• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 /* SPDX-License-Identifier: GPL-2.0 */
2 #ifndef _LINUX_FS_H
3 #define _LINUX_FS_H
4 
5 #include <linux/linkage.h>
6 #include <linux/wait_bit.h>
7 #include <linux/kdev_t.h>
8 #include <linux/dcache.h>
9 #include <linux/path.h>
10 #include <linux/stat.h>
11 #include <linux/cache.h>
12 #include <linux/list.h>
13 #include <linux/list_lru.h>
14 #include <linux/llist.h>
15 #include <linux/radix-tree.h>
16 #include <linux/xarray.h>
17 #include <linux/rbtree.h>
18 #include <linux/init.h>
19 #include <linux/pid.h>
20 #include <linux/bug.h>
21 #include <linux/mutex.h>
22 #include <linux/rwsem.h>
23 #include <linux/mm_types.h>
24 #include <linux/capability.h>
25 #include <linux/semaphore.h>
26 #include <linux/fcntl.h>
27 #include <linux/rculist_bl.h>
28 #include <linux/atomic.h>
29 #include <linux/shrinker.h>
30 #include <linux/migrate_mode.h>
31 #include <linux/uidgid.h>
32 #include <linux/lockdep.h>
33 #include <linux/percpu-rwsem.h>
34 #include <linux/workqueue.h>
35 #include <linux/delayed_call.h>
36 #include <linux/uuid.h>
37 #include <linux/errseq.h>
38 #include <linux/ioprio.h>
39 #include <linux/fs_types.h>
40 #include <linux/build_bug.h>
41 #include <linux/stddef.h>
42 #include <linux/mount.h>
43 #include <linux/cred.h>
44 #include <linux/mnt_idmapping.h>
45 #include <linux/android_kabi.h>
46 
47 #include <asm/byteorder.h>
48 #include <uapi/linux/fs.h>
49 
50 struct backing_dev_info;
51 struct bdi_writeback;
52 struct bio;
53 struct export_operations;
54 struct fiemap_extent_info;
55 struct hd_geometry;
56 struct iovec;
57 struct kiocb;
58 struct kobject;
59 struct pipe_inode_info;
60 struct poll_table_struct;
61 struct kstatfs;
62 struct vm_area_struct;
63 struct vfsmount;
64 struct cred;
65 struct swap_info_struct;
66 struct seq_file;
67 struct workqueue_struct;
68 struct iov_iter;
69 struct fscrypt_info;
70 struct fscrypt_operations;
71 struct fsverity_info;
72 struct fsverity_operations;
73 struct fs_context;
74 struct fs_parameter_spec;
75 struct fileattr;
76 
77 extern void __init inode_init(void);
78 extern void __init inode_init_early(void);
79 extern void __init files_init(void);
80 extern void __init files_maxfiles_init(void);
81 
82 extern struct files_stat_struct files_stat;
83 extern unsigned long get_max_files(void);
84 extern unsigned int sysctl_nr_open;
85 extern struct inodes_stat_t inodes_stat;
86 extern int leases_enable, lease_break_time;
87 extern int sysctl_protected_symlinks;
88 extern int sysctl_protected_hardlinks;
89 extern int sysctl_protected_fifos;
90 extern int sysctl_protected_regular;
91 
92 typedef __kernel_rwf_t rwf_t;
93 
94 struct buffer_head;
95 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
96 			struct buffer_head *bh_result, int create);
97 typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
98 			ssize_t bytes, void *private);
99 
100 #define MAY_EXEC		0x00000001
101 #define MAY_WRITE		0x00000002
102 #define MAY_READ		0x00000004
103 #define MAY_APPEND		0x00000008
104 #define MAY_ACCESS		0x00000010
105 #define MAY_OPEN		0x00000020
106 #define MAY_CHDIR		0x00000040
107 /* called from RCU mode, don't block */
108 #define MAY_NOT_BLOCK		0x00000080
109 
110 /*
111  * flags in file.f_mode.  Note that FMODE_READ and FMODE_WRITE must correspond
112  * to O_WRONLY and O_RDWR via the strange trick in do_dentry_open()
113  */
114 
115 /* file is open for reading */
116 #define FMODE_READ		((__force fmode_t)0x1)
117 /* file is open for writing */
118 #define FMODE_WRITE		((__force fmode_t)0x2)
119 /* file is seekable */
120 #define FMODE_LSEEK		((__force fmode_t)0x4)
121 /* file can be accessed using pread */
122 #define FMODE_PREAD		((__force fmode_t)0x8)
123 /* file can be accessed using pwrite */
124 #define FMODE_PWRITE		((__force fmode_t)0x10)
125 /* File is opened for execution with sys_execve / sys_uselib */
126 #define FMODE_EXEC		((__force fmode_t)0x20)
127 /* File is opened with O_NDELAY (only set for block devices) */
128 #define FMODE_NDELAY		((__force fmode_t)0x40)
129 /* File is opened with O_EXCL (only set for block devices) */
130 #define FMODE_EXCL		((__force fmode_t)0x80)
131 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
132    (specialy hack for floppy.c) */
133 #define FMODE_WRITE_IOCTL	((__force fmode_t)0x100)
134 /* 32bit hashes as llseek() offset (for directories) */
135 #define FMODE_32BITHASH         ((__force fmode_t)0x200)
136 /* 64bit hashes as llseek() offset (for directories) */
137 #define FMODE_64BITHASH         ((__force fmode_t)0x400)
138 
139 /*
140  * Don't update ctime and mtime.
141  *
142  * Currently a special hack for the XFS open_by_handle ioctl, but we'll
143  * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
144  */
145 #define FMODE_NOCMTIME		((__force fmode_t)0x800)
146 
147 /* Expect random access pattern */
148 #define FMODE_RANDOM		((__force fmode_t)0x1000)
149 
150 /* File is huge (eg. /dev/mem): treat loff_t as unsigned */
151 #define FMODE_UNSIGNED_OFFSET	((__force fmode_t)0x2000)
152 
153 /* File is opened with O_PATH; almost nothing can be done with it */
154 #define FMODE_PATH		((__force fmode_t)0x4000)
155 
156 /* File needs atomic accesses to f_pos */
157 #define FMODE_ATOMIC_POS	((__force fmode_t)0x8000)
158 /* Write access to underlying fs */
159 #define FMODE_WRITER		((__force fmode_t)0x10000)
160 /* Has read method(s) */
161 #define FMODE_CAN_READ          ((__force fmode_t)0x20000)
162 /* Has write method(s) */
163 #define FMODE_CAN_WRITE         ((__force fmode_t)0x40000)
164 
165 #define FMODE_OPENED		((__force fmode_t)0x80000)
166 #define FMODE_CREATED		((__force fmode_t)0x100000)
167 
168 /* File is stream-like */
169 #define FMODE_STREAM		((__force fmode_t)0x200000)
170 
171 #define	FMODE_NOREUSE		((__force fmode_t)0x800000)
172 
173 /* File was opened by fanotify and shouldn't generate fanotify events */
174 #define FMODE_NONOTIFY		((__force fmode_t)0x4000000)
175 
176 /* File is capable of returning -EAGAIN if I/O will block */
177 #define FMODE_NOWAIT		((__force fmode_t)0x8000000)
178 
179 /* File represents mount that needs unmounting */
180 #define FMODE_NEED_UNMOUNT	((__force fmode_t)0x10000000)
181 
182 /* File does not contribute to nr_files count */
183 #define FMODE_NOACCOUNT		((__force fmode_t)0x20000000)
184 
185 /* File supports async buffered reads */
186 #define FMODE_BUF_RASYNC	((__force fmode_t)0x40000000)
187 
188 /*
189  * Attribute flags.  These should be or-ed together to figure out what
190  * has been changed!
191  */
192 #define ATTR_MODE	(1 << 0)
193 #define ATTR_UID	(1 << 1)
194 #define ATTR_GID	(1 << 2)
195 #define ATTR_SIZE	(1 << 3)
196 #define ATTR_ATIME	(1 << 4)
197 #define ATTR_MTIME	(1 << 5)
198 #define ATTR_CTIME	(1 << 6)
199 #define ATTR_ATIME_SET	(1 << 7)
200 #define ATTR_MTIME_SET	(1 << 8)
201 #define ATTR_FORCE	(1 << 9) /* Not a change, but a change it */
202 #define ATTR_KILL_SUID	(1 << 11)
203 #define ATTR_KILL_SGID	(1 << 12)
204 #define ATTR_FILE	(1 << 13)
205 #define ATTR_KILL_PRIV	(1 << 14)
206 #define ATTR_OPEN	(1 << 15) /* Truncating from open(O_TRUNC) */
207 #define ATTR_TIMES_SET	(1 << 16)
208 #define ATTR_TOUCH	(1 << 17)
209 
210 /*
211  * Whiteout is represented by a char device.  The following constants define the
212  * mode and device number to use.
213  */
214 #define WHITEOUT_MODE 0
215 #define WHITEOUT_DEV 0
216 
217 /*
218  * This is the Inode Attributes structure, used for notify_change().  It
219  * uses the above definitions as flags, to know which values have changed.
220  * Also, in this manner, a Filesystem can look at only the values it cares
221  * about.  Basically, these are the attributes that the VFS layer can
222  * request to change from the FS layer.
223  *
224  * Derek Atkins <warlord@MIT.EDU> 94-10-20
225  */
226 struct iattr {
227 	unsigned int	ia_valid;
228 	umode_t		ia_mode;
229 	kuid_t		ia_uid;
230 	kgid_t		ia_gid;
231 	loff_t		ia_size;
232 	struct timespec64 ia_atime;
233 	struct timespec64 ia_mtime;
234 	struct timespec64 ia_ctime;
235 
236 	/*
237 	 * Not an attribute, but an auxiliary info for filesystems wanting to
238 	 * implement an ftruncate() like method.  NOTE: filesystem should
239 	 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
240 	 */
241 	struct file	*ia_file;
242 };
243 
244 /*
245  * Includes for diskquotas.
246  */
247 #include <linux/quota.h>
248 
249 /*
250  * Maximum number of layers of fs stack.  Needs to be limited to
251  * prevent kernel stack overflow
252  */
253 #define FILESYSTEM_MAX_STACK_DEPTH 2
254 
255 /**
256  * enum positive_aop_returns - aop return codes with specific semantics
257  *
258  * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
259  * 			    completed, that the page is still locked, and
260  * 			    should be considered active.  The VM uses this hint
261  * 			    to return the page to the active list -- it won't
262  * 			    be a candidate for writeback again in the near
263  * 			    future.  Other callers must be careful to unlock
264  * 			    the page if they get this return.  Returned by
265  * 			    writepage();
266  *
267  * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
268  *  			unlocked it and the page might have been truncated.
269  *  			The caller should back up to acquiring a new page and
270  *  			trying again.  The aop will be taking reasonable
271  *  			precautions not to livelock.  If the caller held a page
272  *  			reference, it should drop it before retrying.  Returned
273  *  			by readpage().
274  *
275  * address_space_operation functions return these large constants to indicate
276  * special semantics to the caller.  These are much larger than the bytes in a
277  * page to allow for functions that return the number of bytes operated on in a
278  * given page.
279  */
280 
281 enum positive_aop_returns {
282 	AOP_WRITEPAGE_ACTIVATE	= 0x80000,
283 	AOP_TRUNCATED_PAGE	= 0x80001,
284 };
285 
286 #define AOP_FLAG_CONT_EXPAND		0x0001 /* called from cont_expand */
287 #define AOP_FLAG_NOFS			0x0002 /* used by filesystem to direct
288 						* helper code (eg buffer layer)
289 						* to clear GFP_FS from alloc */
290 
291 /*
292  * oh the beauties of C type declarations.
293  */
294 struct page;
295 struct address_space;
296 struct writeback_control;
297 struct readahead_control;
298 
299 /*
300  * Write life time hint values.
301  * Stored in struct inode as u8.
302  */
303 enum rw_hint {
304 	WRITE_LIFE_NOT_SET	= 0,
305 	WRITE_LIFE_NONE		= RWH_WRITE_LIFE_NONE,
306 	WRITE_LIFE_SHORT	= RWH_WRITE_LIFE_SHORT,
307 	WRITE_LIFE_MEDIUM	= RWH_WRITE_LIFE_MEDIUM,
308 	WRITE_LIFE_LONG		= RWH_WRITE_LIFE_LONG,
309 	WRITE_LIFE_EXTREME	= RWH_WRITE_LIFE_EXTREME,
310 };
311 
312 /* Match RWF_* bits to IOCB bits */
313 #define IOCB_HIPRI		(__force int) RWF_HIPRI
314 #define IOCB_DSYNC		(__force int) RWF_DSYNC
315 #define IOCB_SYNC		(__force int) RWF_SYNC
316 #define IOCB_NOWAIT		(__force int) RWF_NOWAIT
317 #define IOCB_APPEND		(__force int) RWF_APPEND
318 
319 /* non-RWF related bits - start at 16 */
320 #define IOCB_EVENTFD		(1 << 16)
321 #define IOCB_DIRECT		(1 << 17)
322 #define IOCB_WRITE		(1 << 18)
323 /* iocb->ki_waitq is valid */
324 #define IOCB_WAITQ		(1 << 19)
325 #define IOCB_NOIO		(1 << 20)
326 /* can use bio alloc cache */
327 #define IOCB_ALLOC_CACHE	(1 << 21)
328 /* kiocb is a read or write operation submitted by fs/aio.c. */
329 #define IOCB_AIO_RW		(1 << 23)
330 
331 struct kiocb {
332 	struct file		*ki_filp;
333 
334 	/* The 'ki_filp' pointer is shared in a union for aio */
335 	randomized_struct_fields_start
336 
337 	loff_t			ki_pos;
338 	void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
339 	void			*private;
340 	int			ki_flags;
341 	u16			ki_hint;
342 	u16			ki_ioprio; /* See linux/ioprio.h */
343 	union {
344 		unsigned int		ki_cookie; /* for ->iopoll */
345 		struct wait_page_queue	*ki_waitq; /* for async buffered IO */
346 	};
347 
348 	randomized_struct_fields_end
349 };
350 
is_sync_kiocb(struct kiocb * kiocb)351 static inline bool is_sync_kiocb(struct kiocb *kiocb)
352 {
353 	return kiocb->ki_complete == NULL;
354 }
355 
356 /*
357  * "descriptor" for what we're up to with a read.
358  * This allows us to use the same read code yet
359  * have multiple different users of the data that
360  * we read from a file.
361  *
362  * The simplest case just copies the data to user
363  * mode.
364  */
365 typedef struct {
366 	size_t written;
367 	size_t count;
368 	union {
369 		char __user *buf;
370 		void *data;
371 	} arg;
372 	int error;
373 } read_descriptor_t;
374 
375 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
376 		unsigned long, unsigned long);
377 
378 struct address_space_operations {
379 	int (*writepage)(struct page *page, struct writeback_control *wbc);
380 	int (*readpage)(struct file *, struct page *);
381 
382 	/* Write back some dirty pages from this mapping. */
383 	int (*writepages)(struct address_space *, struct writeback_control *);
384 
385 	/* Set a page dirty.  Return true if this dirtied it */
386 	int (*set_page_dirty)(struct page *page);
387 
388 	/*
389 	 * Reads in the requested pages. Unlike ->readpage(), this is
390 	 * PURELY used for read-ahead!.
391 	 */
392 	int (*readpages)(struct file *filp, struct address_space *mapping,
393 			struct list_head *pages, unsigned nr_pages);
394 	void (*readahead)(struct readahead_control *);
395 
396 	int (*write_begin)(struct file *, struct address_space *mapping,
397 				loff_t pos, unsigned len, unsigned flags,
398 				struct page **pagep, void **fsdata);
399 	int (*write_end)(struct file *, struct address_space *mapping,
400 				loff_t pos, unsigned len, unsigned copied,
401 				struct page *page, void *fsdata);
402 
403 	/* Unfortunately this kludge is needed for FIBMAP. Don't use it */
404 	sector_t (*bmap)(struct address_space *, sector_t);
405 	void (*invalidatepage) (struct page *, unsigned int, unsigned int);
406 	int (*releasepage) (struct page *, gfp_t);
407 	void (*freepage)(struct page *);
408 	ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
409 	/*
410 	 * migrate the contents of a page to the specified target. If
411 	 * migrate_mode is MIGRATE_ASYNC, it must not block.
412 	 */
413 	int (*migratepage) (struct address_space *,
414 			struct page *, struct page *, enum migrate_mode);
415 	bool (*isolate_page)(struct page *, isolate_mode_t);
416 	void (*putback_page)(struct page *);
417 	int (*launder_page) (struct page *);
418 	int (*is_partially_uptodate) (struct page *, unsigned long,
419 					unsigned long);
420 	void (*is_dirty_writeback) (struct page *, bool *, bool *);
421 	int (*error_remove_page)(struct address_space *, struct page *);
422 
423 	/* swapfile support */
424 	int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
425 				sector_t *span);
426 	void (*swap_deactivate)(struct file *file);
427 
428 	ANDROID_KABI_RESERVE(1);
429 	ANDROID_KABI_RESERVE(2);
430 	ANDROID_KABI_RESERVE(3);
431 	ANDROID_KABI_RESERVE(4);
432 };
433 
434 extern const struct address_space_operations empty_aops;
435 
436 /*
437  * pagecache_write_begin/pagecache_write_end must be used by general code
438  * to write into the pagecache.
439  */
440 int pagecache_write_begin(struct file *, struct address_space *mapping,
441 				loff_t pos, unsigned len, unsigned flags,
442 				struct page **pagep, void **fsdata);
443 
444 int pagecache_write_end(struct file *, struct address_space *mapping,
445 				loff_t pos, unsigned len, unsigned copied,
446 				struct page *page, void *fsdata);
447 
448 /**
449  * struct address_space - Contents of a cacheable, mappable object.
450  * @host: Owner, either the inode or the block_device.
451  * @i_pages: Cached pages.
452  * @invalidate_lock: Guards coherency between page cache contents and
453  *   file offset->disk block mappings in the filesystem during invalidates.
454  *   It is also used to block modification of page cache contents through
455  *   memory mappings.
456  * @gfp_mask: Memory allocation flags to use for allocating pages.
457  * @i_mmap_writable: Number of VM_SHARED mappings.
458  * @nr_thps: Number of THPs in the pagecache (non-shmem only).
459  * @i_mmap: Tree of private and shared mappings.
460  * @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable.
461  * @nrpages: Number of page entries, protected by the i_pages lock.
462  * @writeback_index: Writeback starts here.
463  * @a_ops: Methods.
464  * @flags: Error bits and flags (AS_*).
465  * @wb_err: The most recent error which has occurred.
466  * @private_lock: For use by the owner of the address_space.
467  * @private_list: For use by the owner of the address_space.
468  * @private_data: For use by the owner of the address_space.
469  */
470 struct address_space {
471 	struct inode		*host;
472 	struct xarray		i_pages;
473 	struct rw_semaphore	invalidate_lock;
474 	gfp_t			gfp_mask;
475 	atomic_t		i_mmap_writable;
476 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
477 	/* number of thp, only for non-shmem files */
478 	atomic_t		nr_thps;
479 #endif
480 	struct rb_root_cached	i_mmap;
481 	struct rw_semaphore	i_mmap_rwsem;
482 	unsigned long		nrpages;
483 	pgoff_t			writeback_index;
484 	const struct address_space_operations *a_ops;
485 	unsigned long		flags;
486 	errseq_t		wb_err;
487 	spinlock_t		private_lock;
488 	struct list_head	private_list;
489 	void			*private_data;
490 
491 	ANDROID_KABI_RESERVE(1);
492 	ANDROID_KABI_RESERVE(2);
493 	ANDROID_KABI_RESERVE(3);
494 	ANDROID_KABI_RESERVE(4);
495 } __attribute__((aligned(sizeof(long)))) __randomize_layout;
496 	/*
497 	 * On most architectures that alignment is already the case; but
498 	 * must be enforced here for CRIS, to let the least significant bit
499 	 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
500 	 */
501 
502 /* XArray tags, for tagging dirty and writeback pages in the pagecache. */
503 #define PAGECACHE_TAG_DIRTY	XA_MARK_0
504 #define PAGECACHE_TAG_WRITEBACK	XA_MARK_1
505 #define PAGECACHE_TAG_TOWRITE	XA_MARK_2
506 
507 /*
508  * Returns true if any of the pages in the mapping are marked with the tag.
509  */
mapping_tagged(struct address_space * mapping,xa_mark_t tag)510 static inline bool mapping_tagged(struct address_space *mapping, xa_mark_t tag)
511 {
512 	return xa_marked(&mapping->i_pages, tag);
513 }
514 
i_mmap_lock_write(struct address_space * mapping)515 static inline void i_mmap_lock_write(struct address_space *mapping)
516 {
517 	down_write(&mapping->i_mmap_rwsem);
518 }
519 
i_mmap_trylock_write(struct address_space * mapping)520 static inline int i_mmap_trylock_write(struct address_space *mapping)
521 {
522 	return down_write_trylock(&mapping->i_mmap_rwsem);
523 }
524 
i_mmap_unlock_write(struct address_space * mapping)525 static inline void i_mmap_unlock_write(struct address_space *mapping)
526 {
527 	up_write(&mapping->i_mmap_rwsem);
528 }
529 
i_mmap_trylock_read(struct address_space * mapping)530 static inline int i_mmap_trylock_read(struct address_space *mapping)
531 {
532 	return down_read_trylock(&mapping->i_mmap_rwsem);
533 }
534 
i_mmap_lock_read(struct address_space * mapping)535 static inline void i_mmap_lock_read(struct address_space *mapping)
536 {
537 	down_read(&mapping->i_mmap_rwsem);
538 }
539 
i_mmap_unlock_read(struct address_space * mapping)540 static inline void i_mmap_unlock_read(struct address_space *mapping)
541 {
542 	up_read(&mapping->i_mmap_rwsem);
543 }
544 
i_mmap_assert_locked(struct address_space * mapping)545 static inline void i_mmap_assert_locked(struct address_space *mapping)
546 {
547 	lockdep_assert_held(&mapping->i_mmap_rwsem);
548 }
549 
i_mmap_assert_write_locked(struct address_space * mapping)550 static inline void i_mmap_assert_write_locked(struct address_space *mapping)
551 {
552 	lockdep_assert_held_write(&mapping->i_mmap_rwsem);
553 }
554 
555 /*
556  * Might pages of this file be mapped into userspace?
557  */
mapping_mapped(struct address_space * mapping)558 static inline int mapping_mapped(struct address_space *mapping)
559 {
560 	return	!RB_EMPTY_ROOT(&mapping->i_mmap.rb_root);
561 }
562 
563 /*
564  * Might pages of this file have been modified in userspace?
565  * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap
566  * marks vma as VM_SHARED if it is shared, and the file was opened for
567  * writing i.e. vma may be mprotected writable even if now readonly.
568  *
569  * If i_mmap_writable is negative, no new writable mappings are allowed. You
570  * can only deny writable mappings, if none exists right now.
571  */
mapping_writably_mapped(struct address_space * mapping)572 static inline int mapping_writably_mapped(struct address_space *mapping)
573 {
574 	return atomic_read(&mapping->i_mmap_writable) > 0;
575 }
576 
mapping_map_writable(struct address_space * mapping)577 static inline int mapping_map_writable(struct address_space *mapping)
578 {
579 	return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
580 		0 : -EPERM;
581 }
582 
mapping_unmap_writable(struct address_space * mapping)583 static inline void mapping_unmap_writable(struct address_space *mapping)
584 {
585 	atomic_dec(&mapping->i_mmap_writable);
586 }
587 
mapping_deny_writable(struct address_space * mapping)588 static inline int mapping_deny_writable(struct address_space *mapping)
589 {
590 	return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
591 		0 : -EBUSY;
592 }
593 
mapping_allow_writable(struct address_space * mapping)594 static inline void mapping_allow_writable(struct address_space *mapping)
595 {
596 	atomic_inc(&mapping->i_mmap_writable);
597 }
598 
599 /*
600  * Use sequence counter to get consistent i_size on 32-bit processors.
601  */
602 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
603 #include <linux/seqlock.h>
604 #define __NEED_I_SIZE_ORDERED
605 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
606 #else
607 #define i_size_ordered_init(inode) do { } while (0)
608 #endif
609 
610 struct posix_acl;
611 #define ACL_NOT_CACHED ((void *)(-1))
612 /*
613  * ACL_DONT_CACHE is for stacked filesystems, that rely on underlying fs to
614  * cache the ACL.  This also means that ->get_acl() can be called in RCU mode
615  * with the LOOKUP_RCU flag.
616  */
617 #define ACL_DONT_CACHE ((void *)(-3))
618 
619 static inline struct posix_acl *
uncached_acl_sentinel(struct task_struct * task)620 uncached_acl_sentinel(struct task_struct *task)
621 {
622 	return (void *)task + 1;
623 }
624 
625 static inline bool
is_uncached_acl(struct posix_acl * acl)626 is_uncached_acl(struct posix_acl *acl)
627 {
628 	return (long)acl & 1;
629 }
630 
631 #define IOP_FASTPERM	0x0001
632 #define IOP_LOOKUP	0x0002
633 #define IOP_NOFOLLOW	0x0004
634 #define IOP_XATTR	0x0008
635 #define IOP_DEFAULT_READLINK	0x0010
636 
637 struct fsnotify_mark_connector;
638 
639 /*
640  * Keep mostly read-only and often accessed (especially for
641  * the RCU path lookup and 'stat' data) fields at the beginning
642  * of the 'struct inode'
643  */
644 struct inode {
645 	umode_t			i_mode;
646 	unsigned short		i_opflags;
647 	kuid_t			i_uid;
648 	kgid_t			i_gid;
649 	unsigned int		i_flags;
650 
651 #ifdef CONFIG_FS_POSIX_ACL
652 	struct posix_acl	*i_acl;
653 	struct posix_acl	*i_default_acl;
654 #endif
655 
656 	const struct inode_operations	*i_op;
657 	struct super_block	*i_sb;
658 	struct address_space	*i_mapping;
659 
660 #ifdef CONFIG_SECURITY
661 	void			*i_security;
662 #endif
663 
664 	/* Stat data, not accessed from path walking */
665 	unsigned long		i_ino;
666 	/*
667 	 * Filesystems may only read i_nlink directly.  They shall use the
668 	 * following functions for modification:
669 	 *
670 	 *    (set|clear|inc|drop)_nlink
671 	 *    inode_(inc|dec)_link_count
672 	 */
673 	union {
674 		const unsigned int i_nlink;
675 		unsigned int __i_nlink;
676 	};
677 	dev_t			i_rdev;
678 	loff_t			i_size;
679 	struct timespec64	i_atime;
680 	struct timespec64	i_mtime;
681 	struct timespec64	i_ctime;
682 	spinlock_t		i_lock;	/* i_blocks, i_bytes, maybe i_size */
683 	unsigned short          i_bytes;
684 	u8			i_blkbits;
685 	u8			i_write_hint;
686 	blkcnt_t		i_blocks;
687 
688 #ifdef __NEED_I_SIZE_ORDERED
689 	seqcount_t		i_size_seqcount;
690 #endif
691 
692 	/* Misc */
693 	unsigned long		i_state;
694 	struct rw_semaphore	i_rwsem;
695 
696 	unsigned long		dirtied_when;	/* jiffies of first dirtying */
697 	unsigned long		dirtied_time_when;
698 
699 	struct hlist_node	i_hash;
700 	struct list_head	i_io_list;	/* backing dev IO list */
701 #ifdef CONFIG_CGROUP_WRITEBACK
702 	struct bdi_writeback	*i_wb;		/* the associated cgroup wb */
703 
704 	/* foreign inode detection, see wbc_detach_inode() */
705 	int			i_wb_frn_winner;
706 	u16			i_wb_frn_avg_time;
707 	u16			i_wb_frn_history;
708 #endif
709 	struct list_head	i_lru;		/* inode LRU list */
710 	struct list_head	i_sb_list;
711 	struct list_head	i_wb_list;	/* backing dev writeback list */
712 	union {
713 		struct hlist_head	i_dentry;
714 		struct rcu_head		i_rcu;
715 	};
716 	atomic64_t		i_version;
717 	atomic64_t		i_sequence; /* see futex */
718 	atomic_t		i_count;
719 	atomic_t		i_dio_count;
720 	atomic_t		i_writecount;
721 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
722 	atomic_t		i_readcount; /* struct files open RO */
723 #endif
724 	union {
725 		const struct file_operations	*i_fop;	/* former ->i_op->default_file_ops */
726 		void (*free_inode)(struct inode *);
727 	};
728 	struct file_lock_context	*i_flctx;
729 	struct address_space	i_data;
730 	struct list_head	i_devices;
731 	union {
732 		struct pipe_inode_info	*i_pipe;
733 		struct cdev		*i_cdev;
734 		char			*i_link;
735 		unsigned		i_dir_seq;
736 	};
737 
738 	__u32			i_generation;
739 
740 #ifdef CONFIG_FSNOTIFY
741 	__u32			i_fsnotify_mask; /* all events this inode cares about */
742 	struct fsnotify_mark_connector __rcu	*i_fsnotify_marks;
743 #endif
744 
745 #ifdef CONFIG_FS_ENCRYPTION
746 	struct fscrypt_info	*i_crypt_info;
747 #endif
748 
749 #ifdef CONFIG_FS_VERITY
750 	struct fsverity_info	*i_verity_info;
751 #endif
752 
753 	void			*i_private; /* fs or device private pointer */
754 
755 	ANDROID_KABI_RESERVE(1);
756 	ANDROID_KABI_RESERVE(2);
757 } __randomize_layout;
758 
759 struct timespec64 timestamp_truncate(struct timespec64 t, struct inode *inode);
760 
i_blocksize(const struct inode * node)761 static inline unsigned int i_blocksize(const struct inode *node)
762 {
763 	return (1 << node->i_blkbits);
764 }
765 
inode_unhashed(struct inode * inode)766 static inline int inode_unhashed(struct inode *inode)
767 {
768 	return hlist_unhashed(&inode->i_hash);
769 }
770 
771 /*
772  * __mark_inode_dirty expects inodes to be hashed.  Since we don't
773  * want special inodes in the fileset inode space, we make them
774  * appear hashed, but do not put on any lists.  hlist_del()
775  * will work fine and require no locking.
776  */
inode_fake_hash(struct inode * inode)777 static inline void inode_fake_hash(struct inode *inode)
778 {
779 	hlist_add_fake(&inode->i_hash);
780 }
781 
782 /*
783  * inode->i_mutex nesting subclasses for the lock validator:
784  *
785  * 0: the object of the current VFS operation
786  * 1: parent
787  * 2: child/target
788  * 3: xattr
789  * 4: second non-directory
790  * 5: second parent (when locking independent directories in rename)
791  *
792  * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
793  * non-directories at once.
794  *
795  * The locking order between these classes is
796  * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
797  */
798 enum inode_i_mutex_lock_class
799 {
800 	I_MUTEX_NORMAL,
801 	I_MUTEX_PARENT,
802 	I_MUTEX_CHILD,
803 	I_MUTEX_XATTR,
804 	I_MUTEX_NONDIR2,
805 	I_MUTEX_PARENT2,
806 };
807 
inode_lock(struct inode * inode)808 static inline void inode_lock(struct inode *inode)
809 {
810 	down_write(&inode->i_rwsem);
811 }
812 
inode_unlock(struct inode * inode)813 static inline void inode_unlock(struct inode *inode)
814 {
815 	up_write(&inode->i_rwsem);
816 }
817 
inode_lock_shared(struct inode * inode)818 static inline void inode_lock_shared(struct inode *inode)
819 {
820 	down_read(&inode->i_rwsem);
821 }
822 
inode_unlock_shared(struct inode * inode)823 static inline void inode_unlock_shared(struct inode *inode)
824 {
825 	up_read(&inode->i_rwsem);
826 }
827 
inode_trylock(struct inode * inode)828 static inline int inode_trylock(struct inode *inode)
829 {
830 	return down_write_trylock(&inode->i_rwsem);
831 }
832 
inode_trylock_shared(struct inode * inode)833 static inline int inode_trylock_shared(struct inode *inode)
834 {
835 	return down_read_trylock(&inode->i_rwsem);
836 }
837 
inode_is_locked(struct inode * inode)838 static inline int inode_is_locked(struct inode *inode)
839 {
840 	return rwsem_is_locked(&inode->i_rwsem);
841 }
842 
inode_lock_nested(struct inode * inode,unsigned subclass)843 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
844 {
845 	down_write_nested(&inode->i_rwsem, subclass);
846 }
847 
inode_lock_shared_nested(struct inode * inode,unsigned subclass)848 static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass)
849 {
850 	down_read_nested(&inode->i_rwsem, subclass);
851 }
852 
filemap_invalidate_lock(struct address_space * mapping)853 static inline void filemap_invalidate_lock(struct address_space *mapping)
854 {
855 	down_write(&mapping->invalidate_lock);
856 }
857 
filemap_invalidate_unlock(struct address_space * mapping)858 static inline void filemap_invalidate_unlock(struct address_space *mapping)
859 {
860 	up_write(&mapping->invalidate_lock);
861 }
862 
filemap_invalidate_lock_shared(struct address_space * mapping)863 static inline void filemap_invalidate_lock_shared(struct address_space *mapping)
864 {
865 	down_read(&mapping->invalidate_lock);
866 }
867 
filemap_invalidate_trylock_shared(struct address_space * mapping)868 static inline int filemap_invalidate_trylock_shared(
869 					struct address_space *mapping)
870 {
871 	return down_read_trylock(&mapping->invalidate_lock);
872 }
873 
filemap_invalidate_unlock_shared(struct address_space * mapping)874 static inline void filemap_invalidate_unlock_shared(
875 					struct address_space *mapping)
876 {
877 	up_read(&mapping->invalidate_lock);
878 }
879 
880 void lock_two_nondirectories(struct inode *, struct inode*);
881 void unlock_two_nondirectories(struct inode *, struct inode*);
882 
883 void filemap_invalidate_lock_two(struct address_space *mapping1,
884 				 struct address_space *mapping2);
885 void filemap_invalidate_unlock_two(struct address_space *mapping1,
886 				   struct address_space *mapping2);
887 
888 
889 /*
890  * NOTE: in a 32bit arch with a preemptable kernel and
891  * an UP compile the i_size_read/write must be atomic
892  * with respect to the local cpu (unlike with preempt disabled),
893  * but they don't need to be atomic with respect to other cpus like in
894  * true SMP (so they need either to either locally disable irq around
895  * the read or for example on x86 they can be still implemented as a
896  * cmpxchg8b without the need of the lock prefix). For SMP compiles
897  * and 64bit archs it makes no difference if preempt is enabled or not.
898  */
i_size_read(const struct inode * inode)899 static inline loff_t i_size_read(const struct inode *inode)
900 {
901 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
902 	loff_t i_size;
903 	unsigned int seq;
904 
905 	do {
906 		seq = read_seqcount_begin(&inode->i_size_seqcount);
907 		i_size = inode->i_size;
908 	} while (read_seqcount_retry(&inode->i_size_seqcount, seq));
909 	return i_size;
910 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
911 	loff_t i_size;
912 
913 	preempt_disable();
914 	i_size = inode->i_size;
915 	preempt_enable();
916 	return i_size;
917 #else
918 	return inode->i_size;
919 #endif
920 }
921 
922 /*
923  * NOTE: unlike i_size_read(), i_size_write() does need locking around it
924  * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
925  * can be lost, resulting in subsequent i_size_read() calls spinning forever.
926  */
i_size_write(struct inode * inode,loff_t i_size)927 static inline void i_size_write(struct inode *inode, loff_t i_size)
928 {
929 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
930 	preempt_disable();
931 	write_seqcount_begin(&inode->i_size_seqcount);
932 	inode->i_size = i_size;
933 	write_seqcount_end(&inode->i_size_seqcount);
934 	preempt_enable();
935 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
936 	preempt_disable();
937 	inode->i_size = i_size;
938 	preempt_enable();
939 #else
940 	inode->i_size = i_size;
941 #endif
942 }
943 
iminor(const struct inode * inode)944 static inline unsigned iminor(const struct inode *inode)
945 {
946 	return MINOR(inode->i_rdev);
947 }
948 
imajor(const struct inode * inode)949 static inline unsigned imajor(const struct inode *inode)
950 {
951 	return MAJOR(inode->i_rdev);
952 }
953 
954 struct fown_struct {
955 	rwlock_t lock;          /* protects pid, uid, euid fields */
956 	struct pid *pid;	/* pid or -pgrp where SIGIO should be sent */
957 	enum pid_type pid_type;	/* Kind of process group SIGIO should be sent to */
958 	kuid_t uid, euid;	/* uid/euid of process setting the owner */
959 	int signum;		/* posix.1b rt signal to be delivered on IO */
960 };
961 
962 /**
963  * struct file_ra_state - Track a file's readahead state.
964  * @start: Where the most recent readahead started.
965  * @size: Number of pages read in the most recent readahead.
966  * @async_size: Start next readahead when this many pages are left.
967  * @ra_pages: Maximum size of a readahead request.
968  * @mmap_miss: How many mmap accesses missed in the page cache.
969  * @prev_pos: The last byte in the most recent read request.
970  */
971 struct file_ra_state {
972 	pgoff_t start;
973 	unsigned int size;
974 	unsigned int async_size;
975 	unsigned int ra_pages;
976 	unsigned int mmap_miss;
977 	loff_t prev_pos;
978 };
979 
980 /*
981  * Check if @index falls in the readahead windows.
982  */
ra_has_index(struct file_ra_state * ra,pgoff_t index)983 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
984 {
985 	return (index >= ra->start &&
986 		index <  ra->start + ra->size);
987 }
988 
989 struct file {
990 	union {
991 		struct llist_node	fu_llist;
992 		struct rcu_head 	fu_rcuhead;
993 	} f_u;
994 	struct path		f_path;
995 	struct inode		*f_inode;	/* cached value */
996 	const struct file_operations	*f_op;
997 
998 	/*
999 	 * Protects f_ep, f_flags.
1000 	 * Must not be taken from IRQ context.
1001 	 */
1002 	spinlock_t		f_lock;
1003 	enum rw_hint		f_write_hint;
1004 	atomic_long_t		f_count;
1005 	unsigned int 		f_flags;
1006 	fmode_t			f_mode;
1007 	struct mutex		f_pos_lock;
1008 	loff_t			f_pos;
1009 	struct fown_struct	f_owner;
1010 	const struct cred	*f_cred;
1011 	struct file_ra_state	f_ra;
1012 
1013 	u64			f_version;
1014 #ifdef CONFIG_SECURITY
1015 	void			*f_security;
1016 #endif
1017 	/* needed for tty driver, and maybe others */
1018 	void			*private_data;
1019 
1020 #ifdef CONFIG_EPOLL
1021 	/* Used by fs/eventpoll.c to link all the hooks to this file */
1022 	struct hlist_head	*f_ep;
1023 #endif /* #ifdef CONFIG_EPOLL */
1024 	struct address_space	*f_mapping;
1025 	errseq_t		f_wb_err;
1026 	errseq_t		f_sb_err; /* for syncfs */
1027 
1028 	ANDROID_KABI_RESERVE(1);
1029 	ANDROID_KABI_RESERVE(2);
1030 } __randomize_layout
1031   __attribute__((aligned(4)));	/* lest something weird decides that 2 is OK */
1032 
1033 struct file_handle {
1034 	__u32 handle_bytes;
1035 	int handle_type;
1036 	/* file identifier */
1037 	unsigned char f_handle[];
1038 };
1039 
get_file(struct file * f)1040 static inline struct file *get_file(struct file *f)
1041 {
1042 	atomic_long_inc(&f->f_count);
1043 	return f;
1044 }
1045 #define get_file_rcu_many(x, cnt)	\
1046 	atomic_long_add_unless(&(x)->f_count, (cnt), 0)
1047 #define get_file_rcu(x) get_file_rcu_many((x), 1)
1048 #define file_count(x)	atomic_long_read(&(x)->f_count)
1049 
1050 #define	MAX_NON_LFS	((1UL<<31) - 1)
1051 
1052 /* Page cache limit. The filesystems should put that into their s_maxbytes
1053    limits, otherwise bad things can happen in VM. */
1054 #if BITS_PER_LONG==32
1055 #define MAX_LFS_FILESIZE	((loff_t)ULONG_MAX << PAGE_SHIFT)
1056 #elif BITS_PER_LONG==64
1057 #define MAX_LFS_FILESIZE 	((loff_t)LLONG_MAX)
1058 #endif
1059 
1060 #define FL_POSIX	1
1061 #define FL_FLOCK	2
1062 #define FL_DELEG	4	/* NFSv4 delegation */
1063 #define FL_ACCESS	8	/* not trying to lock, just looking */
1064 #define FL_EXISTS	16	/* when unlocking, test for existence */
1065 #define FL_LEASE	32	/* lease held on this file */
1066 #define FL_CLOSE	64	/* unlock on close */
1067 #define FL_SLEEP	128	/* A blocking lock */
1068 #define FL_DOWNGRADE_PENDING	256 /* Lease is being downgraded */
1069 #define FL_UNLOCK_PENDING	512 /* Lease is being broken */
1070 #define FL_OFDLCK	1024	/* lock is "owned" by struct file */
1071 #define FL_LAYOUT	2048	/* outstanding pNFS layout */
1072 #define FL_RECLAIM	4096	/* reclaiming from a reboot server */
1073 
1074 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
1075 
1076 /*
1077  * Special return value from posix_lock_file() and vfs_lock_file() for
1078  * asynchronous locking.
1079  */
1080 #define FILE_LOCK_DEFERRED 1
1081 
1082 /* legacy typedef, should eventually be removed */
1083 typedef void *fl_owner_t;
1084 
1085 struct file_lock;
1086 
1087 struct file_lock_operations {
1088 	void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1089 	void (*fl_release_private)(struct file_lock *);
1090 
1091 	ANDROID_KABI_RESERVE(1);
1092 	ANDROID_KABI_RESERVE(2);
1093 };
1094 
1095 struct lock_manager_operations {
1096 	fl_owner_t (*lm_get_owner)(fl_owner_t);
1097 	void (*lm_put_owner)(fl_owner_t);
1098 	void (*lm_notify)(struct file_lock *);	/* unblock callback */
1099 	int (*lm_grant)(struct file_lock *, int);
1100 	bool (*lm_break)(struct file_lock *);
1101 	int (*lm_change)(struct file_lock *, int, struct list_head *);
1102 	void (*lm_setup)(struct file_lock *, void **);
1103 	bool (*lm_breaker_owns_lease)(struct file_lock *);
1104 
1105 	ANDROID_KABI_RESERVE(1);
1106 	ANDROID_KABI_RESERVE(2);
1107 };
1108 
1109 struct lock_manager {
1110 	struct list_head list;
1111 	/*
1112 	 * NFSv4 and up also want opens blocked during the grace period;
1113 	 * NLM doesn't care:
1114 	 */
1115 	bool block_opens;
1116 };
1117 
1118 struct net;
1119 void locks_start_grace(struct net *, struct lock_manager *);
1120 void locks_end_grace(struct lock_manager *);
1121 bool locks_in_grace(struct net *);
1122 bool opens_in_grace(struct net *);
1123 
1124 /* that will die - we need it for nfs_lock_info */
1125 #include <linux/nfs_fs_i.h>
1126 
1127 /*
1128  * struct file_lock represents a generic "file lock". It's used to represent
1129  * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1130  * note that the same struct is used to represent both a request for a lock and
1131  * the lock itself, but the same object is never used for both.
1132  *
1133  * FIXME: should we create a separate "struct lock_request" to help distinguish
1134  * these two uses?
1135  *
1136  * The varous i_flctx lists are ordered by:
1137  *
1138  * 1) lock owner
1139  * 2) lock range start
1140  * 3) lock range end
1141  *
1142  * Obviously, the last two criteria only matter for POSIX locks.
1143  */
1144 struct file_lock {
1145 	struct file_lock *fl_blocker;	/* The lock, that is blocking us */
1146 	struct list_head fl_list;	/* link into file_lock_context */
1147 	struct hlist_node fl_link;	/* node in global lists */
1148 	struct list_head fl_blocked_requests;	/* list of requests with
1149 						 * ->fl_blocker pointing here
1150 						 */
1151 	struct list_head fl_blocked_member;	/* node in
1152 						 * ->fl_blocker->fl_blocked_requests
1153 						 */
1154 	fl_owner_t fl_owner;
1155 	unsigned int fl_flags;
1156 	unsigned char fl_type;
1157 	unsigned int fl_pid;
1158 	int fl_link_cpu;		/* what cpu's list is this on? */
1159 	wait_queue_head_t fl_wait;
1160 	struct file *fl_file;
1161 	loff_t fl_start;
1162 	loff_t fl_end;
1163 
1164 	struct fasync_struct *	fl_fasync; /* for lease break notifications */
1165 	/* for lease breaks: */
1166 	unsigned long fl_break_time;
1167 	unsigned long fl_downgrade_time;
1168 
1169 	const struct file_lock_operations *fl_ops;	/* Callbacks for filesystems */
1170 	const struct lock_manager_operations *fl_lmops;	/* Callbacks for lockmanagers */
1171 	union {
1172 		struct nfs_lock_info	nfs_fl;
1173 		struct nfs4_lock_info	nfs4_fl;
1174 		struct {
1175 			struct list_head link;	/* link in AFS vnode's pending_locks list */
1176 			int state;		/* state of grant or error if -ve */
1177 			unsigned int	debug_id;
1178 		} afs;
1179 	} fl_u;
1180 
1181 	struct list_head android_reserved1;	/* not a macro as we might just need it as-is */
1182 	ANDROID_KABI_RESERVE(1);
1183 	ANDROID_KABI_RESERVE(2);
1184 } __randomize_layout;
1185 
1186 struct file_lock_context {
1187 	spinlock_t		flc_lock;
1188 	struct list_head	flc_flock;
1189 	struct list_head	flc_posix;
1190 	struct list_head	flc_lease;
1191 };
1192 
1193 /* The following constant reflects the upper bound of the file/locking space */
1194 #ifndef OFFSET_MAX
1195 #define INT_LIMIT(x)	(~((x)1 << (sizeof(x)*8 - 1)))
1196 #define OFFSET_MAX	INT_LIMIT(loff_t)
1197 #define OFFT_OFFSET_MAX	INT_LIMIT(off_t)
1198 #endif
1199 
1200 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1201 
1202 #define locks_inode(f) file_inode(f)
1203 
1204 #ifdef CONFIG_FILE_LOCKING
1205 extern int fcntl_getlk(struct file *, unsigned int, struct flock *);
1206 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1207 			struct flock *);
1208 
1209 #if BITS_PER_LONG == 32
1210 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *);
1211 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1212 			struct flock64 *);
1213 #endif
1214 
1215 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1216 extern int fcntl_getlease(struct file *filp);
1217 
1218 /* fs/locks.c */
1219 void locks_free_lock_context(struct inode *inode);
1220 void locks_free_lock(struct file_lock *fl);
1221 extern void locks_init_lock(struct file_lock *);
1222 extern struct file_lock * locks_alloc_lock(void);
1223 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1224 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1225 extern void locks_remove_posix(struct file *, fl_owner_t);
1226 extern void locks_remove_file(struct file *);
1227 extern void locks_release_private(struct file_lock *);
1228 extern void posix_test_lock(struct file *, struct file_lock *);
1229 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1230 extern int locks_delete_block(struct file_lock *);
1231 extern int vfs_test_lock(struct file *, struct file_lock *);
1232 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1233 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1234 bool vfs_inode_has_locks(struct inode *inode);
1235 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1236 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1237 extern void lease_get_mtime(struct inode *, struct timespec64 *time);
1238 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1239 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1240 extern int lease_modify(struct file_lock *, int, struct list_head *);
1241 
1242 struct notifier_block;
1243 extern int lease_register_notifier(struct notifier_block *);
1244 extern void lease_unregister_notifier(struct notifier_block *);
1245 
1246 struct files_struct;
1247 extern void show_fd_locks(struct seq_file *f,
1248 			 struct file *filp, struct files_struct *files);
1249 #else /* !CONFIG_FILE_LOCKING */
fcntl_getlk(struct file * file,unsigned int cmd,struct flock __user * user)1250 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1251 			      struct flock __user *user)
1252 {
1253 	return -EINVAL;
1254 }
1255 
fcntl_setlk(unsigned int fd,struct file * file,unsigned int cmd,struct flock __user * user)1256 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1257 			      unsigned int cmd, struct flock __user *user)
1258 {
1259 	return -EACCES;
1260 }
1261 
1262 #if BITS_PER_LONG == 32
fcntl_getlk64(struct file * file,unsigned int cmd,struct flock64 __user * user)1263 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1264 				struct flock64 __user *user)
1265 {
1266 	return -EINVAL;
1267 }
1268 
fcntl_setlk64(unsigned int fd,struct file * file,unsigned int cmd,struct flock64 __user * user)1269 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1270 				unsigned int cmd, struct flock64 __user *user)
1271 {
1272 	return -EACCES;
1273 }
1274 #endif
fcntl_setlease(unsigned int fd,struct file * filp,long arg)1275 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1276 {
1277 	return -EINVAL;
1278 }
1279 
fcntl_getlease(struct file * filp)1280 static inline int fcntl_getlease(struct file *filp)
1281 {
1282 	return F_UNLCK;
1283 }
1284 
1285 static inline void
locks_free_lock_context(struct inode * inode)1286 locks_free_lock_context(struct inode *inode)
1287 {
1288 }
1289 
locks_init_lock(struct file_lock * fl)1290 static inline void locks_init_lock(struct file_lock *fl)
1291 {
1292 	return;
1293 }
1294 
locks_copy_conflock(struct file_lock * new,struct file_lock * fl)1295 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1296 {
1297 	return;
1298 }
1299 
locks_copy_lock(struct file_lock * new,struct file_lock * fl)1300 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1301 {
1302 	return;
1303 }
1304 
locks_remove_posix(struct file * filp,fl_owner_t owner)1305 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1306 {
1307 	return;
1308 }
1309 
locks_remove_file(struct file * filp)1310 static inline void locks_remove_file(struct file *filp)
1311 {
1312 	return;
1313 }
1314 
posix_test_lock(struct file * filp,struct file_lock * fl)1315 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1316 {
1317 	return;
1318 }
1319 
posix_lock_file(struct file * filp,struct file_lock * fl,struct file_lock * conflock)1320 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1321 				  struct file_lock *conflock)
1322 {
1323 	return -ENOLCK;
1324 }
1325 
locks_delete_block(struct file_lock * waiter)1326 static inline int locks_delete_block(struct file_lock *waiter)
1327 {
1328 	return -ENOENT;
1329 }
1330 
vfs_test_lock(struct file * filp,struct file_lock * fl)1331 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1332 {
1333 	return 0;
1334 }
1335 
vfs_lock_file(struct file * filp,unsigned int cmd,struct file_lock * fl,struct file_lock * conf)1336 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1337 				struct file_lock *fl, struct file_lock *conf)
1338 {
1339 	return -ENOLCK;
1340 }
1341 
vfs_cancel_lock(struct file * filp,struct file_lock * fl)1342 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1343 {
1344 	return 0;
1345 }
1346 
vfs_inode_has_locks(struct inode * inode)1347 static inline bool vfs_inode_has_locks(struct inode *inode)
1348 {
1349 	return false;
1350 }
1351 
locks_lock_inode_wait(struct inode * inode,struct file_lock * fl)1352 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1353 {
1354 	return -ENOLCK;
1355 }
1356 
__break_lease(struct inode * inode,unsigned int mode,unsigned int type)1357 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1358 {
1359 	return 0;
1360 }
1361 
lease_get_mtime(struct inode * inode,struct timespec64 * time)1362 static inline void lease_get_mtime(struct inode *inode,
1363 				   struct timespec64 *time)
1364 {
1365 	return;
1366 }
1367 
generic_setlease(struct file * filp,long arg,struct file_lock ** flp,void ** priv)1368 static inline int generic_setlease(struct file *filp, long arg,
1369 				    struct file_lock **flp, void **priv)
1370 {
1371 	return -EINVAL;
1372 }
1373 
vfs_setlease(struct file * filp,long arg,struct file_lock ** lease,void ** priv)1374 static inline int vfs_setlease(struct file *filp, long arg,
1375 			       struct file_lock **lease, void **priv)
1376 {
1377 	return -EINVAL;
1378 }
1379 
lease_modify(struct file_lock * fl,int arg,struct list_head * dispose)1380 static inline int lease_modify(struct file_lock *fl, int arg,
1381 			       struct list_head *dispose)
1382 {
1383 	return -EINVAL;
1384 }
1385 
1386 struct files_struct;
show_fd_locks(struct seq_file * f,struct file * filp,struct files_struct * files)1387 static inline void show_fd_locks(struct seq_file *f,
1388 			struct file *filp, struct files_struct *files) {}
1389 #endif /* !CONFIG_FILE_LOCKING */
1390 
file_inode(const struct file * f)1391 static inline struct inode *file_inode(const struct file *f)
1392 {
1393 	return f->f_inode;
1394 }
1395 
file_dentry(const struct file * file)1396 static inline struct dentry *file_dentry(const struct file *file)
1397 {
1398 	return d_real(file->f_path.dentry, file_inode(file));
1399 }
1400 
locks_lock_file_wait(struct file * filp,struct file_lock * fl)1401 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1402 {
1403 	return locks_lock_inode_wait(locks_inode(filp), fl);
1404 }
1405 
1406 struct fasync_struct {
1407 	rwlock_t		fa_lock;
1408 	int			magic;
1409 	int			fa_fd;
1410 	struct fasync_struct	*fa_next; /* singly linked list */
1411 	struct file		*fa_file;
1412 	struct rcu_head		fa_rcu;
1413 };
1414 
1415 #define FASYNC_MAGIC 0x4601
1416 
1417 /* SMP safe fasync helpers: */
1418 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1419 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1420 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1421 extern struct fasync_struct *fasync_alloc(void);
1422 extern void fasync_free(struct fasync_struct *);
1423 
1424 /* can be called from interrupts */
1425 extern void kill_fasync(struct fasync_struct **, int, int);
1426 
1427 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1428 extern int f_setown(struct file *filp, unsigned long arg, int force);
1429 extern void f_delown(struct file *filp);
1430 extern pid_t f_getown(struct file *filp);
1431 extern int send_sigurg(struct fown_struct *fown);
1432 
1433 /*
1434  * sb->s_flags.  Note that these mirror the equivalent MS_* flags where
1435  * represented in both.
1436  */
1437 #define SB_RDONLY       BIT(0)	/* Mount read-only */
1438 #define SB_NOSUID       BIT(1)	/* Ignore suid and sgid bits */
1439 #define SB_NODEV        BIT(2)	/* Disallow access to device special files */
1440 #define SB_NOEXEC       BIT(3)	/* Disallow program execution */
1441 #define SB_SYNCHRONOUS  BIT(4)	/* Writes are synced at once */
1442 #define SB_MANDLOCK     BIT(6)	/* Allow mandatory locks on an FS */
1443 #define SB_DIRSYNC      BIT(7)	/* Directory modifications are synchronous */
1444 #define SB_NOATIME      BIT(10)	/* Do not update access times. */
1445 #define SB_NODIRATIME   BIT(11)	/* Do not update directory access times */
1446 #define SB_SILENT       BIT(15)
1447 #define SB_POSIXACL     BIT(16)	/* VFS does not apply the umask */
1448 #define SB_INLINECRYPT  BIT(17)	/* Use blk-crypto for encrypted files */
1449 #define SB_KERNMOUNT    BIT(22)	/* this is a kern_mount call */
1450 #define SB_I_VERSION    BIT(23)	/* Update inode I_version field */
1451 #define SB_LAZYTIME     BIT(25)	/* Update the on-disk [acm]times lazily */
1452 
1453 /* These sb flags are internal to the kernel */
1454 #define SB_SUBMOUNT     BIT(26)
1455 #define SB_FORCE        BIT(27)
1456 #define SB_NOSEC        BIT(28)
1457 #define SB_BORN         BIT(29)
1458 #define SB_ACTIVE       BIT(30)
1459 #define SB_NOUSER       BIT(31)
1460 
1461 /* These flags relate to encoding and casefolding */
1462 #define SB_ENC_STRICT_MODE_FL	(1 << 0)
1463 
1464 #define sb_has_strict_encoding(sb) \
1465 	(sb->s_encoding_flags & SB_ENC_STRICT_MODE_FL)
1466 
1467 /*
1468  *	Umount options
1469  */
1470 
1471 #define MNT_FORCE	0x00000001	/* Attempt to forcibily umount */
1472 #define MNT_DETACH	0x00000002	/* Just detach from the tree */
1473 #define MNT_EXPIRE	0x00000004	/* Mark for expiry */
1474 #define UMOUNT_NOFOLLOW	0x00000008	/* Don't follow symlink on umount */
1475 #define UMOUNT_UNUSED	0x80000000	/* Flag guaranteed to be unused */
1476 
1477 /* sb->s_iflags */
1478 #define SB_I_CGROUPWB	0x00000001	/* cgroup-aware writeback enabled */
1479 #define SB_I_NOEXEC	0x00000002	/* Ignore executables on this fs */
1480 #define SB_I_NODEV	0x00000004	/* Ignore devices on this fs */
1481 #define SB_I_STABLE_WRITES 0x00000008	/* don't modify blks until WB is done */
1482 
1483 /* sb->s_iflags to limit user namespace mounts */
1484 #define SB_I_USERNS_VISIBLE		0x00000010 /* fstype already mounted */
1485 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE	0x00000020
1486 #define SB_I_UNTRUSTED_MOUNTER		0x00000040
1487 
1488 #define SB_I_SKIP_SYNC	0x00000100	/* Skip superblock at global sync */
1489 
1490 /* Possible states of 'frozen' field */
1491 enum {
1492 	SB_UNFROZEN = 0,		/* FS is unfrozen */
1493 	SB_FREEZE_WRITE	= 1,		/* Writes, dir ops, ioctls frozen */
1494 	SB_FREEZE_PAGEFAULT = 2,	/* Page faults stopped as well */
1495 	SB_FREEZE_FS = 3,		/* For internal FS use (e.g. to stop
1496 					 * internal threads if needed) */
1497 	SB_FREEZE_COMPLETE = 4,		/* ->freeze_fs finished successfully */
1498 };
1499 
1500 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1501 
1502 struct sb_writers {
1503 	int				frozen;		/* Is sb frozen? */
1504 	wait_queue_head_t		wait_unfrozen;	/* wait for thaw */
1505 	struct percpu_rw_semaphore	rw_sem[SB_FREEZE_LEVELS];
1506 };
1507 
1508 struct super_block {
1509 	struct list_head	s_list;		/* Keep this first */
1510 	dev_t			s_dev;		/* search index; _not_ kdev_t */
1511 	unsigned char		s_blocksize_bits;
1512 	unsigned long		s_blocksize;
1513 	loff_t			s_maxbytes;	/* Max file size */
1514 	struct file_system_type	*s_type;
1515 	const struct super_operations	*s_op;
1516 	const struct dquot_operations	*dq_op;
1517 	const struct quotactl_ops	*s_qcop;
1518 	const struct export_operations *s_export_op;
1519 	unsigned long		s_flags;
1520 	unsigned long		s_iflags;	/* internal SB_I_* flags */
1521 	unsigned long		s_magic;
1522 	struct dentry		*s_root;
1523 	struct rw_semaphore	s_umount;
1524 	int			s_count;
1525 	atomic_t		s_active;
1526 #ifdef CONFIG_SECURITY
1527 	void                    *s_security;
1528 #endif
1529 	const struct xattr_handler **s_xattr;
1530 #ifdef CONFIG_FS_ENCRYPTION
1531 	const struct fscrypt_operations	*s_cop;
1532 	struct fscrypt_keyring	*s_master_keys; /* master crypto keys in use */
1533 #endif
1534 #ifdef CONFIG_FS_VERITY
1535 	const struct fsverity_operations *s_vop;
1536 #endif
1537 #ifdef CONFIG_UNICODE
1538 	struct unicode_map *s_encoding;
1539 	__u16 s_encoding_flags;
1540 #endif
1541 	struct hlist_bl_head	s_roots;	/* alternate root dentries for NFS */
1542 	struct list_head	s_mounts;	/* list of mounts; _not_ for fs use */
1543 	struct block_device	*s_bdev;
1544 	struct backing_dev_info *s_bdi;
1545 	struct mtd_info		*s_mtd;
1546 	struct hlist_node	s_instances;
1547 	unsigned int		s_quota_types;	/* Bitmask of supported quota types */
1548 	struct quota_info	s_dquot;	/* Diskquota specific options */
1549 
1550 	struct sb_writers	s_writers;
1551 
1552 	/*
1553 	 * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and
1554 	 * s_fsnotify_marks together for cache efficiency. They are frequently
1555 	 * accessed and rarely modified.
1556 	 */
1557 	void			*s_fs_info;	/* Filesystem private info */
1558 
1559 	/* Granularity of c/m/atime in ns (cannot be worse than a second) */
1560 	u32			s_time_gran;
1561 	/* Time limits for c/m/atime in seconds */
1562 	time64_t		   s_time_min;
1563 	time64_t		   s_time_max;
1564 #ifdef CONFIG_FSNOTIFY
1565 	__u32			s_fsnotify_mask;
1566 	struct fsnotify_mark_connector __rcu	*s_fsnotify_marks;
1567 #endif
1568 
1569 	char			s_id[32];	/* Informational name */
1570 	uuid_t			s_uuid;		/* UUID */
1571 
1572 	unsigned int		s_max_links;
1573 	fmode_t			s_mode;
1574 
1575 	/*
1576 	 * The next field is for VFS *only*. No filesystems have any business
1577 	 * even looking at it. You had been warned.
1578 	 */
1579 	struct mutex s_vfs_rename_mutex;	/* Kludge */
1580 
1581 	/*
1582 	 * Filesystem subtype.  If non-empty the filesystem type field
1583 	 * in /proc/mounts will be "type.subtype"
1584 	 */
1585 	const char *s_subtype;
1586 
1587 	const struct dentry_operations *s_d_op; /* default d_op for dentries */
1588 
1589 	/*
1590 	 * Saved pool identifier for cleancache (-1 means none)
1591 	 */
1592 	int cleancache_poolid;
1593 
1594 	struct shrinker s_shrink;	/* per-sb shrinker handle */
1595 
1596 	/* Number of inodes with nlink == 0 but still referenced */
1597 	atomic_long_t s_remove_count;
1598 
1599 	/*
1600 	 * Number of inode/mount/sb objects that are being watched, note that
1601 	 * inodes objects are currently double-accounted.
1602 	 */
1603 	atomic_long_t s_fsnotify_connectors;
1604 
1605 	/* Being remounted read-only */
1606 	int s_readonly_remount;
1607 
1608 	/* per-sb errseq_t for reporting writeback errors via syncfs */
1609 	errseq_t s_wb_err;
1610 
1611 	/* AIO completions deferred from interrupt context */
1612 	struct workqueue_struct *s_dio_done_wq;
1613 	struct hlist_head s_pins;
1614 
1615 	/*
1616 	 * Owning user namespace and default context in which to
1617 	 * interpret filesystem uids, gids, quotas, device nodes,
1618 	 * xattrs and security labels.
1619 	 */
1620 	struct user_namespace *s_user_ns;
1621 
1622 	/*
1623 	 * The list_lru structure is essentially just a pointer to a table
1624 	 * of per-node lru lists, each of which has its own spinlock.
1625 	 * There is no need to put them into separate cachelines.
1626 	 */
1627 	struct list_lru		s_dentry_lru;
1628 	struct list_lru		s_inode_lru;
1629 	struct rcu_head		rcu;
1630 	struct work_struct	destroy_work;
1631 
1632 	struct mutex		s_sync_lock;	/* sync serialisation lock */
1633 
1634 	/*
1635 	 * Indicates how deep in a filesystem stack this SB is
1636 	 */
1637 	int s_stack_depth;
1638 
1639 	/* s_inode_list_lock protects s_inodes */
1640 	spinlock_t		s_inode_list_lock ____cacheline_aligned_in_smp;
1641 	struct list_head	s_inodes;	/* all inodes */
1642 
1643 	spinlock_t		s_inode_wblist_lock;
1644 	struct list_head	s_inodes_wb;	/* writeback inodes */
1645 
1646 	ANDROID_KABI_RESERVE(1);
1647 	ANDROID_KABI_RESERVE(2);
1648 	ANDROID_KABI_RESERVE(3);
1649 	ANDROID_KABI_RESERVE(4);
1650 } __randomize_layout;
1651 
i_user_ns(const struct inode * inode)1652 static inline struct user_namespace *i_user_ns(const struct inode *inode)
1653 {
1654 	return inode->i_sb->s_user_ns;
1655 }
1656 
1657 /* Helper functions so that in most cases filesystems will
1658  * not need to deal directly with kuid_t and kgid_t and can
1659  * instead deal with the raw numeric values that are stored
1660  * in the filesystem.
1661  */
i_uid_read(const struct inode * inode)1662 static inline uid_t i_uid_read(const struct inode *inode)
1663 {
1664 	return from_kuid(i_user_ns(inode), inode->i_uid);
1665 }
1666 
i_gid_read(const struct inode * inode)1667 static inline gid_t i_gid_read(const struct inode *inode)
1668 {
1669 	return from_kgid(i_user_ns(inode), inode->i_gid);
1670 }
1671 
i_uid_write(struct inode * inode,uid_t uid)1672 static inline void i_uid_write(struct inode *inode, uid_t uid)
1673 {
1674 	inode->i_uid = make_kuid(i_user_ns(inode), uid);
1675 }
1676 
i_gid_write(struct inode * inode,gid_t gid)1677 static inline void i_gid_write(struct inode *inode, gid_t gid)
1678 {
1679 	inode->i_gid = make_kgid(i_user_ns(inode), gid);
1680 }
1681 
1682 /**
1683  * i_uid_into_mnt - map an inode's i_uid down into a mnt_userns
1684  * @mnt_userns: user namespace of the mount the inode was found from
1685  * @inode: inode to map
1686  *
1687  * Return: the inode's i_uid mapped down according to @mnt_userns.
1688  * If the inode's i_uid has no mapping INVALID_UID is returned.
1689  */
i_uid_into_mnt(struct user_namespace * mnt_userns,const struct inode * inode)1690 static inline kuid_t i_uid_into_mnt(struct user_namespace *mnt_userns,
1691 				    const struct inode *inode)
1692 {
1693 	return mapped_kuid_fs(mnt_userns, i_user_ns(inode), inode->i_uid);
1694 }
1695 
1696 /**
1697  * i_gid_into_mnt - map an inode's i_gid down into a mnt_userns
1698  * @mnt_userns: user namespace of the mount the inode was found from
1699  * @inode: inode to map
1700  *
1701  * Return: the inode's i_gid mapped down according to @mnt_userns.
1702  * If the inode's i_gid has no mapping INVALID_GID is returned.
1703  */
i_gid_into_mnt(struct user_namespace * mnt_userns,const struct inode * inode)1704 static inline kgid_t i_gid_into_mnt(struct user_namespace *mnt_userns,
1705 				    const struct inode *inode)
1706 {
1707 	return mapped_kgid_fs(mnt_userns, i_user_ns(inode), inode->i_gid);
1708 }
1709 
1710 /**
1711  * inode_fsuid_set - initialize inode's i_uid field with callers fsuid
1712  * @inode: inode to initialize
1713  * @mnt_userns: user namespace of the mount the inode was found from
1714  *
1715  * Initialize the i_uid field of @inode. If the inode was found/created via
1716  * an idmapped mount map the caller's fsuid according to @mnt_users.
1717  */
inode_fsuid_set(struct inode * inode,struct user_namespace * mnt_userns)1718 static inline void inode_fsuid_set(struct inode *inode,
1719 				   struct user_namespace *mnt_userns)
1720 {
1721 	inode->i_uid = mapped_fsuid(mnt_userns, i_user_ns(inode));
1722 }
1723 
1724 /**
1725  * inode_fsgid_set - initialize inode's i_gid field with callers fsgid
1726  * @inode: inode to initialize
1727  * @mnt_userns: user namespace of the mount the inode was found from
1728  *
1729  * Initialize the i_gid field of @inode. If the inode was found/created via
1730  * an idmapped mount map the caller's fsgid according to @mnt_users.
1731  */
inode_fsgid_set(struct inode * inode,struct user_namespace * mnt_userns)1732 static inline void inode_fsgid_set(struct inode *inode,
1733 				   struct user_namespace *mnt_userns)
1734 {
1735 	inode->i_gid = mapped_fsgid(mnt_userns, i_user_ns(inode));
1736 }
1737 
1738 /**
1739  * fsuidgid_has_mapping() - check whether caller's fsuid/fsgid is mapped
1740  * @sb: the superblock we want a mapping in
1741  * @mnt_userns: user namespace of the relevant mount
1742  *
1743  * Check whether the caller's fsuid and fsgid have a valid mapping in the
1744  * s_user_ns of the superblock @sb. If the caller is on an idmapped mount map
1745  * the caller's fsuid and fsgid according to the @mnt_userns first.
1746  *
1747  * Return: true if fsuid and fsgid is mapped, false if not.
1748  */
fsuidgid_has_mapping(struct super_block * sb,struct user_namespace * mnt_userns)1749 static inline bool fsuidgid_has_mapping(struct super_block *sb,
1750 					struct user_namespace *mnt_userns)
1751 {
1752 	struct user_namespace *fs_userns = sb->s_user_ns;
1753 	kuid_t kuid;
1754 	kgid_t kgid;
1755 
1756 	kuid = mapped_fsuid(mnt_userns, fs_userns);
1757 	if (!uid_valid(kuid))
1758 		return false;
1759 	kgid = mapped_fsgid(mnt_userns, fs_userns);
1760 	if (!gid_valid(kgid))
1761 		return false;
1762 	return kuid_has_mapping(fs_userns, kuid) &&
1763 	       kgid_has_mapping(fs_userns, kgid);
1764 }
1765 
1766 struct timespec64 current_time(struct inode *inode);
1767 struct timespec64 inode_set_ctime_current(struct inode *inode);
1768 
1769 /**
1770  * inode_get_ctime - fetch the current ctime from the inode
1771  * @inode: inode from which to fetch ctime
1772  *
1773  * Grab the current ctime from the inode and return it.
1774  */
inode_get_ctime(const struct inode * inode)1775 static inline struct timespec64 inode_get_ctime(const struct inode *inode)
1776 {
1777 	return inode->i_ctime;
1778 }
1779 
1780 /**
1781  * inode_set_ctime_to_ts - set the ctime in the inode
1782  * @inode: inode in which to set the ctime
1783  * @ts: value to set in the ctime field
1784  *
1785  * Set the ctime in @inode to @ts
1786  */
inode_set_ctime_to_ts(struct inode * inode,struct timespec64 ts)1787 static inline struct timespec64 inode_set_ctime_to_ts(struct inode *inode,
1788 						      struct timespec64 ts)
1789 {
1790 	inode->i_ctime = ts;
1791 	return ts;
1792 }
1793 
1794 /**
1795  * inode_set_ctime - set the ctime in the inode
1796  * @inode: inode in which to set the ctime
1797  * @sec: tv_sec value to set
1798  * @nsec: tv_nsec value to set
1799  *
1800  * Set the ctime in @inode to { @sec, @nsec }
1801  */
inode_set_ctime(struct inode * inode,time64_t sec,long nsec)1802 static inline struct timespec64 inode_set_ctime(struct inode *inode,
1803 						time64_t sec, long nsec)
1804 {
1805 	struct timespec64 ts = { .tv_sec  = sec,
1806 				 .tv_nsec = nsec };
1807 
1808 	return inode_set_ctime_to_ts(inode, ts);
1809 }
1810 
1811 /*
1812  * Snapshotting support.
1813  */
1814 
1815 /*
1816  * These are internal functions, please use sb_start_{write,pagefault,intwrite}
1817  * instead.
1818  */
__sb_end_write(struct super_block * sb,int level)1819 static inline void __sb_end_write(struct super_block *sb, int level)
1820 {
1821 	percpu_up_read(sb->s_writers.rw_sem + level-1);
1822 }
1823 
__sb_start_write(struct super_block * sb,int level)1824 static inline void __sb_start_write(struct super_block *sb, int level)
1825 {
1826 	percpu_down_read(sb->s_writers.rw_sem + level - 1);
1827 }
1828 
__sb_start_write_trylock(struct super_block * sb,int level)1829 static inline bool __sb_start_write_trylock(struct super_block *sb, int level)
1830 {
1831 	return percpu_down_read_trylock(sb->s_writers.rw_sem + level - 1);
1832 }
1833 
1834 #define __sb_writers_acquired(sb, lev)	\
1835 	percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1836 #define __sb_writers_release(sb, lev)	\
1837 	percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1838 
1839 /**
1840  * sb_end_write - drop write access to a superblock
1841  * @sb: the super we wrote to
1842  *
1843  * Decrement number of writers to the filesystem. Wake up possible waiters
1844  * wanting to freeze the filesystem.
1845  */
sb_end_write(struct super_block * sb)1846 static inline void sb_end_write(struct super_block *sb)
1847 {
1848 	__sb_end_write(sb, SB_FREEZE_WRITE);
1849 }
1850 
1851 /**
1852  * sb_end_pagefault - drop write access to a superblock from a page fault
1853  * @sb: the super we wrote to
1854  *
1855  * Decrement number of processes handling write page fault to the filesystem.
1856  * Wake up possible waiters wanting to freeze the filesystem.
1857  */
sb_end_pagefault(struct super_block * sb)1858 static inline void sb_end_pagefault(struct super_block *sb)
1859 {
1860 	__sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1861 }
1862 
1863 /**
1864  * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1865  * @sb: the super we wrote to
1866  *
1867  * Decrement fs-internal number of writers to the filesystem.  Wake up possible
1868  * waiters wanting to freeze the filesystem.
1869  */
sb_end_intwrite(struct super_block * sb)1870 static inline void sb_end_intwrite(struct super_block *sb)
1871 {
1872 	__sb_end_write(sb, SB_FREEZE_FS);
1873 }
1874 
1875 /**
1876  * sb_start_write - get write access to a superblock
1877  * @sb: the super we write to
1878  *
1879  * When a process wants to write data or metadata to a file system (i.e. dirty
1880  * a page or an inode), it should embed the operation in a sb_start_write() -
1881  * sb_end_write() pair to get exclusion against file system freezing. This
1882  * function increments number of writers preventing freezing. If the file
1883  * system is already frozen, the function waits until the file system is
1884  * thawed.
1885  *
1886  * Since freeze protection behaves as a lock, users have to preserve
1887  * ordering of freeze protection and other filesystem locks. Generally,
1888  * freeze protection should be the outermost lock. In particular, we have:
1889  *
1890  * sb_start_write
1891  *   -> i_mutex			(write path, truncate, directory ops, ...)
1892  *   -> s_umount		(freeze_super, thaw_super)
1893  */
sb_start_write(struct super_block * sb)1894 static inline void sb_start_write(struct super_block *sb)
1895 {
1896 	__sb_start_write(sb, SB_FREEZE_WRITE);
1897 }
1898 
sb_start_write_trylock(struct super_block * sb)1899 static inline bool sb_start_write_trylock(struct super_block *sb)
1900 {
1901 	return __sb_start_write_trylock(sb, SB_FREEZE_WRITE);
1902 }
1903 
1904 /**
1905  * sb_start_pagefault - get write access to a superblock from a page fault
1906  * @sb: the super we write to
1907  *
1908  * When a process starts handling write page fault, it should embed the
1909  * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1910  * exclusion against file system freezing. This is needed since the page fault
1911  * is going to dirty a page. This function increments number of running page
1912  * faults preventing freezing. If the file system is already frozen, the
1913  * function waits until the file system is thawed.
1914  *
1915  * Since page fault freeze protection behaves as a lock, users have to preserve
1916  * ordering of freeze protection and other filesystem locks. It is advised to
1917  * put sb_start_pagefault() close to mmap_lock in lock ordering. Page fault
1918  * handling code implies lock dependency:
1919  *
1920  * mmap_lock
1921  *   -> sb_start_pagefault
1922  */
sb_start_pagefault(struct super_block * sb)1923 static inline void sb_start_pagefault(struct super_block *sb)
1924 {
1925 	__sb_start_write(sb, SB_FREEZE_PAGEFAULT);
1926 }
1927 
1928 /**
1929  * sb_start_intwrite - get write access to a superblock for internal fs purposes
1930  * @sb: the super we write to
1931  *
1932  * This is the third level of protection against filesystem freezing. It is
1933  * free for use by a filesystem. The only requirement is that it must rank
1934  * below sb_start_pagefault.
1935  *
1936  * For example filesystem can call sb_start_intwrite() when starting a
1937  * transaction which somewhat eases handling of freezing for internal sources
1938  * of filesystem changes (internal fs threads, discarding preallocation on file
1939  * close, etc.).
1940  */
sb_start_intwrite(struct super_block * sb)1941 static inline void sb_start_intwrite(struct super_block *sb)
1942 {
1943 	__sb_start_write(sb, SB_FREEZE_FS);
1944 }
1945 
sb_start_intwrite_trylock(struct super_block * sb)1946 static inline bool sb_start_intwrite_trylock(struct super_block *sb)
1947 {
1948 	return __sb_start_write_trylock(sb, SB_FREEZE_FS);
1949 }
1950 
1951 bool inode_owner_or_capable(struct user_namespace *mnt_userns,
1952 			    const struct inode *inode);
1953 
1954 /*
1955  * VFS helper functions..
1956  */
1957 int vfs_create(struct user_namespace *, struct inode *,
1958 	       struct dentry *, umode_t, bool);
1959 int vfs_mkdir(struct user_namespace *, struct inode *,
1960 	      struct dentry *, umode_t);
1961 int vfs_mknod(struct user_namespace *, struct inode *, struct dentry *,
1962               umode_t, dev_t);
1963 int vfs_symlink(struct user_namespace *, struct inode *,
1964 		struct dentry *, const char *);
1965 int vfs_link(struct dentry *, struct user_namespace *, struct inode *,
1966 	     struct dentry *, struct inode **);
1967 int vfs_rmdir(struct user_namespace *, struct inode *, struct dentry *);
1968 int vfs_unlink(struct user_namespace *, struct inode *, struct dentry *,
1969 	       struct inode **);
1970 
1971 /**
1972  * struct renamedata - contains all information required for renaming
1973  * @old_mnt_userns:    old user namespace of the mount the inode was found from
1974  * @old_dir:           parent of source
1975  * @old_dentry:                source
1976  * @new_mnt_userns:    new user namespace of the mount the inode was found from
1977  * @new_dir:           parent of destination
1978  * @new_dentry:                destination
1979  * @delegated_inode:   returns an inode needing a delegation break
1980  * @flags:             rename flags
1981  */
1982 struct renamedata {
1983 	struct user_namespace *old_mnt_userns;
1984 	struct inode *old_dir;
1985 	struct dentry *old_dentry;
1986 	struct user_namespace *new_mnt_userns;
1987 	struct inode *new_dir;
1988 	struct dentry *new_dentry;
1989 	struct inode **delegated_inode;
1990 	unsigned int flags;
1991 } __randomize_layout;
1992 
1993 int vfs_rename(struct renamedata *);
1994 
vfs_whiteout(struct user_namespace * mnt_userns,struct inode * dir,struct dentry * dentry)1995 static inline int vfs_whiteout(struct user_namespace *mnt_userns,
1996 			       struct inode *dir, struct dentry *dentry)
1997 {
1998 	return vfs_mknod(mnt_userns, dir, dentry, S_IFCHR | WHITEOUT_MODE,
1999 			 WHITEOUT_DEV);
2000 }
2001 
2002 struct dentry *vfs_tmpfile(struct user_namespace *mnt_userns,
2003 			   struct dentry *dentry, umode_t mode, int open_flag);
2004 
2005 int vfs_mkobj(struct dentry *, umode_t,
2006 		int (*f)(struct dentry *, umode_t, void *),
2007 		void *);
2008 
2009 int vfs_fchown(struct file *file, uid_t user, gid_t group);
2010 int vfs_fchmod(struct file *file, umode_t mode);
2011 int vfs_utimes(const struct path *path, struct timespec64 *times);
2012 
2013 extern long vfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
2014 
2015 #ifdef CONFIG_COMPAT
2016 extern long compat_ptr_ioctl(struct file *file, unsigned int cmd,
2017 					unsigned long arg);
2018 #else
2019 #define compat_ptr_ioctl NULL
2020 #endif
2021 
2022 /*
2023  * VFS file helper functions.
2024  */
2025 void inode_init_owner(struct user_namespace *mnt_userns, struct inode *inode,
2026 		      const struct inode *dir, umode_t mode);
2027 extern bool may_open_dev(const struct path *path);
2028 umode_t mode_strip_sgid(struct user_namespace *mnt_userns,
2029 			const struct inode *dir, umode_t mode);
2030 
2031 /*
2032  * This is the "filldir" function type, used by readdir() to let
2033  * the kernel specify what kind of dirent layout it wants to have.
2034  * This allows the kernel to read directories into kernel space or
2035  * to have different dirent layouts depending on the binary type.
2036  */
2037 struct dir_context;
2038 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
2039 			 unsigned);
2040 
2041 struct dir_context {
2042 	filldir_t actor;
2043 	loff_t pos;
2044 };
2045 
2046 /*
2047  * These flags let !MMU mmap() govern direct device mapping vs immediate
2048  * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
2049  *
2050  * NOMMU_MAP_COPY:	Copy can be mapped (MAP_PRIVATE)
2051  * NOMMU_MAP_DIRECT:	Can be mapped directly (MAP_SHARED)
2052  * NOMMU_MAP_READ:	Can be mapped for reading
2053  * NOMMU_MAP_WRITE:	Can be mapped for writing
2054  * NOMMU_MAP_EXEC:	Can be mapped for execution
2055  */
2056 #define NOMMU_MAP_COPY		0x00000001
2057 #define NOMMU_MAP_DIRECT	0x00000008
2058 #define NOMMU_MAP_READ		VM_MAYREAD
2059 #define NOMMU_MAP_WRITE		VM_MAYWRITE
2060 #define NOMMU_MAP_EXEC		VM_MAYEXEC
2061 
2062 #define NOMMU_VMFLAGS \
2063 	(NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
2064 
2065 /*
2066  * These flags control the behavior of the remap_file_range function pointer.
2067  * If it is called with len == 0 that means "remap to end of source file".
2068  * See Documentation/filesystems/vfs.rst for more details about this call.
2069  *
2070  * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate)
2071  * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request
2072  */
2073 #define REMAP_FILE_DEDUP		(1 << 0)
2074 #define REMAP_FILE_CAN_SHORTEN		(1 << 1)
2075 
2076 /*
2077  * These flags signal that the caller is ok with altering various aspects of
2078  * the behavior of the remap operation.  The changes must be made by the
2079  * implementation; the vfs remap helper functions can take advantage of them.
2080  * Flags in this category exist to preserve the quirky behavior of the hoisted
2081  * btrfs clone/dedupe ioctls.
2082  */
2083 #define REMAP_FILE_ADVISORY		(REMAP_FILE_CAN_SHORTEN)
2084 
2085 /*
2086  * These flags control the behavior of vfs_copy_file_range().
2087  * They are not available to the user via syscall.
2088  *
2089  * COPY_FILE_SPLICE: call splice direct instead of fs clone/copy ops
2090  */
2091 #define COPY_FILE_SPLICE		(1 << 0)
2092 
2093 struct iov_iter;
2094 
2095 struct file_operations {
2096 	struct module *owner;
2097 	loff_t (*llseek) (struct file *, loff_t, int);
2098 	ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
2099 	ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
2100 	ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
2101 	ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
2102 	int (*iopoll)(struct kiocb *kiocb, bool spin);
2103 	int (*iterate) (struct file *, struct dir_context *);
2104 	int (*iterate_shared) (struct file *, struct dir_context *);
2105 	__poll_t (*poll) (struct file *, struct poll_table_struct *);
2106 	long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
2107 	long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
2108 	int (*mmap) (struct file *, struct vm_area_struct *);
2109 	unsigned long mmap_supported_flags;
2110 	int (*open) (struct inode *, struct file *);
2111 	int (*flush) (struct file *, fl_owner_t id);
2112 	int (*release) (struct inode *, struct file *);
2113 	int (*fsync) (struct file *, loff_t, loff_t, int datasync);
2114 	int (*fasync) (int, struct file *, int);
2115 	int (*lock) (struct file *, int, struct file_lock *);
2116 	ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
2117 	unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
2118 	int (*check_flags)(int);
2119 	int (*flock) (struct file *, int, struct file_lock *);
2120 	ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
2121 	ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
2122 	int (*setlease)(struct file *, long, struct file_lock **, void **);
2123 	long (*fallocate)(struct file *file, int mode, loff_t offset,
2124 			  loff_t len);
2125 	void (*show_fdinfo)(struct seq_file *m, struct file *f);
2126 #ifndef CONFIG_MMU
2127 	unsigned (*mmap_capabilities)(struct file *);
2128 #endif
2129 	ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
2130 			loff_t, size_t, unsigned int);
2131 	loff_t (*remap_file_range)(struct file *file_in, loff_t pos_in,
2132 				   struct file *file_out, loff_t pos_out,
2133 				   loff_t len, unsigned int remap_flags);
2134 	int (*fadvise)(struct file *, loff_t, loff_t, int);
2135 
2136 	ANDROID_KABI_RESERVE(1);
2137 	ANDROID_KABI_RESERVE(2);
2138 	ANDROID_KABI_RESERVE(3);
2139 	ANDROID_KABI_RESERVE(4);
2140 } __randomize_layout;
2141 
2142 struct inode_operations {
2143 	struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
2144 	const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
2145 	int (*permission) (struct user_namespace *, struct inode *, int);
2146 	struct posix_acl * (*get_acl)(struct inode *, int, bool);
2147 
2148 	int (*readlink) (struct dentry *, char __user *,int);
2149 
2150 	int (*create) (struct user_namespace *, struct inode *,struct dentry *,
2151 		       umode_t, bool);
2152 	int (*link) (struct dentry *,struct inode *,struct dentry *);
2153 	int (*unlink) (struct inode *,struct dentry *);
2154 	int (*symlink) (struct user_namespace *, struct inode *,struct dentry *,
2155 			const char *);
2156 	int (*mkdir) (struct user_namespace *, struct inode *,struct dentry *,
2157 		      umode_t);
2158 	int (*rmdir) (struct inode *,struct dentry *);
2159 	int (*mknod) (struct user_namespace *, struct inode *,struct dentry *,
2160 		      umode_t,dev_t);
2161 	int (*rename) (struct user_namespace *, struct inode *, struct dentry *,
2162 			struct inode *, struct dentry *, unsigned int);
2163 	int (*setattr) (struct user_namespace *, struct dentry *,
2164 			struct iattr *);
2165 	int (*getattr) (struct user_namespace *, const struct path *,
2166 			struct kstat *, u32, unsigned int);
2167 	ssize_t (*listxattr) (struct dentry *, char *, size_t);
2168 	int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
2169 		      u64 len);
2170 	int (*update_time)(struct inode *, struct timespec64 *, int);
2171 	int (*atomic_open)(struct inode *, struct dentry *,
2172 			   struct file *, unsigned open_flag,
2173 			   umode_t create_mode);
2174 	int (*tmpfile) (struct user_namespace *, struct inode *,
2175 			struct dentry *, umode_t);
2176 	int (*set_acl)(struct user_namespace *, struct inode *,
2177 		       struct posix_acl *, int);
2178 	int (*fileattr_set)(struct user_namespace *mnt_userns,
2179 			    struct dentry *dentry, struct fileattr *fa);
2180 	int (*fileattr_get)(struct dentry *dentry, struct fileattr *fa);
2181 
2182 	ANDROID_KABI_RESERVE(1);
2183 	ANDROID_KABI_RESERVE(2);
2184 	ANDROID_KABI_RESERVE(3);
2185 	ANDROID_KABI_RESERVE(4);
2186 } ____cacheline_aligned;
2187 
call_read_iter(struct file * file,struct kiocb * kio,struct iov_iter * iter)2188 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
2189 				     struct iov_iter *iter)
2190 {
2191 	return file->f_op->read_iter(kio, iter);
2192 }
2193 
call_write_iter(struct file * file,struct kiocb * kio,struct iov_iter * iter)2194 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
2195 				      struct iov_iter *iter)
2196 {
2197 	return file->f_op->write_iter(kio, iter);
2198 }
2199 
call_mmap(struct file * file,struct vm_area_struct * vma)2200 static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
2201 {
2202 	return file->f_op->mmap(file, vma);
2203 }
2204 
2205 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
2206 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
2207 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
2208 				   loff_t, size_t, unsigned int);
2209 extern ssize_t generic_copy_file_range(struct file *file_in, loff_t pos_in,
2210 				       struct file *file_out, loff_t pos_out,
2211 				       size_t len, unsigned int flags);
2212 extern int generic_remap_file_range_prep(struct file *file_in, loff_t pos_in,
2213 					 struct file *file_out, loff_t pos_out,
2214 					 loff_t *count,
2215 					 unsigned int remap_flags);
2216 extern loff_t do_clone_file_range(struct file *file_in, loff_t pos_in,
2217 				  struct file *file_out, loff_t pos_out,
2218 				  loff_t len, unsigned int remap_flags);
2219 extern loff_t vfs_clone_file_range(struct file *file_in, loff_t pos_in,
2220 				   struct file *file_out, loff_t pos_out,
2221 				   loff_t len, unsigned int remap_flags);
2222 extern int vfs_dedupe_file_range(struct file *file,
2223 				 struct file_dedupe_range *same);
2224 extern loff_t vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos,
2225 					struct file *dst_file, loff_t dst_pos,
2226 					loff_t len, unsigned int remap_flags);
2227 
2228 
2229 struct super_operations {
2230    	struct inode *(*alloc_inode)(struct super_block *sb);
2231 	void (*destroy_inode)(struct inode *);
2232 	void (*free_inode)(struct inode *);
2233 
2234    	void (*dirty_inode) (struct inode *, int flags);
2235 	int (*write_inode) (struct inode *, struct writeback_control *wbc);
2236 	int (*drop_inode) (struct inode *);
2237 	void (*evict_inode) (struct inode *);
2238 	void (*put_super) (struct super_block *);
2239 	int (*sync_fs)(struct super_block *sb, int wait);
2240 	int (*freeze_super) (struct super_block *);
2241 	int (*freeze_fs) (struct super_block *);
2242 	int (*thaw_super) (struct super_block *);
2243 	int (*unfreeze_fs) (struct super_block *);
2244 	int (*statfs) (struct dentry *, struct kstatfs *);
2245 	int (*remount_fs) (struct super_block *, int *, char *);
2246 	void (*umount_begin) (struct super_block *);
2247 
2248 	int (*show_options)(struct seq_file *, struct dentry *);
2249 	int (*show_devname)(struct seq_file *, struct dentry *);
2250 	int (*show_path)(struct seq_file *, struct dentry *);
2251 	int (*show_stats)(struct seq_file *, struct dentry *);
2252 #ifdef CONFIG_QUOTA
2253 	ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
2254 	ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
2255 	struct dquot **(*get_dquots)(struct inode *);
2256 #endif
2257 	long (*nr_cached_objects)(struct super_block *,
2258 				  struct shrink_control *);
2259 	long (*free_cached_objects)(struct super_block *,
2260 				    struct shrink_control *);
2261 
2262 	ANDROID_KABI_RESERVE(1);
2263 	ANDROID_KABI_RESERVE(2);
2264 	ANDROID_KABI_RESERVE(3);
2265 	ANDROID_KABI_RESERVE(4);
2266 };
2267 
2268 /*
2269  * Inode flags - they have no relation to superblock flags now
2270  */
2271 #define S_SYNC		(1 << 0)  /* Writes are synced at once */
2272 #define S_NOATIME	(1 << 1)  /* Do not update access times */
2273 #define S_APPEND	(1 << 2)  /* Append-only file */
2274 #define S_IMMUTABLE	(1 << 3)  /* Immutable file */
2275 #define S_DEAD		(1 << 4)  /* removed, but still open directory */
2276 #define S_NOQUOTA	(1 << 5)  /* Inode is not counted to quota */
2277 #define S_DIRSYNC	(1 << 6)  /* Directory modifications are synchronous */
2278 #define S_NOCMTIME	(1 << 7)  /* Do not update file c/mtime */
2279 #define S_SWAPFILE	(1 << 8)  /* Do not truncate: swapon got its bmaps */
2280 #define S_PRIVATE	(1 << 9)  /* Inode is fs-internal */
2281 #define S_IMA		(1 << 10) /* Inode has an associated IMA struct */
2282 #define S_AUTOMOUNT	(1 << 11) /* Automount/referral quasi-directory */
2283 #define S_NOSEC		(1 << 12) /* no suid or xattr security attributes */
2284 #ifdef CONFIG_FS_DAX
2285 #define S_DAX		(1 << 13) /* Direct Access, avoiding the page cache */
2286 #else
2287 #define S_DAX		0	  /* Make all the DAX code disappear */
2288 #endif
2289 #define S_ENCRYPTED	(1 << 14) /* Encrypted file (using fs/crypto/) */
2290 #define S_CASEFOLD	(1 << 15) /* Casefolded file */
2291 #define S_VERITY	(1 << 16) /* Verity file (using fs/verity/) */
2292 
2293 /*
2294  * Note that nosuid etc flags are inode-specific: setting some file-system
2295  * flags just means all the inodes inherit those flags by default. It might be
2296  * possible to override it selectively if you really wanted to with some
2297  * ioctl() that is not currently implemented.
2298  *
2299  * Exception: SB_RDONLY is always applied to the entire file system.
2300  *
2301  * Unfortunately, it is possible to change a filesystems flags with it mounted
2302  * with files in use.  This means that all of the inodes will not have their
2303  * i_flags updated.  Hence, i_flags no longer inherit the superblock mount
2304  * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
2305  */
2306 #define __IS_FLG(inode, flg)	((inode)->i_sb->s_flags & (flg))
2307 
sb_rdonly(const struct super_block * sb)2308 static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; }
2309 #define IS_RDONLY(inode)	sb_rdonly((inode)->i_sb)
2310 #define IS_SYNC(inode)		(__IS_FLG(inode, SB_SYNCHRONOUS) || \
2311 					((inode)->i_flags & S_SYNC))
2312 #define IS_DIRSYNC(inode)	(__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
2313 					((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
2314 #define IS_MANDLOCK(inode)	__IS_FLG(inode, SB_MANDLOCK)
2315 #define IS_NOATIME(inode)	__IS_FLG(inode, SB_RDONLY|SB_NOATIME)
2316 #define IS_I_VERSION(inode)	__IS_FLG(inode, SB_I_VERSION)
2317 
2318 #define IS_NOQUOTA(inode)	((inode)->i_flags & S_NOQUOTA)
2319 #define IS_APPEND(inode)	((inode)->i_flags & S_APPEND)
2320 #define IS_IMMUTABLE(inode)	((inode)->i_flags & S_IMMUTABLE)
2321 #define IS_POSIXACL(inode)	__IS_FLG(inode, SB_POSIXACL)
2322 
2323 #define IS_DEADDIR(inode)	((inode)->i_flags & S_DEAD)
2324 #define IS_NOCMTIME(inode)	((inode)->i_flags & S_NOCMTIME)
2325 #define IS_SWAPFILE(inode)	((inode)->i_flags & S_SWAPFILE)
2326 #define IS_PRIVATE(inode)	((inode)->i_flags & S_PRIVATE)
2327 #define IS_IMA(inode)		((inode)->i_flags & S_IMA)
2328 #define IS_AUTOMOUNT(inode)	((inode)->i_flags & S_AUTOMOUNT)
2329 #define IS_NOSEC(inode)		((inode)->i_flags & S_NOSEC)
2330 #define IS_DAX(inode)		((inode)->i_flags & S_DAX)
2331 #define IS_ENCRYPTED(inode)	((inode)->i_flags & S_ENCRYPTED)
2332 #define IS_CASEFOLDED(inode)	((inode)->i_flags & S_CASEFOLD)
2333 #define IS_VERITY(inode)	((inode)->i_flags & S_VERITY)
2334 
2335 #define IS_WHITEOUT(inode)	(S_ISCHR(inode->i_mode) && \
2336 				 (inode)->i_rdev == WHITEOUT_DEV)
2337 
HAS_UNMAPPED_ID(struct user_namespace * mnt_userns,struct inode * inode)2338 static inline bool HAS_UNMAPPED_ID(struct user_namespace *mnt_userns,
2339 				   struct inode *inode)
2340 {
2341 	return !uid_valid(i_uid_into_mnt(mnt_userns, inode)) ||
2342 	       !gid_valid(i_gid_into_mnt(mnt_userns, inode));
2343 }
2344 
file_write_hint(struct file * file)2345 static inline enum rw_hint file_write_hint(struct file *file)
2346 {
2347 	if (file->f_write_hint != WRITE_LIFE_NOT_SET)
2348 		return file->f_write_hint;
2349 
2350 	return file_inode(file)->i_write_hint;
2351 }
2352 
2353 static inline int iocb_flags(struct file *file);
2354 
ki_hint_validate(enum rw_hint hint)2355 static inline u16 ki_hint_validate(enum rw_hint hint)
2356 {
2357 	typeof(((struct kiocb *)0)->ki_hint) max_hint = -1;
2358 
2359 	if (hint <= max_hint)
2360 		return hint;
2361 	return 0;
2362 }
2363 
init_sync_kiocb(struct kiocb * kiocb,struct file * filp)2364 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
2365 {
2366 	*kiocb = (struct kiocb) {
2367 		.ki_filp = filp,
2368 		.ki_flags = iocb_flags(filp),
2369 		.ki_hint = ki_hint_validate(file_write_hint(filp)),
2370 		.ki_ioprio = get_current_ioprio(),
2371 	};
2372 }
2373 
kiocb_clone(struct kiocb * kiocb,struct kiocb * kiocb_src,struct file * filp)2374 static inline void kiocb_clone(struct kiocb *kiocb, struct kiocb *kiocb_src,
2375 			       struct file *filp)
2376 {
2377 	*kiocb = (struct kiocb) {
2378 		.ki_filp = filp,
2379 		.ki_flags = kiocb_src->ki_flags,
2380 		.ki_hint = kiocb_src->ki_hint,
2381 		.ki_ioprio = kiocb_src->ki_ioprio,
2382 		.ki_pos = kiocb_src->ki_pos,
2383 	};
2384 }
2385 
2386 /*
2387  * Inode state bits.  Protected by inode->i_lock
2388  *
2389  * Four bits determine the dirty state of the inode: I_DIRTY_SYNC,
2390  * I_DIRTY_DATASYNC, I_DIRTY_PAGES, and I_DIRTY_TIME.
2391  *
2392  * Four bits define the lifetime of an inode.  Initially, inodes are I_NEW,
2393  * until that flag is cleared.  I_WILL_FREE, I_FREEING and I_CLEAR are set at
2394  * various stages of removing an inode.
2395  *
2396  * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2397  *
2398  * I_DIRTY_SYNC		Inode is dirty, but doesn't have to be written on
2399  *			fdatasync() (unless I_DIRTY_DATASYNC is also set).
2400  *			Timestamp updates are the usual cause.
2401  * I_DIRTY_DATASYNC	Data-related inode changes pending.  We keep track of
2402  *			these changes separately from I_DIRTY_SYNC so that we
2403  *			don't have to write inode on fdatasync() when only
2404  *			e.g. the timestamps have changed.
2405  * I_DIRTY_PAGES	Inode has dirty pages.  Inode itself may be clean.
2406  * I_DIRTY_TIME		The inode itself has dirty timestamps, and the
2407  *			lazytime mount option is enabled.  We keep track of this
2408  *			separately from I_DIRTY_SYNC in order to implement
2409  *			lazytime.  This gets cleared if I_DIRTY_INODE
2410  *			(I_DIRTY_SYNC and/or I_DIRTY_DATASYNC) gets set. But
2411  *			I_DIRTY_TIME can still be set if I_DIRTY_SYNC is already
2412  *			in place because writeback might already be in progress
2413  *			and we don't want to lose the time update
2414  * I_NEW		Serves as both a mutex and completion notification.
2415  *			New inodes set I_NEW.  If two processes both create
2416  *			the same inode, one of them will release its inode and
2417  *			wait for I_NEW to be released before returning.
2418  *			Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2419  *			also cause waiting on I_NEW, without I_NEW actually
2420  *			being set.  find_inode() uses this to prevent returning
2421  *			nearly-dead inodes.
2422  * I_WILL_FREE		Must be set when calling write_inode_now() if i_count
2423  *			is zero.  I_FREEING must be set when I_WILL_FREE is
2424  *			cleared.
2425  * I_FREEING		Set when inode is about to be freed but still has dirty
2426  *			pages or buffers attached or the inode itself is still
2427  *			dirty.
2428  * I_CLEAR		Added by clear_inode().  In this state the inode is
2429  *			clean and can be destroyed.  Inode keeps I_FREEING.
2430  *
2431  *			Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2432  *			prohibited for many purposes.  iget() must wait for
2433  *			the inode to be completely released, then create it
2434  *			anew.  Other functions will just ignore such inodes,
2435  *			if appropriate.  I_NEW is used for waiting.
2436  *
2437  * I_SYNC		Writeback of inode is running. The bit is set during
2438  *			data writeback, and cleared with a wakeup on the bit
2439  *			address once it is done. The bit is also used to pin
2440  *			the inode in memory for flusher thread.
2441  *
2442  * I_REFERENCED		Marks the inode as recently references on the LRU list.
2443  *
2444  * I_DIO_WAKEUP		Never set.  Only used as a key for wait_on_bit().
2445  *
2446  * I_WB_SWITCH		Cgroup bdi_writeback switching in progress.  Used to
2447  *			synchronize competing switching instances and to tell
2448  *			wb stat updates to grab the i_pages lock.  See
2449  *			inode_switch_wbs_work_fn() for details.
2450  *
2451  * I_OVL_INUSE		Used by overlayfs to get exclusive ownership on upper
2452  *			and work dirs among overlayfs mounts.
2453  *
2454  * I_CREATING		New object's inode in the middle of setting up.
2455  *
2456  * I_DONTCACHE		Evict inode as soon as it is not used anymore.
2457  *
2458  * I_SYNC_QUEUED	Inode is queued in b_io or b_more_io writeback lists.
2459  *			Used to detect that mark_inode_dirty() should not move
2460  * 			inode between dirty lists.
2461  *
2462  * Q: What is the difference between I_WILL_FREE and I_FREEING?
2463  */
2464 #define I_DIRTY_SYNC		(1 << 0)
2465 #define I_DIRTY_DATASYNC	(1 << 1)
2466 #define I_DIRTY_PAGES		(1 << 2)
2467 #define __I_NEW			3
2468 #define I_NEW			(1 << __I_NEW)
2469 #define I_WILL_FREE		(1 << 4)
2470 #define I_FREEING		(1 << 5)
2471 #define I_CLEAR			(1 << 6)
2472 #define __I_SYNC		7
2473 #define I_SYNC			(1 << __I_SYNC)
2474 #define I_REFERENCED		(1 << 8)
2475 #define __I_DIO_WAKEUP		9
2476 #define I_DIO_WAKEUP		(1 << __I_DIO_WAKEUP)
2477 #define I_LINKABLE		(1 << 10)
2478 #define I_DIRTY_TIME		(1 << 11)
2479 #define I_WB_SWITCH		(1 << 13)
2480 #define I_OVL_INUSE		(1 << 14)
2481 #define I_CREATING		(1 << 15)
2482 #define I_DONTCACHE		(1 << 16)
2483 #define I_SYNC_QUEUED		(1 << 17)
2484 
2485 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2486 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2487 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2488 
2489 extern void __mark_inode_dirty(struct inode *, int);
mark_inode_dirty(struct inode * inode)2490 static inline void mark_inode_dirty(struct inode *inode)
2491 {
2492 	__mark_inode_dirty(inode, I_DIRTY);
2493 }
2494 
mark_inode_dirty_sync(struct inode * inode)2495 static inline void mark_inode_dirty_sync(struct inode *inode)
2496 {
2497 	__mark_inode_dirty(inode, I_DIRTY_SYNC);
2498 }
2499 
2500 /*
2501  * Returns true if the given inode itself only has dirty timestamps (its pages
2502  * may still be dirty) and isn't currently being allocated or freed.
2503  * Filesystems should call this if when writing an inode when lazytime is
2504  * enabled, they want to opportunistically write the timestamps of other inodes
2505  * located very nearby on-disk, e.g. in the same inode block.  This returns true
2506  * if the given inode is in need of such an opportunistic update.  Requires
2507  * i_lock, or at least later re-checking under i_lock.
2508  */
inode_is_dirtytime_only(struct inode * inode)2509 static inline bool inode_is_dirtytime_only(struct inode *inode)
2510 {
2511 	return (inode->i_state & (I_DIRTY_TIME | I_NEW |
2512 				  I_FREEING | I_WILL_FREE)) == I_DIRTY_TIME;
2513 }
2514 
2515 extern void inc_nlink(struct inode *inode);
2516 extern void drop_nlink(struct inode *inode);
2517 extern void clear_nlink(struct inode *inode);
2518 extern void set_nlink(struct inode *inode, unsigned int nlink);
2519 
inode_inc_link_count(struct inode * inode)2520 static inline void inode_inc_link_count(struct inode *inode)
2521 {
2522 	inc_nlink(inode);
2523 	mark_inode_dirty(inode);
2524 }
2525 
inode_dec_link_count(struct inode * inode)2526 static inline void inode_dec_link_count(struct inode *inode)
2527 {
2528 	drop_nlink(inode);
2529 	mark_inode_dirty(inode);
2530 }
2531 
2532 enum file_time_flags {
2533 	S_ATIME = 1,
2534 	S_MTIME = 2,
2535 	S_CTIME = 4,
2536 	S_VERSION = 8,
2537 };
2538 
2539 extern bool atime_needs_update(const struct path *, struct inode *);
2540 extern void touch_atime(const struct path *);
2541 int inode_update_time(struct inode *inode, struct timespec64 *time, int flags);
2542 
file_accessed(struct file * file)2543 static inline void file_accessed(struct file *file)
2544 {
2545 	if (!(file->f_flags & O_NOATIME))
2546 		touch_atime(&file->f_path);
2547 }
2548 
2549 extern int file_modified(struct file *file);
2550 
2551 int sync_inode_metadata(struct inode *inode, int wait);
2552 
2553 struct file_system_type {
2554 	const char *name;
2555 	int fs_flags;
2556 #define FS_REQUIRES_DEV		1
2557 #define FS_BINARY_MOUNTDATA	2
2558 #define FS_HAS_SUBTYPE		4
2559 #define FS_USERNS_MOUNT		8	/* Can be mounted by userns root */
2560 #define FS_DISALLOW_NOTIFY_PERM	16	/* Disable fanotify permission events */
2561 #define FS_ALLOW_IDMAP         32      /* FS has been updated to handle vfs idmappings. */
2562 #define FS_THP_SUPPORT		8192	/* Remove once all fs converted */
2563 #define FS_RENAME_DOES_D_MOVE	32768	/* FS will handle d_move() during rename() internally. */
2564 	int (*init_fs_context)(struct fs_context *);
2565 	const struct fs_parameter_spec *parameters;
2566 	struct dentry *(*mount) (struct file_system_type *, int,
2567 		       const char *, void *);
2568 	void (*kill_sb) (struct super_block *);
2569 	struct module *owner;
2570 	struct file_system_type * next;
2571 	struct hlist_head fs_supers;
2572 
2573 	struct lock_class_key s_lock_key;
2574 	struct lock_class_key s_umount_key;
2575 	struct lock_class_key s_vfs_rename_key;
2576 	struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2577 
2578 	struct lock_class_key i_lock_key;
2579 	struct lock_class_key i_mutex_key;
2580 	struct lock_class_key invalidate_lock_key;
2581 	struct lock_class_key i_mutex_dir_key;
2582 
2583 	ANDROID_KABI_RESERVE(1);
2584 	ANDROID_KABI_RESERVE(2);
2585 	ANDROID_KABI_RESERVE(3);
2586 	ANDROID_KABI_RESERVE(4);
2587 };
2588 
2589 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2590 
2591 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2592 	int flags, const char *dev_name, void *data,
2593 	int (*fill_super)(struct super_block *, void *, int));
2594 extern struct dentry *mount_single(struct file_system_type *fs_type,
2595 	int flags, void *data,
2596 	int (*fill_super)(struct super_block *, void *, int));
2597 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2598 	int flags, void *data,
2599 	int (*fill_super)(struct super_block *, void *, int));
2600 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2601 void generic_shutdown_super(struct super_block *sb);
2602 void kill_block_super(struct super_block *sb);
2603 void kill_anon_super(struct super_block *sb);
2604 void kill_litter_super(struct super_block *sb);
2605 void deactivate_super(struct super_block *sb);
2606 void deactivate_locked_super(struct super_block *sb);
2607 int set_anon_super(struct super_block *s, void *data);
2608 int set_anon_super_fc(struct super_block *s, struct fs_context *fc);
2609 int get_anon_bdev(dev_t *);
2610 void free_anon_bdev(dev_t);
2611 struct super_block *sget_fc(struct fs_context *fc,
2612 			    int (*test)(struct super_block *, struct fs_context *),
2613 			    int (*set)(struct super_block *, struct fs_context *));
2614 struct super_block *sget(struct file_system_type *type,
2615 			int (*test)(struct super_block *,void *),
2616 			int (*set)(struct super_block *,void *),
2617 			int flags, void *data);
2618 
2619 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2620 #define fops_get(fops) \
2621 	(((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2622 #define fops_put(fops) \
2623 	do { if (fops) module_put((fops)->owner); } while(0)
2624 /*
2625  * This one is to be used *ONLY* from ->open() instances.
2626  * fops must be non-NULL, pinned down *and* module dependencies
2627  * should be sufficient to pin the caller down as well.
2628  */
2629 #define replace_fops(f, fops) \
2630 	do {	\
2631 		struct file *__file = (f); \
2632 		fops_put(__file->f_op); \
2633 		BUG_ON(!(__file->f_op = (fops))); \
2634 	} while(0)
2635 
2636 extern int register_filesystem(struct file_system_type *);
2637 extern int unregister_filesystem(struct file_system_type *);
2638 extern struct vfsmount *kern_mount(struct file_system_type *);
2639 extern void kern_unmount(struct vfsmount *mnt);
2640 extern int may_umount_tree(struct vfsmount *);
2641 extern int may_umount(struct vfsmount *);
2642 extern long do_mount(const char *, const char __user *,
2643 		     const char *, unsigned long, void *);
2644 extern struct vfsmount *collect_mounts(const struct path *);
2645 extern void drop_collected_mounts(struct vfsmount *);
2646 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2647 			  struct vfsmount *);
2648 extern int vfs_statfs(const struct path *, struct kstatfs *);
2649 extern int user_statfs(const char __user *, struct kstatfs *);
2650 extern int fd_statfs(int, struct kstatfs *);
2651 extern int freeze_super(struct super_block *super);
2652 extern int thaw_super(struct super_block *super);
2653 extern bool our_mnt(struct vfsmount *mnt);
2654 extern __printf(2, 3)
2655 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2656 extern int super_setup_bdi(struct super_block *sb);
2657 
2658 extern int current_umask(void);
2659 
2660 extern void ihold(struct inode * inode);
2661 extern void iput(struct inode *);
2662 extern int generic_update_time(struct inode *, struct timespec64 *, int);
2663 
2664 /* /sys/fs */
2665 extern struct kobject *fs_kobj;
2666 
2667 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2668 
2669 #ifdef CONFIG_FILE_LOCKING
break_lease(struct inode * inode,unsigned int mode)2670 static inline int break_lease(struct inode *inode, unsigned int mode)
2671 {
2672 	/*
2673 	 * Since this check is lockless, we must ensure that any refcounts
2674 	 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2675 	 * could end up racing with tasks trying to set a new lease on this
2676 	 * file.
2677 	 */
2678 	smp_mb();
2679 	if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2680 		return __break_lease(inode, mode, FL_LEASE);
2681 	return 0;
2682 }
2683 
break_deleg(struct inode * inode,unsigned int mode)2684 static inline int break_deleg(struct inode *inode, unsigned int mode)
2685 {
2686 	/*
2687 	 * Since this check is lockless, we must ensure that any refcounts
2688 	 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2689 	 * could end up racing with tasks trying to set a new lease on this
2690 	 * file.
2691 	 */
2692 	smp_mb();
2693 	if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2694 		return __break_lease(inode, mode, FL_DELEG);
2695 	return 0;
2696 }
2697 
try_break_deleg(struct inode * inode,struct inode ** delegated_inode)2698 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2699 {
2700 	int ret;
2701 
2702 	ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2703 	if (ret == -EWOULDBLOCK && delegated_inode) {
2704 		*delegated_inode = inode;
2705 		ihold(inode);
2706 	}
2707 	return ret;
2708 }
2709 
break_deleg_wait(struct inode ** delegated_inode)2710 static inline int break_deleg_wait(struct inode **delegated_inode)
2711 {
2712 	int ret;
2713 
2714 	ret = break_deleg(*delegated_inode, O_WRONLY);
2715 	iput(*delegated_inode);
2716 	*delegated_inode = NULL;
2717 	return ret;
2718 }
2719 
break_layout(struct inode * inode,bool wait)2720 static inline int break_layout(struct inode *inode, bool wait)
2721 {
2722 	smp_mb();
2723 	if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2724 		return __break_lease(inode,
2725 				wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2726 				FL_LAYOUT);
2727 	return 0;
2728 }
2729 
2730 #else /* !CONFIG_FILE_LOCKING */
break_lease(struct inode * inode,unsigned int mode)2731 static inline int break_lease(struct inode *inode, unsigned int mode)
2732 {
2733 	return 0;
2734 }
2735 
break_deleg(struct inode * inode,unsigned int mode)2736 static inline int break_deleg(struct inode *inode, unsigned int mode)
2737 {
2738 	return 0;
2739 }
2740 
try_break_deleg(struct inode * inode,struct inode ** delegated_inode)2741 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2742 {
2743 	return 0;
2744 }
2745 
break_deleg_wait(struct inode ** delegated_inode)2746 static inline int break_deleg_wait(struct inode **delegated_inode)
2747 {
2748 	BUG();
2749 	return 0;
2750 }
2751 
break_layout(struct inode * inode,bool wait)2752 static inline int break_layout(struct inode *inode, bool wait)
2753 {
2754 	return 0;
2755 }
2756 
2757 #endif /* CONFIG_FILE_LOCKING */
2758 
2759 /* fs/open.c */
2760 struct audit_names;
2761 struct filename {
2762 	const char		*name;	/* pointer to actual string */
2763 	const __user char	*uptr;	/* original userland pointer */
2764 	int			refcnt;
2765 	struct audit_names	*aname;
2766 	const char		iname[];
2767 };
2768 static_assert(offsetof(struct filename, iname) % sizeof(long) == 0);
2769 
file_mnt_user_ns(struct file * file)2770 static inline struct user_namespace *file_mnt_user_ns(struct file *file)
2771 {
2772 	return mnt_user_ns(file->f_path.mnt);
2773 }
2774 
2775 /**
2776  * is_idmapped_mnt - check whether a mount is mapped
2777  * @mnt: the mount to check
2778  *
2779  * If @mnt has an idmapping attached different from the
2780  * filesystem's idmapping then @mnt is mapped.
2781  *
2782  * Return: true if mount is mapped, false if not.
2783  */
is_idmapped_mnt(const struct vfsmount * mnt)2784 static inline bool is_idmapped_mnt(const struct vfsmount *mnt)
2785 {
2786 	return mnt_user_ns(mnt) != mnt->mnt_sb->s_user_ns;
2787 }
2788 
2789 extern long vfs_truncate(const struct path *, loff_t);
2790 int do_truncate(struct user_namespace *, struct dentry *, loff_t start,
2791 		unsigned int time_attrs, struct file *filp);
2792 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2793 			loff_t len);
2794 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2795 			umode_t mode);
2796 extern struct file *file_open_name(struct filename *, int, umode_t);
2797 extern struct file *filp_open(const char *, int, umode_t);
2798 extern struct file *filp_open_block(const char *, int, umode_t);
2799 extern struct file *file_open_root(const struct path *,
2800 				   const char *, int, umode_t);
file_open_root_mnt(struct vfsmount * mnt,const char * name,int flags,umode_t mode)2801 static inline struct file *file_open_root_mnt(struct vfsmount *mnt,
2802 				   const char *name, int flags, umode_t mode)
2803 {
2804 	return file_open_root(&(struct path){.mnt = mnt, .dentry = mnt->mnt_root},
2805 			      name, flags, mode);
2806 }
2807 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2808 extern struct file * open_with_fake_path(const struct path *, int,
2809 					 struct inode*, const struct cred *);
file_clone_open(struct file * file)2810 static inline struct file *file_clone_open(struct file *file)
2811 {
2812 	return dentry_open(&file->f_path, file->f_flags, file->f_cred);
2813 }
2814 extern int filp_close(struct file *, fl_owner_t id);
2815 
2816 extern struct filename *getname_flags(const char __user *, int, int *);
2817 extern struct filename *getname_uflags(const char __user *, int);
2818 extern struct filename *getname(const char __user *);
2819 extern struct filename *getname_kernel(const char *);
2820 extern void putname(struct filename *name);
2821 
2822 extern int finish_open(struct file *file, struct dentry *dentry,
2823 			int (*open)(struct inode *, struct file *));
2824 extern int finish_no_open(struct file *file, struct dentry *dentry);
2825 
2826 /* fs/dcache.c */
2827 extern void __init vfs_caches_init_early(void);
2828 extern void __init vfs_caches_init(void);
2829 
2830 extern struct kmem_cache *names_cachep;
2831 
2832 #define __getname()		kmem_cache_alloc(names_cachep, GFP_KERNEL)
2833 #define __putname(name)		kmem_cache_free(names_cachep, (void *)(name))
2834 
2835 extern struct super_block *blockdev_superblock;
sb_is_blkdev_sb(struct super_block * sb)2836 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2837 {
2838 	return IS_ENABLED(CONFIG_BLOCK) && sb == blockdev_superblock;
2839 }
2840 
2841 void emergency_thaw_all(void);
2842 extern int sync_filesystem(struct super_block *);
2843 extern const struct file_operations def_blk_fops;
2844 extern const struct file_operations def_chr_fops;
2845 
2846 /* fs/char_dev.c */
2847 #define CHRDEV_MAJOR_MAX 512
2848 /* Marks the bottom of the first segment of free char majors */
2849 #define CHRDEV_MAJOR_DYN_END 234
2850 /* Marks the top and bottom of the second segment of free char majors */
2851 #define CHRDEV_MAJOR_DYN_EXT_START 511
2852 #define CHRDEV_MAJOR_DYN_EXT_END 384
2853 
2854 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2855 extern int register_chrdev_region(dev_t, unsigned, const char *);
2856 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2857 			     unsigned int count, const char *name,
2858 			     const struct file_operations *fops);
2859 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2860 				unsigned int count, const char *name);
2861 extern void unregister_chrdev_region(dev_t, unsigned);
2862 extern void chrdev_show(struct seq_file *,off_t);
2863 
register_chrdev(unsigned int major,const char * name,const struct file_operations * fops)2864 static inline int register_chrdev(unsigned int major, const char *name,
2865 				  const struct file_operations *fops)
2866 {
2867 	return __register_chrdev(major, 0, 256, name, fops);
2868 }
2869 
unregister_chrdev(unsigned int major,const char * name)2870 static inline void unregister_chrdev(unsigned int major, const char *name)
2871 {
2872 	__unregister_chrdev(major, 0, 256, name);
2873 }
2874 
2875 extern void init_special_inode(struct inode *, umode_t, dev_t);
2876 
2877 /* Invalid inode operations -- fs/bad_inode.c */
2878 extern void make_bad_inode(struct inode *);
2879 extern bool is_bad_inode(struct inode *);
2880 
2881 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2882 					pgoff_t start, pgoff_t end);
2883 
2884 void invalidate_mapping_pagevec(struct address_space *mapping,
2885 				pgoff_t start, pgoff_t end,
2886 				unsigned long *nr_pagevec);
2887 
invalidate_remote_inode(struct inode * inode)2888 static inline void invalidate_remote_inode(struct inode *inode)
2889 {
2890 	if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2891 	    S_ISLNK(inode->i_mode))
2892 		invalidate_mapping_pages(inode->i_mapping, 0, -1);
2893 }
2894 extern int invalidate_inode_pages2(struct address_space *mapping);
2895 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2896 					 pgoff_t start, pgoff_t end);
2897 extern int write_inode_now(struct inode *, int);
2898 extern int filemap_fdatawrite(struct address_space *);
2899 extern int filemap_flush(struct address_space *);
2900 extern int filemap_fdatawait_keep_errors(struct address_space *mapping);
2901 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2902 				   loff_t lend);
2903 extern int filemap_fdatawait_range_keep_errors(struct address_space *mapping,
2904 		loff_t start_byte, loff_t end_byte);
2905 
filemap_fdatawait(struct address_space * mapping)2906 static inline int filemap_fdatawait(struct address_space *mapping)
2907 {
2908 	return filemap_fdatawait_range(mapping, 0, LLONG_MAX);
2909 }
2910 
2911 extern bool filemap_range_has_page(struct address_space *, loff_t lstart,
2912 				  loff_t lend);
2913 extern bool filemap_range_needs_writeback(struct address_space *,
2914 					  loff_t lstart, loff_t lend);
2915 extern int filemap_write_and_wait_range(struct address_space *mapping,
2916 				        loff_t lstart, loff_t lend);
2917 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2918 				loff_t start, loff_t end, int sync_mode);
2919 extern int filemap_fdatawrite_range(struct address_space *mapping,
2920 				loff_t start, loff_t end);
2921 extern int filemap_check_errors(struct address_space *mapping);
2922 extern void __filemap_set_wb_err(struct address_space *mapping, int err);
2923 int filemap_fdatawrite_wbc(struct address_space *mapping,
2924 			   struct writeback_control *wbc);
2925 
filemap_write_and_wait(struct address_space * mapping)2926 static inline int filemap_write_and_wait(struct address_space *mapping)
2927 {
2928 	return filemap_write_and_wait_range(mapping, 0, LLONG_MAX);
2929 }
2930 
2931 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2932 						loff_t lend);
2933 extern int __must_check file_check_and_advance_wb_err(struct file *file);
2934 extern int __must_check file_write_and_wait_range(struct file *file,
2935 						loff_t start, loff_t end);
2936 
file_write_and_wait(struct file * file)2937 static inline int file_write_and_wait(struct file *file)
2938 {
2939 	return file_write_and_wait_range(file, 0, LLONG_MAX);
2940 }
2941 
2942 /**
2943  * filemap_set_wb_err - set a writeback error on an address_space
2944  * @mapping: mapping in which to set writeback error
2945  * @err: error to be set in mapping
2946  *
2947  * When writeback fails in some way, we must record that error so that
2948  * userspace can be informed when fsync and the like are called.  We endeavor
2949  * to report errors on any file that was open at the time of the error.  Some
2950  * internal callers also need to know when writeback errors have occurred.
2951  *
2952  * When a writeback error occurs, most filesystems will want to call
2953  * filemap_set_wb_err to record the error in the mapping so that it will be
2954  * automatically reported whenever fsync is called on the file.
2955  */
filemap_set_wb_err(struct address_space * mapping,int err)2956 static inline void filemap_set_wb_err(struct address_space *mapping, int err)
2957 {
2958 	/* Fastpath for common case of no error */
2959 	if (unlikely(err))
2960 		__filemap_set_wb_err(mapping, err);
2961 }
2962 
2963 /**
2964  * filemap_check_wb_err - has an error occurred since the mark was sampled?
2965  * @mapping: mapping to check for writeback errors
2966  * @since: previously-sampled errseq_t
2967  *
2968  * Grab the errseq_t value from the mapping, and see if it has changed "since"
2969  * the given value was sampled.
2970  *
2971  * If it has then report the latest error set, otherwise return 0.
2972  */
filemap_check_wb_err(struct address_space * mapping,errseq_t since)2973 static inline int filemap_check_wb_err(struct address_space *mapping,
2974 					errseq_t since)
2975 {
2976 	return errseq_check(&mapping->wb_err, since);
2977 }
2978 
2979 /**
2980  * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2981  * @mapping: mapping to be sampled
2982  *
2983  * Writeback errors are always reported relative to a particular sample point
2984  * in the past. This function provides those sample points.
2985  */
filemap_sample_wb_err(struct address_space * mapping)2986 static inline errseq_t filemap_sample_wb_err(struct address_space *mapping)
2987 {
2988 	return errseq_sample(&mapping->wb_err);
2989 }
2990 
2991 /**
2992  * file_sample_sb_err - sample the current errseq_t to test for later errors
2993  * @file: file pointer to be sampled
2994  *
2995  * Grab the most current superblock-level errseq_t value for the given
2996  * struct file.
2997  */
file_sample_sb_err(struct file * file)2998 static inline errseq_t file_sample_sb_err(struct file *file)
2999 {
3000 	return errseq_sample(&file->f_path.dentry->d_sb->s_wb_err);
3001 }
3002 
3003 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
3004 			   int datasync);
3005 extern int vfs_fsync(struct file *file, int datasync);
3006 
3007 extern int sync_file_range(struct file *file, loff_t offset, loff_t nbytes,
3008 				unsigned int flags);
3009 
3010 /*
3011  * Sync the bytes written if this was a synchronous write.  Expect ki_pos
3012  * to already be updated for the write, and will return either the amount
3013  * of bytes passed in, or an error if syncing the file failed.
3014  */
generic_write_sync(struct kiocb * iocb,ssize_t count)3015 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
3016 {
3017 	if (iocb->ki_flags & IOCB_DSYNC) {
3018 		int ret = vfs_fsync_range(iocb->ki_filp,
3019 				iocb->ki_pos - count, iocb->ki_pos - 1,
3020 				(iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
3021 		if (ret)
3022 			return ret;
3023 	}
3024 
3025 	return count;
3026 }
3027 
3028 extern void emergency_sync(void);
3029 extern void emergency_remount(void);
3030 
3031 #ifdef CONFIG_BLOCK
3032 extern int bmap(struct inode *inode, sector_t *block);
3033 #else
bmap(struct inode * inode,sector_t * block)3034 static inline int bmap(struct inode *inode,  sector_t *block)
3035 {
3036 	return -EINVAL;
3037 }
3038 #endif
3039 
3040 int notify_change(struct user_namespace *, struct dentry *,
3041 		  struct iattr *, struct inode **);
3042 int inode_permission(struct user_namespace *, struct inode *, int);
3043 int generic_permission(struct user_namespace *, struct inode *, int);
file_permission(struct file * file,int mask)3044 static inline int file_permission(struct file *file, int mask)
3045 {
3046 	return inode_permission(file_mnt_user_ns(file),
3047 				file_inode(file), mask);
3048 }
path_permission(const struct path * path,int mask)3049 static inline int path_permission(const struct path *path, int mask)
3050 {
3051 	return inode_permission(mnt_user_ns(path->mnt),
3052 				d_inode(path->dentry), mask);
3053 }
3054 int __check_sticky(struct user_namespace *mnt_userns, struct inode *dir,
3055 		   struct inode *inode);
3056 
execute_ok(struct inode * inode)3057 static inline bool execute_ok(struct inode *inode)
3058 {
3059 	return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
3060 }
3061 
inode_wrong_type(const struct inode * inode,umode_t mode)3062 static inline bool inode_wrong_type(const struct inode *inode, umode_t mode)
3063 {
3064 	return (inode->i_mode ^ mode) & S_IFMT;
3065 }
3066 
file_start_write(struct file * file)3067 static inline void file_start_write(struct file *file)
3068 {
3069 	if (!S_ISREG(file_inode(file)->i_mode))
3070 		return;
3071 	sb_start_write(file_inode(file)->i_sb);
3072 }
3073 
file_start_write_trylock(struct file * file)3074 static inline bool file_start_write_trylock(struct file *file)
3075 {
3076 	if (!S_ISREG(file_inode(file)->i_mode))
3077 		return true;
3078 	return sb_start_write_trylock(file_inode(file)->i_sb);
3079 }
3080 
file_end_write(struct file * file)3081 static inline void file_end_write(struct file *file)
3082 {
3083 	if (!S_ISREG(file_inode(file)->i_mode))
3084 		return;
3085 	__sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
3086 }
3087 
3088 /*
3089  * This is used for regular files where some users -- especially the
3090  * currently executed binary in a process, previously handled via
3091  * VM_DENYWRITE -- cannot handle concurrent write (and maybe mmap
3092  * read-write shared) accesses.
3093  *
3094  * get_write_access() gets write permission for a file.
3095  * put_write_access() releases this write permission.
3096  * deny_write_access() denies write access to a file.
3097  * allow_write_access() re-enables write access to a file.
3098  *
3099  * The i_writecount field of an inode can have the following values:
3100  * 0: no write access, no denied write access
3101  * < 0: (-i_writecount) users that denied write access to the file.
3102  * > 0: (i_writecount) users that have write access to the file.
3103  *
3104  * Normally we operate on that counter with atomic_{inc,dec} and it's safe
3105  * except for the cases where we don't hold i_writecount yet. Then we need to
3106  * use {get,deny}_write_access() - these functions check the sign and refuse
3107  * to do the change if sign is wrong.
3108  */
get_write_access(struct inode * inode)3109 static inline int get_write_access(struct inode *inode)
3110 {
3111 	return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
3112 }
deny_write_access(struct file * file)3113 static inline int deny_write_access(struct file *file)
3114 {
3115 	struct inode *inode = file_inode(file);
3116 	return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
3117 }
put_write_access(struct inode * inode)3118 static inline void put_write_access(struct inode * inode)
3119 {
3120 	atomic_dec(&inode->i_writecount);
3121 }
allow_write_access(struct file * file)3122 static inline void allow_write_access(struct file *file)
3123 {
3124 	if (file)
3125 		atomic_inc(&file_inode(file)->i_writecount);
3126 }
inode_is_open_for_write(const struct inode * inode)3127 static inline bool inode_is_open_for_write(const struct inode *inode)
3128 {
3129 	return atomic_read(&inode->i_writecount) > 0;
3130 }
3131 
3132 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
i_readcount_dec(struct inode * inode)3133 static inline void i_readcount_dec(struct inode *inode)
3134 {
3135 	BUG_ON(!atomic_read(&inode->i_readcount));
3136 	atomic_dec(&inode->i_readcount);
3137 }
i_readcount_inc(struct inode * inode)3138 static inline void i_readcount_inc(struct inode *inode)
3139 {
3140 	atomic_inc(&inode->i_readcount);
3141 }
3142 #else
i_readcount_dec(struct inode * inode)3143 static inline void i_readcount_dec(struct inode *inode)
3144 {
3145 	return;
3146 }
i_readcount_inc(struct inode * inode)3147 static inline void i_readcount_inc(struct inode *inode)
3148 {
3149 	return;
3150 }
3151 #endif
3152 extern int do_pipe_flags(int *, int);
3153 
3154 extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
3155 ssize_t __kernel_read(struct file *file, void *buf, size_t count, loff_t *pos);
3156 extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
3157 extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
3158 extern struct file * open_exec(const char *);
3159 
3160 /* fs/dcache.c -- generic fs support functions */
3161 extern bool is_subdir(struct dentry *, struct dentry *);
3162 extern bool path_is_under(const struct path *, const struct path *);
3163 
3164 extern char *file_path(struct file *, char *, int);
3165 
3166 #include <linux/err.h>
3167 
3168 /* needed for stackable file system support */
3169 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
3170 
3171 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
3172 
3173 extern int inode_init_always(struct super_block *, struct inode *);
3174 extern void inode_init_once(struct inode *);
3175 extern void address_space_init_once(struct address_space *mapping);
3176 extern struct inode * igrab(struct inode *);
3177 extern ino_t iunique(struct super_block *, ino_t);
3178 extern int inode_needs_sync(struct inode *inode);
3179 extern int generic_delete_inode(struct inode *inode);
generic_drop_inode(struct inode * inode)3180 static inline int generic_drop_inode(struct inode *inode)
3181 {
3182 	return !inode->i_nlink || inode_unhashed(inode);
3183 }
3184 extern void d_mark_dontcache(struct inode *inode);
3185 
3186 extern struct inode *ilookup5_nowait(struct super_block *sb,
3187 		unsigned long hashval, int (*test)(struct inode *, void *),
3188 		void *data);
3189 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
3190 		int (*test)(struct inode *, void *), void *data);
3191 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
3192 
3193 extern struct inode *inode_insert5(struct inode *inode, unsigned long hashval,
3194 		int (*test)(struct inode *, void *),
3195 		int (*set)(struct inode *, void *),
3196 		void *data);
3197 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
3198 extern struct inode * iget_locked(struct super_block *, unsigned long);
3199 extern struct inode *find_inode_nowait(struct super_block *,
3200 				       unsigned long,
3201 				       int (*match)(struct inode *,
3202 						    unsigned long, void *),
3203 				       void *data);
3204 extern struct inode *find_inode_rcu(struct super_block *, unsigned long,
3205 				    int (*)(struct inode *, void *), void *);
3206 extern struct inode *find_inode_by_ino_rcu(struct super_block *, unsigned long);
3207 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
3208 extern int insert_inode_locked(struct inode *);
3209 #ifdef CONFIG_DEBUG_LOCK_ALLOC
3210 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
3211 #else
lockdep_annotate_inode_mutex_key(struct inode * inode)3212 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
3213 #endif
3214 extern void unlock_new_inode(struct inode *);
3215 extern void discard_new_inode(struct inode *);
3216 extern unsigned int get_next_ino(void);
3217 extern void evict_inodes(struct super_block *sb);
3218 
3219 /*
3220  * Userspace may rely on the the inode number being non-zero. For example, glibc
3221  * simply ignores files with zero i_ino in unlink() and other places.
3222  *
3223  * As an additional complication, if userspace was compiled with
3224  * _FILE_OFFSET_BITS=32 on a 64-bit kernel we'll only end up reading out the
3225  * lower 32 bits, so we need to check that those aren't zero explicitly. With
3226  * _FILE_OFFSET_BITS=64, this may cause some harmless false-negatives, but
3227  * better safe than sorry.
3228  */
is_zero_ino(ino_t ino)3229 static inline bool is_zero_ino(ino_t ino)
3230 {
3231 	return (u32)ino == 0;
3232 }
3233 
3234 extern void __iget(struct inode * inode);
3235 extern void iget_failed(struct inode *);
3236 extern void clear_inode(struct inode *);
3237 extern void __destroy_inode(struct inode *);
3238 extern struct inode *new_inode_pseudo(struct super_block *sb);
3239 extern struct inode *new_inode(struct super_block *sb);
3240 extern void free_inode_nonrcu(struct inode *inode);
3241 extern int setattr_should_drop_suidgid(struct user_namespace *, struct inode *);
3242 extern int file_remove_privs(struct file *);
3243 int setattr_should_drop_sgid(struct user_namespace *mnt_userns,
3244 			     const struct inode *inode);
3245 
3246 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
insert_inode_hash(struct inode * inode)3247 static inline void insert_inode_hash(struct inode *inode)
3248 {
3249 	__insert_inode_hash(inode, inode->i_ino);
3250 }
3251 
3252 extern void __remove_inode_hash(struct inode *);
remove_inode_hash(struct inode * inode)3253 static inline void remove_inode_hash(struct inode *inode)
3254 {
3255 	if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
3256 		__remove_inode_hash(inode);
3257 }
3258 
3259 extern void inode_sb_list_add(struct inode *inode);
3260 
3261 extern int sb_set_blocksize(struct super_block *, int);
3262 extern int sb_min_blocksize(struct super_block *, int);
3263 
3264 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
3265 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
3266 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
3267 extern int generic_write_check_limits(struct file *file, loff_t pos,
3268 		loff_t *count);
3269 extern int generic_file_rw_checks(struct file *file_in, struct file *file_out);
3270 ssize_t filemap_read(struct kiocb *iocb, struct iov_iter *to,
3271 		ssize_t already_read);
3272 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
3273 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
3274 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
3275 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
3276 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
3277 
3278 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
3279 		rwf_t flags);
3280 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
3281 		rwf_t flags);
3282 ssize_t vfs_iocb_iter_read(struct file *file, struct kiocb *iocb,
3283 			   struct iov_iter *iter);
3284 ssize_t vfs_iocb_iter_write(struct file *file, struct kiocb *iocb,
3285 			    struct iov_iter *iter);
3286 
3287 /* fs/splice.c */
3288 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
3289 		struct pipe_inode_info *, size_t, unsigned int);
3290 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
3291 		struct file *, loff_t *, size_t, unsigned int);
3292 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
3293 		struct file *out, loff_t *, size_t len, unsigned int flags);
3294 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
3295 		loff_t *opos, size_t len, unsigned int flags);
3296 
3297 
3298 extern void
3299 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
3300 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
3301 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
3302 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
3303 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
3304 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
3305 		int whence, loff_t maxsize, loff_t eof);
3306 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
3307 		int whence, loff_t size);
3308 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
3309 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
3310 extern int generic_file_open(struct inode * inode, struct file * filp);
3311 extern int nonseekable_open(struct inode * inode, struct file * filp);
3312 extern int stream_open(struct inode * inode, struct file * filp);
3313 
3314 #ifdef CONFIG_BLOCK
3315 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
3316 			    loff_t file_offset);
3317 
3318 enum {
3319 	/* need locking between buffered and direct access */
3320 	DIO_LOCKING	= 0x01,
3321 
3322 	/* filesystem does not support filling holes */
3323 	DIO_SKIP_HOLES	= 0x02,
3324 };
3325 
3326 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
3327 			     struct block_device *bdev, struct iov_iter *iter,
3328 			     get_block_t get_block,
3329 			     dio_iodone_t end_io, dio_submit_t submit_io,
3330 			     int flags);
3331 
blockdev_direct_IO(struct kiocb * iocb,struct inode * inode,struct iov_iter * iter,get_block_t get_block)3332 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
3333 					 struct inode *inode,
3334 					 struct iov_iter *iter,
3335 					 get_block_t get_block)
3336 {
3337 	return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
3338 			get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
3339 }
3340 #endif
3341 
3342 void inode_dio_wait(struct inode *inode);
3343 
3344 /**
3345  * inode_dio_begin - signal start of a direct I/O requests
3346  * @inode: inode the direct I/O happens on
3347  *
3348  * This is called once we've finished processing a direct I/O request,
3349  * and is used to wake up callers waiting for direct I/O to be quiesced.
3350  */
inode_dio_begin(struct inode * inode)3351 static inline void inode_dio_begin(struct inode *inode)
3352 {
3353 	atomic_inc(&inode->i_dio_count);
3354 }
3355 
3356 /**
3357  * inode_dio_end - signal finish of a direct I/O requests
3358  * @inode: inode the direct I/O happens on
3359  *
3360  * This is called once we've finished processing a direct I/O request,
3361  * and is used to wake up callers waiting for direct I/O to be quiesced.
3362  */
inode_dio_end(struct inode * inode)3363 static inline void inode_dio_end(struct inode *inode)
3364 {
3365 	if (atomic_dec_and_test(&inode->i_dio_count))
3366 		wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
3367 }
3368 
3369 /*
3370  * Warn about a page cache invalidation failure diring a direct I/O write.
3371  */
3372 void dio_warn_stale_pagecache(struct file *filp);
3373 
3374 extern void inode_set_flags(struct inode *inode, unsigned int flags,
3375 			    unsigned int mask);
3376 
3377 extern const struct file_operations generic_ro_fops;
3378 
3379 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3380 
3381 extern int readlink_copy(char __user *, int, const char *);
3382 extern int page_readlink(struct dentry *, char __user *, int);
3383 extern const char *page_get_link(struct dentry *, struct inode *,
3384 				 struct delayed_call *);
3385 extern void page_put_link(void *);
3386 extern int __page_symlink(struct inode *inode, const char *symname, int len,
3387 		int nofs);
3388 extern int page_symlink(struct inode *inode, const char *symname, int len);
3389 extern const struct inode_operations page_symlink_inode_operations;
3390 extern void kfree_link(void *);
3391 void generic_fillattr(struct user_namespace *, struct inode *, struct kstat *);
3392 void generic_fill_statx_attr(struct inode *inode, struct kstat *stat);
3393 extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
3394 extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
3395 void __inode_add_bytes(struct inode *inode, loff_t bytes);
3396 void inode_add_bytes(struct inode *inode, loff_t bytes);
3397 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
3398 void inode_sub_bytes(struct inode *inode, loff_t bytes);
__inode_get_bytes(struct inode * inode)3399 static inline loff_t __inode_get_bytes(struct inode *inode)
3400 {
3401 	return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes;
3402 }
3403 loff_t inode_get_bytes(struct inode *inode);
3404 void inode_set_bytes(struct inode *inode, loff_t bytes);
3405 const char *simple_get_link(struct dentry *, struct inode *,
3406 			    struct delayed_call *);
3407 extern const struct inode_operations simple_symlink_inode_operations;
3408 
3409 extern int iterate_dir(struct file *, struct dir_context *);
3410 
3411 int vfs_fstatat(int dfd, const char __user *filename, struct kstat *stat,
3412 		int flags);
3413 int vfs_fstat(int fd, struct kstat *stat);
3414 
vfs_stat(const char __user * filename,struct kstat * stat)3415 static inline int vfs_stat(const char __user *filename, struct kstat *stat)
3416 {
3417 	return vfs_fstatat(AT_FDCWD, filename, stat, 0);
3418 }
vfs_lstat(const char __user * name,struct kstat * stat)3419 static inline int vfs_lstat(const char __user *name, struct kstat *stat)
3420 {
3421 	return vfs_fstatat(AT_FDCWD, name, stat, AT_SYMLINK_NOFOLLOW);
3422 }
3423 
3424 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
3425 extern int vfs_readlink(struct dentry *, char __user *, int);
3426 
3427 extern struct file_system_type *get_filesystem(struct file_system_type *fs);
3428 extern void put_filesystem(struct file_system_type *fs);
3429 extern struct file_system_type *get_fs_type(const char *name);
3430 extern struct super_block *get_super(struct block_device *);
3431 extern struct super_block *get_active_super(struct block_device *bdev);
3432 extern void drop_super(struct super_block *sb);
3433 extern void drop_super_exclusive(struct super_block *sb);
3434 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
3435 extern void iterate_supers_type(struct file_system_type *,
3436 			        void (*)(struct super_block *, void *), void *);
3437 
3438 extern int dcache_dir_open(struct inode *, struct file *);
3439 extern int dcache_dir_close(struct inode *, struct file *);
3440 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
3441 extern int dcache_readdir(struct file *, struct dir_context *);
3442 extern int simple_setattr(struct user_namespace *, struct dentry *,
3443 			  struct iattr *);
3444 extern int simple_getattr(struct user_namespace *, const struct path *,
3445 			  struct kstat *, u32, unsigned int);
3446 extern int simple_statfs(struct dentry *, struct kstatfs *);
3447 extern int simple_open(struct inode *inode, struct file *file);
3448 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3449 extern int simple_unlink(struct inode *, struct dentry *);
3450 extern int simple_rmdir(struct inode *, struct dentry *);
3451 extern int simple_rename(struct user_namespace *, struct inode *,
3452 			 struct dentry *, struct inode *, struct dentry *,
3453 			 unsigned int);
3454 extern void simple_recursive_removal(struct dentry *,
3455                               void (*callback)(struct dentry *));
3456 extern int noop_fsync(struct file *, loff_t, loff_t, int);
3457 extern void noop_invalidatepage(struct page *page, unsigned int offset,
3458 		unsigned int length);
3459 extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter);
3460 extern int simple_empty(struct dentry *);
3461 extern int simple_write_begin(struct file *file, struct address_space *mapping,
3462 			loff_t pos, unsigned len, unsigned flags,
3463 			struct page **pagep, void **fsdata);
3464 extern const struct address_space_operations ram_aops;
3465 extern int always_delete_dentry(const struct dentry *);
3466 extern struct inode *alloc_anon_inode(struct super_block *);
3467 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
3468 extern const struct dentry_operations simple_dentry_operations;
3469 
3470 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3471 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3472 extern const struct file_operations simple_dir_operations;
3473 extern const struct inode_operations simple_dir_inode_operations;
3474 extern void make_empty_dir_inode(struct inode *inode);
3475 extern bool is_empty_dir_inode(struct inode *inode);
3476 struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
3477 struct dentry *d_alloc_name(struct dentry *, const char *);
3478 extern int simple_fill_super(struct super_block *, unsigned long,
3479 			     const struct tree_descr *);
3480 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3481 extern void simple_release_fs(struct vfsmount **mount, int *count);
3482 
3483 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3484 			loff_t *ppos, const void *from, size_t available);
3485 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3486 		const void __user *from, size_t count);
3487 
3488 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3489 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3490 
3491 extern int generic_check_addressable(unsigned, u64);
3492 
3493 extern void generic_set_encrypted_ci_d_ops(struct dentry *dentry);
3494 
3495 #ifdef CONFIG_MIGRATION
3496 extern int buffer_migrate_page(struct address_space *,
3497 				struct page *, struct page *,
3498 				enum migrate_mode);
3499 extern int buffer_migrate_page_norefs(struct address_space *,
3500 				struct page *, struct page *,
3501 				enum migrate_mode);
3502 #else
3503 #define buffer_migrate_page NULL
3504 #define buffer_migrate_page_norefs NULL
3505 #endif
3506 
3507 int may_setattr(struct user_namespace *mnt_userns, struct inode *inode,
3508 		unsigned int ia_valid);
3509 int setattr_prepare(struct user_namespace *, struct dentry *, struct iattr *);
3510 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3511 void setattr_copy(struct user_namespace *, struct inode *inode,
3512 		  const struct iattr *attr);
3513 
3514 extern int file_update_time(struct file *file);
3515 
vma_is_dax(const struct vm_area_struct * vma)3516 static inline bool vma_is_dax(const struct vm_area_struct *vma)
3517 {
3518 	return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3519 }
3520 
vma_is_fsdax(struct vm_area_struct * vma)3521 static inline bool vma_is_fsdax(struct vm_area_struct *vma)
3522 {
3523 	struct inode *inode;
3524 
3525 	if (!IS_ENABLED(CONFIG_FS_DAX) || !vma->vm_file)
3526 		return false;
3527 	if (!vma_is_dax(vma))
3528 		return false;
3529 	inode = file_inode(vma->vm_file);
3530 	if (S_ISCHR(inode->i_mode))
3531 		return false; /* device-dax */
3532 	return true;
3533 }
3534 
iocb_flags(struct file * file)3535 static inline int iocb_flags(struct file *file)
3536 {
3537 	int res = 0;
3538 	if (file->f_flags & O_APPEND)
3539 		res |= IOCB_APPEND;
3540 	if (file->f_flags & O_DIRECT)
3541 		res |= IOCB_DIRECT;
3542 	if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3543 		res |= IOCB_DSYNC;
3544 	if (file->f_flags & __O_SYNC)
3545 		res |= IOCB_SYNC;
3546 	return res;
3547 }
3548 
kiocb_set_rw_flags(struct kiocb * ki,rwf_t flags)3549 static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
3550 {
3551 	int kiocb_flags = 0;
3552 
3553 	/* make sure there's no overlap between RWF and private IOCB flags */
3554 	BUILD_BUG_ON((__force int) RWF_SUPPORTED & IOCB_EVENTFD);
3555 
3556 	if (!flags)
3557 		return 0;
3558 	if (unlikely(flags & ~RWF_SUPPORTED))
3559 		return -EOPNOTSUPP;
3560 
3561 	if (flags & RWF_NOWAIT) {
3562 		if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
3563 			return -EOPNOTSUPP;
3564 		kiocb_flags |= IOCB_NOIO;
3565 	}
3566 	kiocb_flags |= (__force int) (flags & RWF_SUPPORTED);
3567 	if (flags & RWF_SYNC)
3568 		kiocb_flags |= IOCB_DSYNC;
3569 
3570 	ki->ki_flags |= kiocb_flags;
3571 	return 0;
3572 }
3573 
iocb_to_rw_flags(int ifl,int iocb_mask)3574 static inline rwf_t iocb_to_rw_flags(int ifl, int iocb_mask)
3575 {
3576 	return ifl & iocb_mask;
3577 }
3578 
parent_ino(struct dentry * dentry)3579 static inline ino_t parent_ino(struct dentry *dentry)
3580 {
3581 	ino_t res;
3582 
3583 	/*
3584 	 * Don't strictly need d_lock here? If the parent ino could change
3585 	 * then surely we'd have a deeper race in the caller?
3586 	 */
3587 	spin_lock(&dentry->d_lock);
3588 	res = dentry->d_parent->d_inode->i_ino;
3589 	spin_unlock(&dentry->d_lock);
3590 	return res;
3591 }
3592 
3593 /* Transaction based IO helpers */
3594 
3595 /*
3596  * An argresp is stored in an allocated page and holds the
3597  * size of the argument or response, along with its content
3598  */
3599 struct simple_transaction_argresp {
3600 	ssize_t size;
3601 	char data[];
3602 };
3603 
3604 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3605 
3606 char *simple_transaction_get(struct file *file, const char __user *buf,
3607 				size_t size);
3608 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3609 				size_t size, loff_t *pos);
3610 int simple_transaction_release(struct inode *inode, struct file *file);
3611 
3612 void simple_transaction_set(struct file *file, size_t n);
3613 
3614 /*
3615  * simple attribute files
3616  *
3617  * These attributes behave similar to those in sysfs:
3618  *
3619  * Writing to an attribute immediately sets a value, an open file can be
3620  * written to multiple times.
3621  *
3622  * Reading from an attribute creates a buffer from the value that might get
3623  * read with multiple read calls. When the attribute has been read
3624  * completely, no further read calls are possible until the file is opened
3625  * again.
3626  *
3627  * All attributes contain a text representation of a numeric value
3628  * that are accessed with the get() and set() functions.
3629  */
3630 #define DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, __is_signed)	\
3631 static int __fops ## _open(struct inode *inode, struct file *file)	\
3632 {									\
3633 	__simple_attr_check_format(__fmt, 0ull);			\
3634 	return simple_attr_open(inode, file, __get, __set, __fmt);	\
3635 }									\
3636 static const struct file_operations __fops = {				\
3637 	.owner	 = THIS_MODULE,						\
3638 	.open	 = __fops ## _open,					\
3639 	.release = simple_attr_release,					\
3640 	.read	 = simple_attr_read,					\
3641 	.write	 = (__is_signed) ? simple_attr_write_signed : simple_attr_write,	\
3642 	.llseek	 = generic_file_llseek,					\
3643 }
3644 
3645 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt)		\
3646 	DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, false)
3647 
3648 #define DEFINE_SIMPLE_ATTRIBUTE_SIGNED(__fops, __get, __set, __fmt)	\
3649 	DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, true)
3650 
3651 static inline __printf(1, 2)
__simple_attr_check_format(const char * fmt,...)3652 void __simple_attr_check_format(const char *fmt, ...)
3653 {
3654 	/* don't do anything, just let the compiler check the arguments; */
3655 }
3656 
3657 int simple_attr_open(struct inode *inode, struct file *file,
3658 		     int (*get)(void *, u64 *), int (*set)(void *, u64),
3659 		     const char *fmt);
3660 int simple_attr_release(struct inode *inode, struct file *file);
3661 ssize_t simple_attr_read(struct file *file, char __user *buf,
3662 			 size_t len, loff_t *ppos);
3663 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3664 			  size_t len, loff_t *ppos);
3665 ssize_t simple_attr_write_signed(struct file *file, const char __user *buf,
3666 				 size_t len, loff_t *ppos);
3667 
3668 struct ctl_table;
3669 int proc_nr_files(struct ctl_table *table, int write,
3670 		  void *buffer, size_t *lenp, loff_t *ppos);
3671 int proc_nr_dentry(struct ctl_table *table, int write,
3672 		  void *buffer, size_t *lenp, loff_t *ppos);
3673 int proc_nr_inodes(struct ctl_table *table, int write,
3674 		   void *buffer, size_t *lenp, loff_t *ppos);
3675 int __init list_bdev_fs_names(char *buf, size_t size);
3676 
3677 #define __FMODE_EXEC		((__force int) FMODE_EXEC)
3678 #define __FMODE_NONOTIFY	((__force int) FMODE_NONOTIFY)
3679 
3680 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3681 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3682 					    (flag & __FMODE_NONOTIFY)))
3683 
is_sxid(umode_t mode)3684 static inline bool is_sxid(umode_t mode)
3685 {
3686 	return mode & (S_ISUID | S_ISGID);
3687 }
3688 
check_sticky(struct user_namespace * mnt_userns,struct inode * dir,struct inode * inode)3689 static inline int check_sticky(struct user_namespace *mnt_userns,
3690 			       struct inode *dir, struct inode *inode)
3691 {
3692 	if (!(dir->i_mode & S_ISVTX))
3693 		return 0;
3694 
3695 	return __check_sticky(mnt_userns, dir, inode);
3696 }
3697 
inode_has_no_xattr(struct inode * inode)3698 static inline void inode_has_no_xattr(struct inode *inode)
3699 {
3700 	if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
3701 		inode->i_flags |= S_NOSEC;
3702 }
3703 
is_root_inode(struct inode * inode)3704 static inline bool is_root_inode(struct inode *inode)
3705 {
3706 	return inode == inode->i_sb->s_root->d_inode;
3707 }
3708 
dir_emit(struct dir_context * ctx,const char * name,int namelen,u64 ino,unsigned type)3709 static inline bool dir_emit(struct dir_context *ctx,
3710 			    const char *name, int namelen,
3711 			    u64 ino, unsigned type)
3712 {
3713 	return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3714 }
dir_emit_dot(struct file * file,struct dir_context * ctx)3715 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3716 {
3717 	return ctx->actor(ctx, ".", 1, ctx->pos,
3718 			  file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3719 }
dir_emit_dotdot(struct file * file,struct dir_context * ctx)3720 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3721 {
3722 	return ctx->actor(ctx, "..", 2, ctx->pos,
3723 			  parent_ino(file->f_path.dentry), DT_DIR) == 0;
3724 }
dir_emit_dots(struct file * file,struct dir_context * ctx)3725 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3726 {
3727 	if (ctx->pos == 0) {
3728 		if (!dir_emit_dot(file, ctx))
3729 			return false;
3730 		ctx->pos = 1;
3731 	}
3732 	if (ctx->pos == 1) {
3733 		if (!dir_emit_dotdot(file, ctx))
3734 			return false;
3735 		ctx->pos = 2;
3736 	}
3737 	return true;
3738 }
dir_relax(struct inode * inode)3739 static inline bool dir_relax(struct inode *inode)
3740 {
3741 	inode_unlock(inode);
3742 	inode_lock(inode);
3743 	return !IS_DEADDIR(inode);
3744 }
3745 
dir_relax_shared(struct inode * inode)3746 static inline bool dir_relax_shared(struct inode *inode)
3747 {
3748 	inode_unlock_shared(inode);
3749 	inode_lock_shared(inode);
3750 	return !IS_DEADDIR(inode);
3751 }
3752 
3753 extern bool path_noexec(const struct path *path);
3754 extern void inode_nohighmem(struct inode *inode);
3755 
3756 /* mm/fadvise.c */
3757 extern int vfs_fadvise(struct file *file, loff_t offset, loff_t len,
3758 		       int advice);
3759 extern int generic_fadvise(struct file *file, loff_t offset, loff_t len,
3760 			   int advice);
3761 
3762 /*
3763  * Flush file data before changing attributes.  Caller must hold any locks
3764  * required to prevent further writes to this file until we're done setting
3765  * flags.
3766  */
inode_drain_writes(struct inode * inode)3767 static inline int inode_drain_writes(struct inode *inode)
3768 {
3769 	inode_dio_wait(inode);
3770 	return filemap_write_and_wait(inode->i_mapping);
3771 }
3772 
3773 #endif /* _LINUX_FS_H */
3774