inode->i_writecount);
3184	}
3185	static inline void allow_write_access(struct file *file)
3186	{
3187	if (file)
3188	atomic_inc(v: &file_inode(f: file)->i_writecount);
3189	}
3190
3191	/*
3192	* Do not prevent write to executable file when watched by pre-content events.
3193	*
3194	* Note that FMODE_FSNOTIFY_HSM mode is set depending on pre-content watches at
3195	* the time of file open and remains constant for entire lifetime of the file,
3196	* so if pre-content watches are added post execution or removed before the end
3197	* of the execution, it will not cause i_writecount reference leak.
3198	*/
3199	static inline int exe_file_deny_write_access(struct file *exe_file)
3200	{
3201	if (unlikely(FMODE_FSNOTIFY_HSM(exe_file->f_mode)))
3202	return 0;
3203	return deny_write_access(file: exe_file);
3204	}
3205	static inline void exe_file_allow_write_access(struct file *exe_file)
3206	{
3207	if (unlikely(!exe_file || FMODE_FSNOTIFY_HSM(exe_file->f_mode)))
3208	return;
3209	allow_write_access(file: exe_file);
3210	}
3211
3212	static inline void file_set_fsnotify_mode(struct file *file, fmode_t mode)
3213	{
3214	file->f_mode &= ~FMODE_FSNOTIFY_MASK;
3215	file->f_mode |= mode;
3216	}
3217
3218	static inline bool inode_is_open_for_write(const struct inode *inode)
3219	{
3220	return atomic_read(v: &inode->i_writecount) > 0;
3221	}
3222
3223	#if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
3224	static inline void i_readcount_dec(struct inode *inode)
3225	{
3226	BUG_ON(atomic_dec_return(&inode->i_readcount) < 0);
3227	}
3228	static inline void i_readcount_inc(struct inode *inode)
3229	{
3230	atomic_inc(v: &inode->i_readcount);
3231	}
3232	#else
3233	static inline void i_readcount_dec(struct inode *inode)
3234	{
3235	return;
3236	}
3237	static inline void i_readcount_inc(struct inode *inode)
3238	{
3239	return;
3240	}
3241	#endif
3242	extern int do_pipe_flags(int *, int);
3243
3244	extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
3245	ssize_t __kernel_read(struct file *file, void *buf, size_t count, loff_t *pos);
3246	extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
3247	extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
3248	extern struct file * open_exec(const char *);
3249
3250	/* fs/dcache.c -- generic fs support functions */
3251	extern bool is_subdir(struct dentry *, struct dentry *);
3252	extern bool path_is_under(const struct path *, const struct path *);
3253
3254	extern char *file_path(struct file *, char *, int);
3255
3256	/**
3257	* is_dot_dotdot - returns true only if @name is "." or ".."
3258	* @name: file name to check
3259	* @len: length of file name, in bytes
3260	*/
3261	static inline bool is_dot_dotdot(const char *name, size_t len)
3262	{
3263	return len && unlikely(name[0] == '.') &&
3264	(len == 1 || (len == 2 && name[1] == '.'));
3265	}
3266
3267	#include <linux/err.h>
3268
3269	/* needed for stackable file system support */
3270	extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
3271
3272	extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
3273
3274	extern int inode_init_always_gfp(struct super_block *, struct inode *, gfp_t);
3275	static inline int inode_init_always(struct super_block *sb, struct inode *inode)
3276	{
3277	return inode_init_always_gfp(sb, inode, GFP_NOFS);
3278	}
3279
3280	extern void inode_init_once(struct inode *);
3281	extern void address_space_init_once(struct address_space *mapping);
3282	extern struct inode * igrab(struct inode *);
3283	extern ino_t iunique(struct super_block *, ino_t);
3284	extern int inode_needs_sync(struct inode *inode);
3285	extern int generic_delete_inode(struct inode *inode);
3286	static inline int generic_drop_inode(struct inode *inode)
3287	{
3288	return !inode->i_nlink || inode_unhashed(inode);
3289	}
3290	extern void d_mark_dontcache(struct inode *inode);
3291
3292	extern struct inode *ilookup5_nowait(struct super_block *sb,
3293	unsigned long hashval, int (*test)(struct inode *, void *),
3294	void *data);
3295	extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
3296	int (*test)(struct inode *, void *), void *data);
3297	extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
3298
3299	extern struct inode *inode_insert5(struct inode *inode, unsigned long hashval,
3300	int (*test)(struct inode *, void *),
3301	int (*set)(struct inode *, void *),
3302	void *data);
3303	struct inode *iget5_locked(struct super_block *, unsigned long,
3304	int (*test)(struct inode *, void *),
3305	int (*set)(struct inode *, void *), void *);
3306	struct inode *iget5_locked_rcu(struct super_block *, unsigned long,
3307	int (*test)(struct inode *, void *),
3308	int (*set)(struct inode *, void *), void *);
3309	extern struct inode * iget_locked(struct super_block *, unsigned long);
3310	extern struct inode *find_inode_nowait(struct super_block *,
3311	unsigned long,
3312	int (*match)(struct inode *,
3313	unsigned long, void *),
3314	void *data);
3315	extern struct inode *find_inode_rcu(struct super_block *, unsigned long,
3316	int (*)(struct inode *, void *), void *);
3317	extern struct inode *find_inode_by_ino_rcu(struct super_block *, unsigned long);
3318	extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
3319	extern int insert_inode_locked(struct inode *);
3320	#ifdef CONFIG_DEBUG_LOCK_ALLOC
3321	extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
3322	#else
3323	static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
3324	#endif
3325	extern void unlock_new_inode(struct inode *);
3326	extern void discard_new_inode(struct inode *);
3327	extern unsigned int get_next_ino(void);
3328	extern void evict_inodes(struct super_block *sb);
3329	void dump_mapping(const struct address_space *);
3330
3331	/*
3332	* Userspace may rely on the inode number being non-zero. For example, glibc
3333	* simply ignores files with zero i_ino in unlink() and other places.
3334	*
3335	* As an additional complication, if userspace was compiled with
3336	* _FILE_OFFSET_BITS=32 on a 64-bit kernel we'll only end up reading out the
3337	* lower 32 bits, so we need to check that those aren't zero explicitly. With
3338	* _FILE_OFFSET_BITS=64, this may cause some harmless false-negatives, but
3339	* better safe than sorry.
3340	*/
3341	static inline bool is_zero_ino(ino_t ino)
3342	{
3343	return (u32)ino == 0;
3344	}
3345
3346	/*
3347	* inode->i_lock must be held
3348	*/
3349	static inline void __iget(struct inode *inode)
3350	{
3351	atomic_inc(v: &inode->i_count);
3352	}
3353
3354	extern void iget_failed(struct inode *);
3355	extern void clear_inode(struct inode *);
3356	extern void __destroy_inode(struct inode *);
3357	struct inode *alloc_inode(struct super_block *sb);
3358	static inline struct inode *new_inode_pseudo(struct super_block *sb)
3359	{
3360	return alloc_inode(sb);
3361	}
3362	extern struct inode *new_inode(struct super_block *sb);
3363	extern void free_inode_nonrcu(struct inode *inode);
3364	extern int setattr_should_drop_suidgid(struct mnt_idmap *, struct inode *);
3365	extern int file_remove_privs_flags(struct file *file, unsigned int flags);
3366	extern int file_remove_privs(struct file *);
3367	int setattr_should_drop_sgid(struct mnt_idmap *idmap,
3368	const struct inode *inode);
3369
3370	/*
3371	* This must be used for allocating filesystems specific inodes to set
3372	* up the inode reclaim context correctly.
3373	*/
3374	#define alloc_inode_sb(_sb, _cache, _gfp) kmem_cache_alloc_lru(_cache, &_sb->s_inode_lru, _gfp)
3375
3376	extern void __insert_inode_hash(struct inode *, unsigned long hashval);
3377	static inline void insert_inode_hash(struct inode *inode)
3378	{
3379	__insert_inode_hash(inode, hashval: inode->i_ino);
3380	}
3381
3382	extern void __remove_inode_hash(struct inode *);
3383	static inline void remove_inode_hash(struct inode *inode)
3384	{
3385	if (!inode_unhashed(inode) && !hlist_fake(h: &inode->i_hash))
3386	__remove_inode_hash(inode);
3387	}
3388
3389	extern void inode_sb_list_add(struct inode *inode);
3390	extern void inode_add_lru(struct inode *inode);
3391
3392	extern int sb_set_blocksize(struct super_block *, int);
3393	extern int sb_min_blocksize(struct super_block *, int);
3394
3395	extern int generic_file_mmap(struct file *, struct vm_area_struct *);
3396	extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
3397	extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
3398	int generic_write_checks_count(struct kiocb *iocb, loff_t *count);
3399	extern int generic_write_check_limits(struct file *file, loff_t pos,
3400	loff_t *count);
3401	extern int generic_file_rw_checks(struct file *file_in, struct file *file_out);
3402	ssize_t filemap_read(struct kiocb *iocb, struct iov_iter *to,
3403	ssize_t already_read);
3404	extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
3405	extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
3406	extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
3407	extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
3408	ssize_t generic_perform_write(struct kiocb *, struct iov_iter *);
3409	ssize_t direct_write_fallback(struct kiocb *iocb, struct iov_iter *iter,
3410	ssize_t direct_written, ssize_t buffered_written);
3411
3412	ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
3413	rwf_t flags);
3414	ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
3415	rwf_t flags);
3416	ssize_t vfs_iocb_iter_read(struct file *file, struct kiocb *iocb,
3417	struct iov_iter *iter);
3418	ssize_t vfs_iocb_iter_write(struct file *file, struct kiocb *iocb,
3419	struct iov_iter *iter);
3420
3421	/* fs/splice.c */
3422	ssize_t filemap_splice_read(struct file *in, loff_t *ppos,
3423	struct pipe_inode_info *pipe,
3424	size_t len, unsigned int flags);
3425	ssize_t copy_splice_read(struct file *in, loff_t *ppos,
3426	struct pipe_inode_info *pipe,
3427	size_t len, unsigned int flags);
3428	extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
3429	struct file *, loff_t *, size_t, unsigned int);
3430
3431
3432	extern void
3433	file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
3434	extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
3435	extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
3436	extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
3437	extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
3438	int whence, loff_t maxsize, loff_t eof);
3439	loff_t generic_llseek_cookie(struct file *file, loff_t offset, int whence,
3440	u64 *cookie);
3441	extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
3442	int whence, loff_t size);
3443	extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
3444	extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
3445	int rw_verify_area(int, struct file *, const loff_t *, size_t);
3446	extern int generic_file_open(struct inode * inode, struct file * filp);
3447	extern int nonseekable_open(struct inode * inode, struct file * filp);
3448	extern int stream_open(struct inode * inode, struct file * filp);
3449
3450	#ifdef CONFIG_BLOCK
3451	typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
3452	loff_t file_offset);
3453
3454	enum {
3455	/* need locking between buffered and direct access */
3456	DIO_LOCKING = 0x01,
3457
3458	/* filesystem does not support filling holes */
3459	DIO_SKIP_HOLES = 0x02,
3460	};
3461
3462	ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
3463	struct block_device *bdev, struct iov_iter *iter,
3464	get_block_t get_block,
3465	dio_iodone_t end_io,
3466	int flags);
3467
3468	static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
3469	struct inode *inode,
3470	struct iov_iter *iter,
3471	get_block_t get_block)
3472	{
3473	return __blockdev_direct_IO(iocb, inode, bdev: inode->i_sb->s_bdev, iter,
3474	get_block, NULL, flags: DIO_LOCKING | DIO_SKIP_HOLES);
3475	}
3476	#endif
3477
3478	bool inode_dio_finished(const struct inode *inode);
3479	void inode_dio_wait(struct inode *inode);
3480	void inode_dio_wait_interruptible(struct inode *inode);
3481
3482	/**
3483	* inode_dio_begin - signal start of a direct I/O requests
3484	* @inode: inode the direct I/O happens on
3485	*
3486	* This is called once we've finished processing a direct I/O request,
3487	* and is used to wake up callers waiting for direct I/O to be quiesced.
3488	*/
3489	static inline void inode_dio_begin(struct inode *inode)
3490	{
3491	atomic_inc(v: &inode->i_dio_count);
3492	}
3493
3494	/**
3495	* inode_dio_end - signal finish of a direct I/O requests
3496	* @inode: inode the direct I/O happens on
3497	*
3498	* This is called once we've finished processing a direct I/O request,
3499	* and is used to wake up callers waiting for direct I/O to be quiesced.
3500	*/
3501	static inline void inode_dio_end(struct inode *inode)
3502	{
3503	if (atomic_dec_and_test(v: &inode->i_dio_count))
3504	wake_up_var(var: &inode->i_dio_count);
3505	}
3506
3507	extern void inode_set_flags(struct inode *inode, unsigned int flags,
3508	unsigned int mask);
3509
3510	extern const struct file_operations generic_ro_fops;
3511
3512	#define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3513
3514	extern int readlink_copy(char __user *, int, const char *, int);
3515	extern int page_readlink(struct dentry *, char __user *, int);
3516	extern const char *page_get_link_raw(struct dentry *, struct inode *,
3517	struct delayed_call *);
3518	extern const char *page_get_link(struct dentry *, struct inode *,
3519	struct delayed_call *);
3520	extern void page_put_link(void *);
3521	extern int page_symlink(struct inode *inode, const char *symname, int len);
3522	extern const struct inode_operations page_symlink_inode_operations;
3523	extern void kfree_link(void *);
3524	void fill_mg_cmtime(struct kstat *stat, u32 request_mask, struct inode *inode);
3525	void generic_fillattr(struct mnt_idmap *, u32, struct inode *, struct kstat *);
3526	void generic_fill_statx_attr(struct inode *inode, struct kstat *stat);
3527	void generic_fill_statx_atomic_writes(struct kstat *stat,
3528	unsigned int unit_min,
3529	unsigned int unit_max,
3530	unsigned int unit_max_opt);
3531	extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
3532	extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
3533	void __inode_add_bytes(struct inode *inode, loff_t bytes);
3534	void inode_add_bytes(struct inode *inode, loff_t bytes);
3535	void __inode_sub_bytes(struct inode *inode, loff_t bytes);
3536	void inode_sub_bytes(struct inode *inode, loff_t bytes);
3537	static inline loff_t __inode_get_bytes(struct inode *inode)
3538	{
3539	return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes;
3540	}
3541	loff_t inode_get_bytes(struct inode *inode);
3542	void inode_set_bytes(struct inode *inode, loff_t bytes);
3543	const char *simple_get_link(struct dentry *, struct inode *,
3544	struct delayed_call *);
3545	extern const struct inode_operations simple_symlink_inode_operations;
3546
3547	extern int iterate_dir(struct file *, struct dir_context *);
3548
3549	int vfs_fstatat(int dfd, const char __user *filename, struct kstat *stat,
3550	int flags);
3551	int vfs_fstat(int fd, struct kstat *stat);
3552
3553	static inline int vfs_stat(const char __user *filename, struct kstat *stat)
3554	{
3555	return vfs_fstatat(AT_FDCWD, filename, stat, flags: 0);
3556	}
3557	static inline int vfs_lstat(const char __user *name, struct kstat *stat)
3558	{
3559	return vfs_fstatat(AT_FDCWD, filename: name, stat, AT_SYMLINK_NOFOLLOW);
3560	}
3561
3562	extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
3563	extern int vfs_readlink(struct dentry *, char __user *, int);
3564
3565	extern struct file_system_type *get_filesystem(struct file_system_type *fs);
3566	extern void put_filesystem(struct file_system_type *fs);
3567	extern struct file_system_type *get_fs_type(const char *name);
3568	extern void drop_super(struct super_block *sb);
3569	extern void drop_super_exclusive(struct super_block *sb);
3570	extern void iterate_supers(void (*f)(struct super_block *, void *), void *arg);
3571	extern void iterate_supers_type(struct file_system_type *,
3572	void (*)(struct super_block *, void *), void *);
3573	void filesystems_freeze(void);
3574	void filesystems_thaw(void);
3575
3576	extern int dcache_dir_open(struct inode *, struct file *);
3577	extern int dcache_dir_close(struct inode *, struct file *);
3578	extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
3579	extern int dcache_readdir(struct file *, struct dir_context *);
3580	extern int simple_setattr(struct mnt_idmap *, struct dentry *,
3581	struct iattr *);
3582	extern int simple_getattr(struct mnt_idmap *, const struct path *,
3583	struct kstat *, u32, unsigned int);
3584	extern int simple_statfs(struct dentry *, struct kstatfs *);
3585	extern int simple_open(struct inode *inode, struct file *file);
3586	extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3587	extern int simple_unlink(struct inode *, struct dentry *);
3588	extern int simple_rmdir(struct inode *, struct dentry *);
3589	void simple_rename_timestamp(struct inode *old_dir, struct dentry *old_dentry,
3590	struct inode *new_dir, struct dentry *new_dentry);
3591	extern int simple_rename_exchange(struct inode *old_dir, struct dentry *old_dentry,
3592	struct inode *new_dir, struct dentry *new_dentry);
3593	extern int simple_rename(struct mnt_idmap *, struct inode *,
3594	struct dentry *, struct inode *, struct dentry *,
3595	unsigned int);
3596	extern void simple_recursive_removal(struct dentry *,
3597	void (*callback)(struct dentry *));
3598	extern int noop_fsync(struct file *, loff_t, loff_t, int);
3599	extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter);
3600	extern int simple_empty(struct dentry *);
3601	extern int simple_write_begin(struct file *file, struct address_space *mapping,
3602	loff_t pos, unsigned len,
3603	struct folio **foliop, void **fsdata);
3604	extern const struct address_space_operations ram_aops;
3605	extern int always_delete_dentry(const struct dentry *);
3606	extern struct inode *alloc_anon_inode(struct super_block *);
3607	extern int simple_nosetlease(struct file *, int, struct file_lease **, void **);
3608	extern const struct dentry_operations simple_dentry_operations;
3609
3610	extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3611	extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3612	extern const struct file_operations simple_dir_operations;
3613	extern const struct inode_operations simple_dir_inode_operations;
3614	extern void make_empty_dir_inode(struct inode *inode);
3615	extern bool is_empty_dir_inode(struct inode *inode);
3616	struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
3617	struct dentry *d_alloc_name(struct dentry *, const char *);
3618	extern int simple_fill_super(struct super_block *, unsigned long,
3619	const struct tree_descr *);
3620	extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3621	extern void simple_release_fs(struct vfsmount **mount, int *count);
3622
3623	extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3624	loff_t *ppos, const void *from, size_t available);
3625	extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3626	const void __user *from, size_t count);
3627
3628	struct offset_ctx {
3629	struct maple_tree mt;
3630	unsigned long next_offset;
3631	};
3632
3633	void simple_offset_init(struct offset_ctx *octx);
3634	int simple_offset_add(struct offset_ctx *octx, struct dentry *dentry);
3635	void simple_offset_remove(struct offset_ctx *octx, struct dentry *dentry);
3636	int simple_offset_rename(struct inode *old_dir, struct dentry *old_dentry,
3637	struct inode *new_dir, struct dentry *new_dentry);
3638	int simple_offset_rename_exchange(struct inode *old_dir,
3639	struct dentry *old_dentry,
3640	struct inode *new_dir,
3641	struct dentry *new_dentry);
3642	void simple_offset_destroy(struct offset_ctx *octx);
3643
3644	extern const struct file_operations simple_offset_dir_operations;
3645
3646	extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3647	extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3648
3649	extern int generic_check_addressable(unsigned, u64);
3650
3651	extern void generic_set_sb_d_ops(struct super_block *sb);
3652	extern int generic_ci_match(const struct inode *parent,
3653	const struct qstr *name,
3654	const struct qstr *folded_name,
3655	const u8 *de_name, u32 de_name_len);
3656
3657	#if IS_ENABLED(CONFIG_UNICODE)
3658	int generic_ci_d_hash(const struct dentry *dentry, struct qstr *str);
3659	int generic_ci_d_compare(const struct dentry *dentry, unsigned int len,
3660	const char *str, const struct qstr *name);
3661
3662	/**
3663	* generic_ci_validate_strict_name - Check if a given name is suitable
3664	* for a directory
3665	*
3666	* This functions checks if the proposed filename is valid for the
3667	* parent directory. That means that only valid UTF-8 filenames will be
3668	* accepted for casefold directories from filesystems created with the
3669	* strict encoding flag. That also means that any name will be
3670	* accepted for directories that doesn't have casefold enabled, or
3671	* aren't being strict with the encoding.
3672	*
3673	* @dir: inode of the directory where the new file will be created
3674	* @name: name of the new file
3675	*
3676	* Return:
3677	* * True: if the filename is suitable for this directory. It can be
3678	* true if a given name is not suitable for a strict encoding
3679	* directory, but the directory being used isn't strict
3680	* * False if the filename isn't suitable for this directory. This only
3681	* happens when a directory is casefolded and the filesystem is strict
3682	* about its encoding.
3683	*/
3684	static inline bool generic_ci_validate_strict_name(struct inode *dir, struct qstr *name)
3685	{
3686	if (!IS_CASEFOLDED(dir) || !sb_has_strict_encoding(dir->i_sb))
3687	return true;
3688
3689	/*
3690	* A casefold dir must have a encoding set, unless the filesystem
3691	* is corrupted
3692	*/
3693	if (WARN_ON_ONCE(!dir->i_sb->s_encoding))
3694	return true;
3695
3696	return !utf8_validate(um: dir->i_sb->s_encoding, str: name);
3697	}
3698	#else
3699	static inline bool generic_ci_validate_strict_name(struct inode *dir, struct qstr *name)
3700	{
3701	return true;
3702	}
3703	#endif
3704
3705	static inline bool sb_has_encoding(const struct super_block *sb)
3706	{
3707	#if IS_ENABLED(CONFIG_UNICODE)
3708	return !!sb->s_encoding;
3709	#else
3710	return false;
3711	#endif
3712	}
3713
3714	int may_setattr(struct mnt_idmap *idmap, struct inode *inode,
3715	unsigned int ia_valid);
3716	int setattr_prepare(struct mnt_idmap *, struct dentry *, struct iattr *);
3717	extern int inode_newsize_ok(const struct inode *, loff_t offset);
3718	void setattr_copy(struct mnt_idmap *, struct inode *inode,
3719	const struct iattr *attr);
3720
3721	extern int file_update_time(struct file *file);
3722
3723	static inline bool vma_is_dax(const struct vm_area_struct *vma)
3724	{
3725	return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3726	}
3727
3728	static inline bool vma_is_fsdax(struct vm_area_struct *vma)
3729	{
3730	struct inode *inode;
3731
3732	if (!IS_ENABLED(CONFIG_FS_DAX) || !vma->vm_file)
3733	return false;
3734	if (!vma_is_dax(vma))
3735	return false;
3736	inode = file_inode(f: vma->vm_file);
3737	if (S_ISCHR(inode->i_mode))
3738	return false; /* device-dax */
3739	return true;
3740	}
3741
3742	static inline int iocb_flags(struct file *file)
3743	{
3744	int res = 0;
3745	if (file->f_flags & O_APPEND)
3746	res |= IOCB_APPEND;
3747	if (file->f_flags & O_DIRECT)
3748	res |= IOCB_DIRECT;
3749	if (file->f_flags & O_DSYNC)
3750	res |= IOCB_DSYNC;
3751	if (file->f_flags & __O_SYNC)
3752	res |= IOCB_SYNC;
3753	return res;
3754	}
3755
3756	static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags,
3757	int rw_type)
3758	{
3759	int kiocb_flags = 0;
3760
3761	/* make sure there's no overlap between RWF and private IOCB flags */
3762	BUILD_BUG_ON((__force int) RWF_SUPPORTED & IOCB_EVENTFD);
3763
3764	if (!flags)
3765	return 0;
3766	if (unlikely(flags & ~RWF_SUPPORTED))
3767	return -EOPNOTSUPP;
3768	if (unlikely((flags & RWF_APPEND) && (flags & RWF_NOAPPEND)))
3769	return -EINVAL;
3770
3771	if (flags & RWF_NOWAIT) {
3772	if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
3773	return -EOPNOTSUPP;
3774	}
3775	if (flags & RWF_ATOMIC) {
3776	if (rw_type != WRITE)
3777	return -EOPNOTSUPP;
3778	if (!(ki->ki_filp->f_mode & FMODE_CAN_ATOMIC_WRITE))
3779	return -EOPNOTSUPP;
3780	}
3781	if (flags & RWF_DONTCACHE) {
3782	/* file system must support it */
3783	if (!(ki->ki_filp->f_op->fop_flags & FOP_DONTCACHE))
3784	return -EOPNOTSUPP;
3785	/* DAX mappings not supported */
3786	if (IS_DAX(ki->ki_filp->f_mapping->host))
3787	return -EOPNOTSUPP;
3788	}
3789	kiocb_flags |= (__force int) (flags & RWF_SUPPORTED);
3790	if (flags & RWF_SYNC)
3791	kiocb_flags |= IOCB_DSYNC;
3792
3793	if ((flags & RWF_NOAPPEND) && (ki->ki_flags & IOCB_APPEND)) {
3794	if (IS_APPEND(file_inode(ki->ki_filp)))
3795	return -EPERM;
3796	ki->ki_flags &= ~IOCB_APPEND;
3797	}
3798
3799	ki->ki_flags |= kiocb_flags;
3800	return 0;
3801	}
3802
3803	/* Transaction based IO helpers */
3804
3805	/*
3806	* An argresp is stored in an allocated page and holds the
3807	* size of the argument or response, along with its content
3808	*/
3809	struct simple_transaction_argresp {
3810	ssize_t size;
3811	char data[];
3812	};
3813
3814	#define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3815
3816	char *simple_transaction_get(struct file *file, const char __user *buf,
3817	size_t size);
3818	ssize_t simple_transaction_read(struct file *file, char __user *buf,
3819	size_t size, loff_t *pos);
3820	int simple_transaction_release(struct inode *inode, struct file *file);
3821
3822	void simple_transaction_set(struct file *file, size_t n);
3823
3824	/*
3825	* simple attribute files
3826	*
3827	* These attributes behave similar to those in sysfs:
3828	*
3829	* Writing to an attribute immediately sets a value, an open file can be
3830	* written to multiple times.
3831	*
3832	* Reading from an attribute creates a buffer from the value that might get
3833	* read with multiple read calls. When the attribute has been read
3834	* completely, no further read calls are possible until the file is opened
3835	* again.
3836	*
3837	* All attributes contain a text representation of a numeric value
3838	* that are accessed with the get() and set() functions.
3839	*/
3840	#define DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, __is_signed) \
3841	static int __fops ## _open(struct inode *inode, struct file *file) \
3842	{ \
3843	__simple_attr_check_format(__fmt, 0ull); \
3844	return simple_attr_open(inode, file, __get, __set, __fmt); \
3845	} \
3846	static const struct file_operations __fops = { \
3847	.owner = THIS_MODULE, \
3848	.open = __fops ## _open, \
3849	.release = simple_attr_release, \
3850	.read = simple_attr_read, \
3851	.write = (__is_signed) ? simple_attr_write_signed : simple_attr_write, \
3852	.llseek = generic_file_llseek, \
3853	}
3854
3855	#define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3856	DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, false)
3857
3858	#define DEFINE_SIMPLE_ATTRIBUTE_SIGNED(__fops, __get, __set, __fmt) \
3859	DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, true)
3860
3861	static inline __printf(1, 2)
3862	void __simple_attr_check_format(const char *fmt, ...)
3863	{
3864	/* don't do anything, just let the compiler check the arguments; */
3865	}
3866
3867	int simple_attr_open(struct inode *inode, struct file *file,
3868	int (*get)(void *, u64 *), int (*set)(void *, u64),
3869	const char *fmt);
3870	int simple_attr_release(struct inode *inode, struct file *file);
3871	ssize_t simple_attr_read(struct file *file, char __user *buf,
3872	size_t len, loff_t *ppos);
3873	ssize_t simple_attr_write(struct file *file, const char __user *buf,
3874	size_t len, loff_t *ppos);
3875	ssize_t simple_attr_write_signed(struct file *file, const char __user *buf,
3876	size_t len, loff_t *ppos);
3877
3878	struct ctl_table;
3879	int __init list_bdev_fs_names(char *buf, size_t size);
3880
3881	#define __FMODE_EXEC ((__force int) FMODE_EXEC)
3882
3883	#define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3884	#define OPEN_FMODE(flag) ((__force fmode_t)((flag + 1) & O_ACCMODE))
3885
3886	static inline bool is_sxid(umode_t mode)
3887	{
3888	return mode & (S_ISUID | S_ISGID);
3889	}
3890
3891	static inline int check_sticky(struct mnt_idmap *idmap,
3892	struct inode *dir, struct inode *inode)
3893	{
3894	if (!(dir->i_mode & S_ISVTX))
3895	return 0;
3896
3897	return __check_sticky(idmap, dir, inode);
3898	}
3899
3900	static inline void inode_has_no_xattr(struct inode *inode)
3901	{
3902	if (!is_sxid(mode: inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
3903	inode->i_flags |= S_NOSEC;
3904	}
3905
3906	static inline bool is_root_inode(struct inode *inode)
3907	{
3908	return inode == inode->i_sb->s_root->d_inode;
3909	}
3910
3911	static inline bool dir_emit(struct dir_context *ctx,
3912	const char *name, int namelen,
3913	u64 ino, unsigned type)
3914	{
3915	return ctx->actor(ctx, name, namelen, ctx->pos, ino, type);
3916	}
3917	static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3918	{
3919	return ctx->actor(ctx, ".", 1, ctx->pos,
3920	file->f_path.dentry->d_inode->i_ino, DT_DIR);
3921	}
3922	static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3923	{
3924	return ctx->actor(ctx, "..", 2, ctx->pos,
3925	d_parent_ino(dentry: file->f_path.dentry), DT_DIR);
3926	}
3927	static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3928	{
3929	if (ctx->pos == 0) {
3930	if (!dir_emit_dot(file, ctx))
3931	return false;
3932	ctx->pos = 1;
3933	}
3934	if (ctx->pos == 1) {
3935	if (!dir_emit_dotdot(file, ctx))
3936	return false;
3937	ctx->pos = 2;
3938	}
3939	return true;
3940	}
3941	static inline bool dir_relax(struct inode *inode)
3942	{
3943	inode_unlock(inode);
3944	inode_lock(inode);
3945	return !IS_DEADDIR(inode);
3946	}
3947
3948	static inline bool dir_relax_shared(struct inode *inode)
3949	{
3950	inode_unlock_shared(inode);
3951	inode_lock_shared(inode);
3952	return !IS_DEADDIR(inode);
3953	}
3954
3955	extern bool path_noexec(const struct path *path);
3956	extern void inode_nohighmem(struct inode *inode);
3957
3958	/* mm/fadvise.c */
3959	extern int vfs_fadvise(struct file *file, loff_t offset, loff_t len,
3960	int advice);
3961	extern int generic_fadvise(struct file *file, loff_t offset, loff_t len,
3962	int advice);
3963
3964	static inline bool vfs_empty_path(int dfd, const char __user *path)
3965	{
3966	char c;
3967
3968	if (dfd < 0)
3969	return false;
3970
3971	/* We now allow NULL to be used for empty path. */
3972	if (!path)
3973	return true;
3974
3975	if (unlikely(get_user(c, path)))
3976	return false;
3977
3978	return !c;
3979	}
3980
3981	int generic_atomic_write_valid(struct kiocb *iocb, struct iov_iter *iter);
3982
3983	#endif /* _LINUX_FS_H */
3984