1 /* SPDX-License-Identifier: GPL-2.0 */
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>
45 #include <asm/byteorder.h>
46 #include <uapi/linux/fs.h>
48 struct backing_dev_info;
51 struct export_operations;
52 struct fiemap_extent_info;
57 struct pipe_inode_info;
58 struct poll_table_struct;
60 struct vm_area_struct;
63 struct swap_info_struct;
65 struct workqueue_struct;
68 struct fscrypt_operations;
70 struct fsverity_operations;
72 struct fs_parameter_spec;
74 extern void __init inode_init(void);
75 extern void __init inode_init_early(void);
76 extern void __init files_init(void);
77 extern void __init files_maxfiles_init(void);
79 extern struct files_stat_struct files_stat;
80 extern unsigned long get_max_files(void);
81 extern unsigned int sysctl_nr_open;
82 extern struct inodes_stat_t inodes_stat;
83 extern int leases_enable, lease_break_time;
84 extern int sysctl_protected_symlinks;
85 extern int sysctl_protected_hardlinks;
86 extern int sysctl_protected_fifos;
87 extern int sysctl_protected_regular;
89 typedef __kernel_rwf_t rwf_t;
92 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
93 struct buffer_head *bh_result, int create);
94 typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
95 ssize_t bytes, void *private);
97 #define MAY_EXEC 0x00000001
98 #define MAY_WRITE 0x00000002
99 #define MAY_READ 0x00000004
100 #define MAY_APPEND 0x00000008
101 #define MAY_ACCESS 0x00000010
102 #define MAY_OPEN 0x00000020
103 #define MAY_CHDIR 0x00000040
104 /* called from RCU mode, don't block */
105 #define MAY_NOT_BLOCK 0x00000080
108 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
109 * to O_WRONLY and O_RDWR via the strange trick in do_dentry_open()
112 /* file is open for reading */
113 #define FMODE_READ ((__force fmode_t)0x1)
114 /* file is open for writing */
115 #define FMODE_WRITE ((__force fmode_t)0x2)
116 /* file is seekable */
117 #define FMODE_LSEEK ((__force fmode_t)0x4)
118 /* file can be accessed using pread */
119 #define FMODE_PREAD ((__force fmode_t)0x8)
120 /* file can be accessed using pwrite */
121 #define FMODE_PWRITE ((__force fmode_t)0x10)
122 /* File is opened for execution with sys_execve / sys_uselib */
123 #define FMODE_EXEC ((__force fmode_t)0x20)
124 /* File is opened with O_NDELAY (only set for block devices) */
125 #define FMODE_NDELAY ((__force fmode_t)0x40)
126 /* File is opened with O_EXCL (only set for block devices) */
127 #define FMODE_EXCL ((__force fmode_t)0x80)
128 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
129 (specialy hack for floppy.c) */
130 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
131 /* 32bit hashes as llseek() offset (for directories) */
132 #define FMODE_32BITHASH ((__force fmode_t)0x200)
133 /* 64bit hashes as llseek() offset (for directories) */
134 #define FMODE_64BITHASH ((__force fmode_t)0x400)
137 * Don't update ctime and mtime.
139 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
140 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
142 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
144 /* Expect random access pattern */
145 #define FMODE_RANDOM ((__force fmode_t)0x1000)
147 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
148 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
150 /* File is opened with O_PATH; almost nothing can be done with it */
151 #define FMODE_PATH ((__force fmode_t)0x4000)
153 /* File needs atomic accesses to f_pos */
154 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
155 /* Write access to underlying fs */
156 #define FMODE_WRITER ((__force fmode_t)0x10000)
157 /* Has read method(s) */
158 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
159 /* Has write method(s) */
160 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
162 #define FMODE_OPENED ((__force fmode_t)0x80000)
163 #define FMODE_CREATED ((__force fmode_t)0x100000)
165 /* File is stream-like */
166 #define FMODE_STREAM ((__force fmode_t)0x200000)
168 /* File was opened by fanotify and shouldn't generate fanotify events */
169 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
171 /* File is capable of returning -EAGAIN if I/O will block */
172 #define FMODE_NOWAIT ((__force fmode_t)0x8000000)
174 /* File represents mount that needs unmounting */
175 #define FMODE_NEED_UNMOUNT ((__force fmode_t)0x10000000)
177 /* File does not contribute to nr_files count */
178 #define FMODE_NOACCOUNT ((__force fmode_t)0x20000000)
180 /* File supports async buffered reads */
181 #define FMODE_BUF_RASYNC ((__force fmode_t)0x40000000)
184 * Attribute flags. These should be or-ed together to figure out what
187 #define ATTR_MODE (1 << 0)
188 #define ATTR_UID (1 << 1)
189 #define ATTR_GID (1 << 2)
190 #define ATTR_SIZE (1 << 3)
191 #define ATTR_ATIME (1 << 4)
192 #define ATTR_MTIME (1 << 5)
193 #define ATTR_CTIME (1 << 6)
194 #define ATTR_ATIME_SET (1 << 7)
195 #define ATTR_MTIME_SET (1 << 8)
196 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
197 #define ATTR_KILL_SUID (1 << 11)
198 #define ATTR_KILL_SGID (1 << 12)
199 #define ATTR_FILE (1 << 13)
200 #define ATTR_KILL_PRIV (1 << 14)
201 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
202 #define ATTR_TIMES_SET (1 << 16)
203 #define ATTR_TOUCH (1 << 17)
206 * Whiteout is represented by a char device. The following constants define the
207 * mode and device number to use.
209 #define WHITEOUT_MODE 0
210 #define WHITEOUT_DEV 0
213 * This is the Inode Attributes structure, used for notify_change(). It
214 * uses the above definitions as flags, to know which values have changed.
215 * Also, in this manner, a Filesystem can look at only the values it cares
216 * about. Basically, these are the attributes that the VFS layer can
217 * request to change from the FS layer.
219 * Derek Atkins <warlord@MIT.EDU> 94-10-20
222 unsigned int ia_valid;
227 struct timespec64 ia_atime;
228 struct timespec64 ia_mtime;
229 struct timespec64 ia_ctime;
232 * Not an attribute, but an auxiliary info for filesystems wanting to
233 * implement an ftruncate() like method. NOTE: filesystem should
234 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
236 struct file *ia_file;
240 * Includes for diskquotas.
242 #include <linux/quota.h>
245 * Maximum number of layers of fs stack. Needs to be limited to
246 * prevent kernel stack overflow
248 #define FILESYSTEM_MAX_STACK_DEPTH 2
251 * enum positive_aop_returns - aop return codes with specific semantics
253 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
254 * completed, that the page is still locked, and
255 * should be considered active. The VM uses this hint
256 * to return the page to the active list -- it won't
257 * be a candidate for writeback again in the near
258 * future. Other callers must be careful to unlock
259 * the page if they get this return. Returned by
262 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
263 * unlocked it and the page might have been truncated.
264 * The caller should back up to acquiring a new page and
265 * trying again. The aop will be taking reasonable
266 * precautions not to livelock. If the caller held a page
267 * reference, it should drop it before retrying. Returned
270 * address_space_operation functions return these large constants to indicate
271 * special semantics to the caller. These are much larger than the bytes in a
272 * page to allow for functions that return the number of bytes operated on in a
276 enum positive_aop_returns {
277 AOP_WRITEPAGE_ACTIVATE = 0x80000,
278 AOP_TRUNCATED_PAGE = 0x80001,
281 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
282 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
283 * helper code (eg buffer layer)
284 * to clear GFP_FS from alloc */
287 * oh the beauties of C type declarations.
290 struct address_space;
291 struct writeback_control;
292 struct readahead_control;
295 * Write life time hint values.
296 * Stored in struct inode as u8.
299 WRITE_LIFE_NOT_SET = 0,
300 WRITE_LIFE_NONE = RWH_WRITE_LIFE_NONE,
301 WRITE_LIFE_SHORT = RWH_WRITE_LIFE_SHORT,
302 WRITE_LIFE_MEDIUM = RWH_WRITE_LIFE_MEDIUM,
303 WRITE_LIFE_LONG = RWH_WRITE_LIFE_LONG,
304 WRITE_LIFE_EXTREME = RWH_WRITE_LIFE_EXTREME,
307 /* Match RWF_* bits to IOCB bits */
308 #define IOCB_HIPRI (__force int) RWF_HIPRI
309 #define IOCB_DSYNC (__force int) RWF_DSYNC
310 #define IOCB_SYNC (__force int) RWF_SYNC
311 #define IOCB_NOWAIT (__force int) RWF_NOWAIT
312 #define IOCB_APPEND (__force int) RWF_APPEND
314 /* non-RWF related bits - start at 16 */
315 #define IOCB_EVENTFD (1 << 16)
316 #define IOCB_DIRECT (1 << 17)
317 #define IOCB_WRITE (1 << 18)
318 /* iocb->ki_waitq is valid */
319 #define IOCB_WAITQ (1 << 19)
320 #define IOCB_NOIO (1 << 20)
323 struct file *ki_filp;
325 /* The 'ki_filp' pointer is shared in a union for aio */
326 randomized_struct_fields_start
329 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
333 u16 ki_ioprio; /* See linux/ioprio.h */
335 unsigned int ki_cookie; /* for ->iopoll */
336 struct wait_page_queue *ki_waitq; /* for async buffered IO */
339 randomized_struct_fields_end
342 static inline bool is_sync_kiocb(struct kiocb *kiocb)
344 return kiocb->ki_complete == NULL;
348 * "descriptor" for what we're up to with a read.
349 * This allows us to use the same read code yet
350 * have multiple different users of the data that
351 * we read from a file.
353 * The simplest case just copies the data to user
366 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
367 unsigned long, unsigned long);
369 struct address_space_operations {
370 int (*writepage)(struct page *page, struct writeback_control *wbc);
371 int (*readpage)(struct file *, struct page *);
373 /* Write back some dirty pages from this mapping. */
374 int (*writepages)(struct address_space *, struct writeback_control *);
376 /* Set a page dirty. Return true if this dirtied it */
377 int (*set_page_dirty)(struct page *page);
380 * Reads in the requested pages. Unlike ->readpage(), this is
381 * PURELY used for read-ahead!.
383 int (*readpages)(struct file *filp, struct address_space *mapping,
384 struct list_head *pages, unsigned nr_pages);
385 void (*readahead)(struct readahead_control *);
387 int (*write_begin)(struct file *, struct address_space *mapping,
388 loff_t pos, unsigned len, unsigned flags,
389 struct page **pagep, void **fsdata);
390 int (*write_end)(struct file *, struct address_space *mapping,
391 loff_t pos, unsigned len, unsigned copied,
392 struct page *page, void *fsdata);
394 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
395 sector_t (*bmap)(struct address_space *, sector_t);
396 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
397 int (*releasepage) (struct page *, gfp_t);
398 void (*freepage)(struct page *);
399 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
401 * migrate the contents of a page to the specified target. If
402 * migrate_mode is MIGRATE_ASYNC, it must not block.
404 int (*migratepage) (struct address_space *,
405 struct page *, struct page *, enum migrate_mode);
406 bool (*isolate_page)(struct page *, isolate_mode_t);
407 void (*putback_page)(struct page *);
408 int (*launder_page) (struct page *);
409 int (*is_partially_uptodate) (struct page *, unsigned long,
411 void (*is_dirty_writeback) (struct page *, bool *, bool *);
412 int (*error_remove_page)(struct address_space *, struct page *);
414 /* swapfile support */
415 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
417 void (*swap_deactivate)(struct file *file);
420 extern const struct address_space_operations empty_aops;
423 * pagecache_write_begin/pagecache_write_end must be used by general code
424 * to write into the pagecache.
426 int pagecache_write_begin(struct file *, struct address_space *mapping,
427 loff_t pos, unsigned len, unsigned flags,
428 struct page **pagep, void **fsdata);
430 int pagecache_write_end(struct file *, struct address_space *mapping,
431 loff_t pos, unsigned len, unsigned copied,
432 struct page *page, void *fsdata);
435 * struct address_space - Contents of a cacheable, mappable object.
436 * @host: Owner, either the inode or the block_device.
437 * @i_pages: Cached pages.
438 * @gfp_mask: Memory allocation flags to use for allocating pages.
439 * @i_mmap_writable: Number of VM_SHARED mappings.
440 * @nr_thps: Number of THPs in the pagecache (non-shmem only).
441 * @i_mmap: Tree of private and shared mappings.
442 * @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable.
443 * @nrpages: Number of page entries, protected by the i_pages lock.
444 * @nrexceptional: Shadow or DAX entries, protected by the i_pages lock.
445 * @writeback_index: Writeback starts here.
447 * @flags: Error bits and flags (AS_*).
448 * @wb_err: The most recent error which has occurred.
449 * @private_lock: For use by the owner of the address_space.
450 * @private_list: For use by the owner of the address_space.
451 * @private_data: For use by the owner of the address_space.
453 struct address_space {
455 struct xarray i_pages;
457 atomic_t i_mmap_writable;
458 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
459 /* number of thp, only for non-shmem files */
462 struct rb_root_cached i_mmap;
463 struct rw_semaphore i_mmap_rwsem;
464 unsigned long nrpages;
465 unsigned long nrexceptional;
466 pgoff_t writeback_index;
467 const struct address_space_operations *a_ops;
470 spinlock_t private_lock;
471 struct list_head private_list;
473 } __attribute__((aligned(sizeof(long)))) __randomize_layout;
475 * On most architectures that alignment is already the case; but
476 * must be enforced here for CRIS, to let the least significant bit
477 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
480 /* XArray tags, for tagging dirty and writeback pages in the pagecache. */
481 #define PAGECACHE_TAG_DIRTY XA_MARK_0
482 #define PAGECACHE_TAG_WRITEBACK XA_MARK_1
483 #define PAGECACHE_TAG_TOWRITE XA_MARK_2
486 * Returns true if any of the pages in the mapping are marked with the tag.
488 static inline bool mapping_tagged(struct address_space *mapping, xa_mark_t tag)
490 return xa_marked(&mapping->i_pages, tag);
493 static inline void i_mmap_lock_write(struct address_space *mapping)
495 down_write(&mapping->i_mmap_rwsem);
498 static inline int i_mmap_trylock_write(struct address_space *mapping)
500 return down_write_trylock(&mapping->i_mmap_rwsem);
503 static inline void i_mmap_unlock_write(struct address_space *mapping)
505 up_write(&mapping->i_mmap_rwsem);
508 static inline void i_mmap_lock_read(struct address_space *mapping)
510 down_read(&mapping->i_mmap_rwsem);
513 static inline void i_mmap_unlock_read(struct address_space *mapping)
515 up_read(&mapping->i_mmap_rwsem);
518 static inline void i_mmap_assert_locked(struct address_space *mapping)
520 lockdep_assert_held(&mapping->i_mmap_rwsem);
523 static inline void i_mmap_assert_write_locked(struct address_space *mapping)
525 lockdep_assert_held_write(&mapping->i_mmap_rwsem);
529 * Might pages of this file be mapped into userspace?
531 static inline int mapping_mapped(struct address_space *mapping)
533 return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root);
537 * Might pages of this file have been modified in userspace?
538 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap
539 * marks vma as VM_SHARED if it is shared, and the file was opened for
540 * writing i.e. vma may be mprotected writable even if now readonly.
542 * If i_mmap_writable is negative, no new writable mappings are allowed. You
543 * can only deny writable mappings, if none exists right now.
545 static inline int mapping_writably_mapped(struct address_space *mapping)
547 return atomic_read(&mapping->i_mmap_writable) > 0;
550 static inline int mapping_map_writable(struct address_space *mapping)
552 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
556 static inline void mapping_unmap_writable(struct address_space *mapping)
558 atomic_dec(&mapping->i_mmap_writable);
561 static inline int mapping_deny_writable(struct address_space *mapping)
563 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
567 static inline void mapping_allow_writable(struct address_space *mapping)
569 atomic_inc(&mapping->i_mmap_writable);
573 * Use sequence counter to get consistent i_size on 32-bit processors.
575 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
576 #include <linux/seqlock.h>
577 #define __NEED_I_SIZE_ORDERED
578 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
580 #define i_size_ordered_init(inode) do { } while (0)
584 #define ACL_NOT_CACHED ((void *)(-1))
585 #define ACL_DONT_CACHE ((void *)(-3))
587 static inline struct posix_acl *
588 uncached_acl_sentinel(struct task_struct *task)
590 return (void *)task + 1;
594 is_uncached_acl(struct posix_acl *acl)
596 return (long)acl & 1;
599 #define IOP_FASTPERM 0x0001
600 #define IOP_LOOKUP 0x0002
601 #define IOP_NOFOLLOW 0x0004
602 #define IOP_XATTR 0x0008
603 #define IOP_DEFAULT_READLINK 0x0010
605 struct fsnotify_mark_connector;
608 * Keep mostly read-only and often accessed (especially for
609 * the RCU path lookup and 'stat' data) fields at the beginning
610 * of the 'struct inode'
614 unsigned short i_opflags;
617 unsigned int i_flags;
619 #ifdef CONFIG_FS_POSIX_ACL
620 struct posix_acl *i_acl;
621 struct posix_acl *i_default_acl;
624 const struct inode_operations *i_op;
625 struct super_block *i_sb;
626 struct address_space *i_mapping;
628 #ifdef CONFIG_SECURITY
632 /* Stat data, not accessed from path walking */
635 * Filesystems may only read i_nlink directly. They shall use the
636 * following functions for modification:
638 * (set|clear|inc|drop)_nlink
639 * inode_(inc|dec)_link_count
642 const unsigned int i_nlink;
643 unsigned int __i_nlink;
647 struct timespec64 i_atime;
648 struct timespec64 i_mtime;
649 struct timespec64 i_ctime;
650 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
651 unsigned short i_bytes;
656 #ifdef __NEED_I_SIZE_ORDERED
657 seqcount_t i_size_seqcount;
661 unsigned long i_state;
662 struct rw_semaphore i_rwsem;
664 unsigned long dirtied_when; /* jiffies of first dirtying */
665 unsigned long dirtied_time_when;
667 struct hlist_node i_hash;
668 struct list_head i_io_list; /* backing dev IO list */
669 #ifdef CONFIG_CGROUP_WRITEBACK
670 struct bdi_writeback *i_wb; /* the associated cgroup wb */
672 /* foreign inode detection, see wbc_detach_inode() */
674 u16 i_wb_frn_avg_time;
675 u16 i_wb_frn_history;
677 struct list_head i_lru; /* inode LRU list */
678 struct list_head i_sb_list;
679 struct list_head i_wb_list; /* backing dev writeback list */
681 struct hlist_head i_dentry;
682 struct rcu_head i_rcu;
684 atomic64_t i_version;
685 atomic64_t i_sequence; /* see futex */
687 atomic_t i_dio_count;
688 atomic_t i_writecount;
689 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
690 atomic_t i_readcount; /* struct files open RO */
693 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
694 void (*free_inode)(struct inode *);
696 struct file_lock_context *i_flctx;
697 struct address_space i_data;
698 struct list_head i_devices;
700 struct pipe_inode_info *i_pipe;
708 #ifdef CONFIG_FSNOTIFY
709 __u32 i_fsnotify_mask; /* all events this inode cares about */
710 struct fsnotify_mark_connector __rcu *i_fsnotify_marks;
713 #ifdef CONFIG_FS_ENCRYPTION
714 struct fscrypt_info *i_crypt_info;
717 #ifdef CONFIG_FS_VERITY
718 struct fsverity_info *i_verity_info;
721 void *i_private; /* fs or device private pointer */
722 } __randomize_layout;
724 struct timespec64 timestamp_truncate(struct timespec64 t, struct inode *inode);
726 static inline unsigned int i_blocksize(const struct inode *node)
728 return (1 << node->i_blkbits);
731 static inline int inode_unhashed(struct inode *inode)
733 return hlist_unhashed(&inode->i_hash);
737 * __mark_inode_dirty expects inodes to be hashed. Since we don't
738 * want special inodes in the fileset inode space, we make them
739 * appear hashed, but do not put on any lists. hlist_del()
740 * will work fine and require no locking.
742 static inline void inode_fake_hash(struct inode *inode)
744 hlist_add_fake(&inode->i_hash);
748 * inode->i_mutex nesting subclasses for the lock validator:
750 * 0: the object of the current VFS operation
754 * 4: second non-directory
755 * 5: second parent (when locking independent directories in rename)
757 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
758 * non-directories at once.
760 * The locking order between these classes is
761 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
763 enum inode_i_mutex_lock_class
773 static inline void inode_lock(struct inode *inode)
775 down_write(&inode->i_rwsem);
778 static inline void inode_unlock(struct inode *inode)
780 up_write(&inode->i_rwsem);
783 static inline void inode_lock_shared(struct inode *inode)
785 down_read(&inode->i_rwsem);
788 static inline void inode_unlock_shared(struct inode *inode)
790 up_read(&inode->i_rwsem);
793 static inline int inode_trylock(struct inode *inode)
795 return down_write_trylock(&inode->i_rwsem);
798 static inline int inode_trylock_shared(struct inode *inode)
800 return down_read_trylock(&inode->i_rwsem);
803 static inline int inode_is_locked(struct inode *inode)
805 return rwsem_is_locked(&inode->i_rwsem);
808 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
810 down_write_nested(&inode->i_rwsem, subclass);
813 static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass)
815 down_read_nested(&inode->i_rwsem, subclass);
818 void lock_two_nondirectories(struct inode *, struct inode*);
819 void unlock_two_nondirectories(struct inode *, struct inode*);
822 * NOTE: in a 32bit arch with a preemptable kernel and
823 * an UP compile the i_size_read/write must be atomic
824 * with respect to the local cpu (unlike with preempt disabled),
825 * but they don't need to be atomic with respect to other cpus like in
826 * true SMP (so they need either to either locally disable irq around
827 * the read or for example on x86 they can be still implemented as a
828 * cmpxchg8b without the need of the lock prefix). For SMP compiles
829 * and 64bit archs it makes no difference if preempt is enabled or not.
831 static inline loff_t i_size_read(const struct inode *inode)
833 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
838 seq = read_seqcount_begin(&inode->i_size_seqcount);
839 i_size = inode->i_size;
840 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
842 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
846 i_size = inode->i_size;
850 return inode->i_size;
855 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
856 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
857 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
859 static inline void i_size_write(struct inode *inode, loff_t i_size)
861 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
863 write_seqcount_begin(&inode->i_size_seqcount);
864 inode->i_size = i_size;
865 write_seqcount_end(&inode->i_size_seqcount);
867 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
869 inode->i_size = i_size;
872 inode->i_size = i_size;
876 static inline unsigned iminor(const struct inode *inode)
878 return MINOR(inode->i_rdev);
881 static inline unsigned imajor(const struct inode *inode)
883 return MAJOR(inode->i_rdev);
887 rwlock_t lock; /* protects pid, uid, euid fields */
888 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
889 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
890 kuid_t uid, euid; /* uid/euid of process setting the owner */
891 int signum; /* posix.1b rt signal to be delivered on IO */
895 * Track a single file's readahead state
897 struct file_ra_state {
898 pgoff_t start; /* where readahead started */
899 unsigned int size; /* # of readahead pages */
900 unsigned int async_size; /* do asynchronous readahead when
901 there are only # of pages ahead */
903 unsigned int ra_pages; /* Maximum readahead window */
904 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
905 loff_t prev_pos; /* Cache last read() position */
909 * Check if @index falls in the readahead windows.
911 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
913 return (index >= ra->start &&
914 index < ra->start + ra->size);
919 struct llist_node fu_llist;
920 struct rcu_head fu_rcuhead;
923 struct inode *f_inode; /* cached value */
924 const struct file_operations *f_op;
927 * Protects f_ep, f_flags.
928 * Must not be taken from IRQ context.
931 enum rw_hint f_write_hint;
932 atomic_long_t f_count;
933 unsigned int f_flags;
935 struct mutex f_pos_lock;
937 struct fown_struct f_owner;
938 const struct cred *f_cred;
939 struct file_ra_state f_ra;
942 #ifdef CONFIG_SECURITY
945 /* needed for tty driver, and maybe others */
949 /* Used by fs/eventpoll.c to link all the hooks to this file */
950 struct hlist_head *f_ep;
951 #endif /* #ifdef CONFIG_EPOLL */
952 struct address_space *f_mapping;
954 errseq_t f_sb_err; /* for syncfs */
956 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
961 /* file identifier */
962 unsigned char f_handle[];
965 static inline struct file *get_file(struct file *f)
967 atomic_long_inc(&f->f_count);
970 #define get_file_rcu_many(x, cnt) \
971 atomic_long_add_unless(&(x)->f_count, (cnt), 0)
972 #define get_file_rcu(x) get_file_rcu_many((x), 1)
973 #define file_count(x) atomic_long_read(&(x)->f_count)
975 #define MAX_NON_LFS ((1UL<<31) - 1)
977 /* Page cache limit. The filesystems should put that into their s_maxbytes
978 limits, otherwise bad things can happen in VM. */
979 #if BITS_PER_LONG==32
980 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
981 #elif BITS_PER_LONG==64
982 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
987 #define FL_DELEG 4 /* NFSv4 delegation */
988 #define FL_ACCESS 8 /* not trying to lock, just looking */
989 #define FL_EXISTS 16 /* when unlocking, test for existence */
990 #define FL_LEASE 32 /* lease held on this file */
991 #define FL_CLOSE 64 /* unlock on close */
992 #define FL_SLEEP 128 /* A blocking lock */
993 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
994 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
995 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
996 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
998 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
1001 * Special return value from posix_lock_file() and vfs_lock_file() for
1002 * asynchronous locking.
1004 #define FILE_LOCK_DEFERRED 1
1006 /* legacy typedef, should eventually be removed */
1007 typedef void *fl_owner_t;
1011 struct file_lock_operations {
1012 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1013 void (*fl_release_private)(struct file_lock *);
1016 struct lock_manager_operations {
1017 fl_owner_t (*lm_get_owner)(fl_owner_t);
1018 void (*lm_put_owner)(fl_owner_t);
1019 void (*lm_notify)(struct file_lock *); /* unblock callback */
1020 int (*lm_grant)(struct file_lock *, int);
1021 bool (*lm_break)(struct file_lock *);
1022 int (*lm_change)(struct file_lock *, int, struct list_head *);
1023 void (*lm_setup)(struct file_lock *, void **);
1024 bool (*lm_breaker_owns_lease)(struct file_lock *);
1027 struct lock_manager {
1028 struct list_head list;
1030 * NFSv4 and up also want opens blocked during the grace period;
1037 void locks_start_grace(struct net *, struct lock_manager *);
1038 void locks_end_grace(struct lock_manager *);
1039 bool locks_in_grace(struct net *);
1040 bool opens_in_grace(struct net *);
1042 /* that will die - we need it for nfs_lock_info */
1043 #include <linux/nfs_fs_i.h>
1046 * struct file_lock represents a generic "file lock". It's used to represent
1047 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1048 * note that the same struct is used to represent both a request for a lock and
1049 * the lock itself, but the same object is never used for both.
1051 * FIXME: should we create a separate "struct lock_request" to help distinguish
1054 * The varous i_flctx lists are ordered by:
1057 * 2) lock range start
1060 * Obviously, the last two criteria only matter for POSIX locks.
1063 struct file_lock *fl_blocker; /* The lock, that is blocking us */
1064 struct list_head fl_list; /* link into file_lock_context */
1065 struct hlist_node fl_link; /* node in global lists */
1066 struct list_head fl_blocked_requests; /* list of requests with
1067 * ->fl_blocker pointing here
1069 struct list_head fl_blocked_member; /* node in
1070 * ->fl_blocker->fl_blocked_requests
1072 fl_owner_t fl_owner;
1073 unsigned int fl_flags;
1074 unsigned char fl_type;
1075 unsigned int fl_pid;
1076 int fl_link_cpu; /* what cpu's list is this on? */
1077 wait_queue_head_t fl_wait;
1078 struct file *fl_file;
1082 struct fasync_struct * fl_fasync; /* for lease break notifications */
1083 /* for lease breaks: */
1084 unsigned long fl_break_time;
1085 unsigned long fl_downgrade_time;
1087 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1088 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1090 struct nfs_lock_info nfs_fl;
1091 struct nfs4_lock_info nfs4_fl;
1093 struct list_head link; /* link in AFS vnode's pending_locks list */
1094 int state; /* state of grant or error if -ve */
1095 unsigned int debug_id;
1098 } __randomize_layout;
1100 struct file_lock_context {
1101 spinlock_t flc_lock;
1102 struct list_head flc_flock;
1103 struct list_head flc_posix;
1104 struct list_head flc_lease;
1107 /* The following constant reflects the upper bound of the file/locking space */
1109 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1110 #define OFFSET_MAX INT_LIMIT(loff_t)
1111 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1114 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1116 #define locks_inode(f) file_inode(f)
1118 #ifdef CONFIG_FILE_LOCKING
1119 extern int fcntl_getlk(struct file *, unsigned int, struct flock *);
1120 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1123 #if BITS_PER_LONG == 32
1124 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *);
1125 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1129 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1130 extern int fcntl_getlease(struct file *filp);
1133 void locks_free_lock_context(struct inode *inode);
1134 void locks_free_lock(struct file_lock *fl);
1135 extern void locks_init_lock(struct file_lock *);
1136 extern struct file_lock * locks_alloc_lock(void);
1137 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1138 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1139 extern void locks_remove_posix(struct file *, fl_owner_t);
1140 extern void locks_remove_file(struct file *);
1141 extern void locks_release_private(struct file_lock *);
1142 extern void posix_test_lock(struct file *, struct file_lock *);
1143 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1144 extern int locks_delete_block(struct file_lock *);
1145 extern int vfs_test_lock(struct file *, struct file_lock *);
1146 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1147 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1148 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1149 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1150 extern void lease_get_mtime(struct inode *, struct timespec64 *time);
1151 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1152 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1153 extern int lease_modify(struct file_lock *, int, struct list_head *);
1155 struct notifier_block;
1156 extern int lease_register_notifier(struct notifier_block *);
1157 extern void lease_unregister_notifier(struct notifier_block *);
1159 struct files_struct;
1160 extern void show_fd_locks(struct seq_file *f,
1161 struct file *filp, struct files_struct *files);
1162 #else /* !CONFIG_FILE_LOCKING */
1163 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1164 struct flock __user *user)
1169 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1170 unsigned int cmd, struct flock __user *user)
1175 #if BITS_PER_LONG == 32
1176 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1177 struct flock64 __user *user)
1182 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1183 unsigned int cmd, struct flock64 __user *user)
1188 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1193 static inline int fcntl_getlease(struct file *filp)
1199 locks_free_lock_context(struct inode *inode)
1203 static inline void locks_init_lock(struct file_lock *fl)
1208 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1213 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1218 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1223 static inline void locks_remove_file(struct file *filp)
1228 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1233 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1234 struct file_lock *conflock)
1239 static inline int locks_delete_block(struct file_lock *waiter)
1244 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1249 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1250 struct file_lock *fl, struct file_lock *conf)
1255 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1260 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1265 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1270 static inline void lease_get_mtime(struct inode *inode,
1271 struct timespec64 *time)
1276 static inline int generic_setlease(struct file *filp, long arg,
1277 struct file_lock **flp, void **priv)
1282 static inline int vfs_setlease(struct file *filp, long arg,
1283 struct file_lock **lease, void **priv)
1288 static inline int lease_modify(struct file_lock *fl, int arg,
1289 struct list_head *dispose)
1294 struct files_struct;
1295 static inline void show_fd_locks(struct seq_file *f,
1296 struct file *filp, struct files_struct *files) {}
1297 #endif /* !CONFIG_FILE_LOCKING */
1299 static inline struct inode *file_inode(const struct file *f)
1304 static inline struct dentry *file_dentry(const struct file *file)
1306 return d_real(file->f_path.dentry, file_inode(file));
1309 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1311 return locks_lock_inode_wait(locks_inode(filp), fl);
1314 struct fasync_struct {
1318 struct fasync_struct *fa_next; /* singly linked list */
1319 struct file *fa_file;
1320 struct rcu_head fa_rcu;
1323 #define FASYNC_MAGIC 0x4601
1325 /* SMP safe fasync helpers: */
1326 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1327 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1328 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1329 extern struct fasync_struct *fasync_alloc(void);
1330 extern void fasync_free(struct fasync_struct *);
1332 /* can be called from interrupts */
1333 extern void kill_fasync(struct fasync_struct **, int, int);
1335 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1336 extern int f_setown(struct file *filp, unsigned long arg, int force);
1337 extern void f_delown(struct file *filp);
1338 extern pid_t f_getown(struct file *filp);
1339 extern int send_sigurg(struct fown_struct *fown);
1342 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1343 * represented in both.
1345 #define SB_RDONLY 1 /* Mount read-only */
1346 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1347 #define SB_NODEV 4 /* Disallow access to device special files */
1348 #define SB_NOEXEC 8 /* Disallow program execution */
1349 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1350 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1351 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1352 #define SB_NOATIME 1024 /* Do not update access times. */
1353 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1354 #define SB_SILENT 32768
1355 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1356 #define SB_INLINECRYPT (1<<17) /* Use blk-crypto for encrypted files */
1357 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1358 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1359 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1361 /* These sb flags are internal to the kernel */
1362 #define SB_SUBMOUNT (1<<26)
1363 #define SB_FORCE (1<<27)
1364 #define SB_NOSEC (1<<28)
1365 #define SB_BORN (1<<29)
1366 #define SB_ACTIVE (1<<30)
1367 #define SB_NOUSER (1<<31)
1369 /* These flags relate to encoding and casefolding */
1370 #define SB_ENC_STRICT_MODE_FL (1 << 0)
1372 #define sb_has_strict_encoding(sb) \
1373 (sb->s_encoding_flags & SB_ENC_STRICT_MODE_FL)
1379 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1380 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1381 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1382 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1383 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1386 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1387 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1388 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1389 #define SB_I_STABLE_WRITES 0x00000008 /* don't modify blks until WB is done */
1391 /* sb->s_iflags to limit user namespace mounts */
1392 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1393 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1394 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1396 #define SB_I_SKIP_SYNC 0x00000100 /* Skip superblock at global sync */
1398 /* Possible states of 'frozen' field */
1400 SB_UNFROZEN = 0, /* FS is unfrozen */
1401 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1402 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1403 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1404 * internal threads if needed) */
1405 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1408 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1411 int frozen; /* Is sb frozen? */
1412 wait_queue_head_t wait_unfrozen; /* wait for thaw */
1413 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1416 struct super_block {
1417 struct list_head s_list; /* Keep this first */
1418 dev_t s_dev; /* search index; _not_ kdev_t */
1419 unsigned char s_blocksize_bits;
1420 unsigned long s_blocksize;
1421 loff_t s_maxbytes; /* Max file size */
1422 struct file_system_type *s_type;
1423 const struct super_operations *s_op;
1424 const struct dquot_operations *dq_op;
1425 const struct quotactl_ops *s_qcop;
1426 const struct export_operations *s_export_op;
1427 unsigned long s_flags;
1428 unsigned long s_iflags; /* internal SB_I_* flags */
1429 unsigned long s_magic;
1430 struct dentry *s_root;
1431 struct rw_semaphore s_umount;
1434 #ifdef CONFIG_SECURITY
1437 const struct xattr_handler **s_xattr;
1438 #ifdef CONFIG_FS_ENCRYPTION
1439 const struct fscrypt_operations *s_cop;
1440 struct key *s_master_keys; /* master crypto keys in use */
1442 #ifdef CONFIG_FS_VERITY
1443 const struct fsverity_operations *s_vop;
1445 #ifdef CONFIG_UNICODE
1446 struct unicode_map *s_encoding;
1447 __u16 s_encoding_flags;
1449 struct hlist_bl_head s_roots; /* alternate root dentries for NFS */
1450 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1451 struct block_device *s_bdev;
1452 struct backing_dev_info *s_bdi;
1453 struct mtd_info *s_mtd;
1454 struct hlist_node s_instances;
1455 unsigned int s_quota_types; /* Bitmask of supported quota types */
1456 struct quota_info s_dquot; /* Diskquota specific options */
1458 struct sb_writers s_writers;
1461 * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and
1462 * s_fsnotify_marks together for cache efficiency. They are frequently
1463 * accessed and rarely modified.
1465 void *s_fs_info; /* Filesystem private info */
1467 /* Granularity of c/m/atime in ns (cannot be worse than a second) */
1469 /* Time limits for c/m/atime in seconds */
1470 time64_t s_time_min;
1471 time64_t s_time_max;
1472 #ifdef CONFIG_FSNOTIFY
1473 __u32 s_fsnotify_mask;
1474 struct fsnotify_mark_connector __rcu *s_fsnotify_marks;
1477 char s_id[32]; /* Informational name */
1478 uuid_t s_uuid; /* UUID */
1480 unsigned int s_max_links;
1484 * The next field is for VFS *only*. No filesystems have any business
1485 * even looking at it. You had been warned.
1487 struct mutex s_vfs_rename_mutex; /* Kludge */
1490 * Filesystem subtype. If non-empty the filesystem type field
1491 * in /proc/mounts will be "type.subtype"
1493 const char *s_subtype;
1495 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1498 * Saved pool identifier for cleancache (-1 means none)
1500 int cleancache_poolid;
1502 struct shrinker s_shrink; /* per-sb shrinker handle */
1504 /* Number of inodes with nlink == 0 but still referenced */
1505 atomic_long_t s_remove_count;
1507 /* Pending fsnotify inode refs */
1508 atomic_long_t s_fsnotify_inode_refs;
1510 /* Being remounted read-only */
1511 int s_readonly_remount;
1513 /* per-sb errseq_t for reporting writeback errors via syncfs */
1516 /* AIO completions deferred from interrupt context */
1517 struct workqueue_struct *s_dio_done_wq;
1518 struct hlist_head s_pins;
1521 * Owning user namespace and default context in which to
1522 * interpret filesystem uids, gids, quotas, device nodes,
1523 * xattrs and security labels.
1525 struct user_namespace *s_user_ns;
1528 * The list_lru structure is essentially just a pointer to a table
1529 * of per-node lru lists, each of which has its own spinlock.
1530 * There is no need to put them into separate cachelines.
1532 struct list_lru s_dentry_lru;
1533 struct list_lru s_inode_lru;
1534 struct rcu_head rcu;
1535 struct work_struct destroy_work;
1537 struct mutex s_sync_lock; /* sync serialisation lock */
1540 * Indicates how deep in a filesystem stack this SB is
1544 /* s_inode_list_lock protects s_inodes */
1545 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1546 struct list_head s_inodes; /* all inodes */
1548 spinlock_t s_inode_wblist_lock;
1549 struct list_head s_inodes_wb; /* writeback inodes */
1550 } __randomize_layout;
1552 /* Helper functions so that in most cases filesystems will
1553 * not need to deal directly with kuid_t and kgid_t and can
1554 * instead deal with the raw numeric values that are stored
1555 * in the filesystem.
1557 static inline uid_t i_uid_read(const struct inode *inode)
1559 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid);
1562 static inline gid_t i_gid_read(const struct inode *inode)
1564 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid);
1567 static inline void i_uid_write(struct inode *inode, uid_t uid)
1569 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid);
1572 static inline void i_gid_write(struct inode *inode, gid_t gid)
1574 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid);
1577 static inline kuid_t kuid_into_mnt(struct user_namespace *mnt_userns,
1580 return make_kuid(mnt_userns, __kuid_val(kuid));
1583 static inline kgid_t kgid_into_mnt(struct user_namespace *mnt_userns,
1586 return make_kgid(mnt_userns, __kgid_val(kgid));
1589 static inline kuid_t i_uid_into_mnt(struct user_namespace *mnt_userns,
1590 const struct inode *inode)
1592 return kuid_into_mnt(mnt_userns, inode->i_uid);
1595 static inline kgid_t i_gid_into_mnt(struct user_namespace *mnt_userns,
1596 const struct inode *inode)
1598 return kgid_into_mnt(mnt_userns, inode->i_gid);
1601 static inline kuid_t kuid_from_mnt(struct user_namespace *mnt_userns,
1604 return KUIDT_INIT(from_kuid(mnt_userns, kuid));
1607 static inline kgid_t kgid_from_mnt(struct user_namespace *mnt_userns,
1610 return KGIDT_INIT(from_kgid(mnt_userns, kgid));
1613 static inline kuid_t fsuid_into_mnt(struct user_namespace *mnt_userns)
1615 return kuid_from_mnt(mnt_userns, current_fsuid());
1618 static inline kgid_t fsgid_into_mnt(struct user_namespace *mnt_userns)
1620 return kgid_from_mnt(mnt_userns, current_fsgid());
1623 extern struct timespec64 current_time(struct inode *inode);
1626 * Snapshotting support.
1630 * These are internal functions, please use sb_start_{write,pagefault,intwrite}
1633 static inline void __sb_end_write(struct super_block *sb, int level)
1635 percpu_up_read(sb->s_writers.rw_sem + level-1);
1638 static inline void __sb_start_write(struct super_block *sb, int level)
1640 percpu_down_read(sb->s_writers.rw_sem + level - 1);
1643 static inline bool __sb_start_write_trylock(struct super_block *sb, int level)
1645 return percpu_down_read_trylock(sb->s_writers.rw_sem + level - 1);
1648 #define __sb_writers_acquired(sb, lev) \
1649 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1650 #define __sb_writers_release(sb, lev) \
1651 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1654 * sb_end_write - drop write access to a superblock
1655 * @sb: the super we wrote to
1657 * Decrement number of writers to the filesystem. Wake up possible waiters
1658 * wanting to freeze the filesystem.
1660 static inline void sb_end_write(struct super_block *sb)
1662 __sb_end_write(sb, SB_FREEZE_WRITE);
1666 * sb_end_pagefault - drop write access to a superblock from a page fault
1667 * @sb: the super we wrote to
1669 * Decrement number of processes handling write page fault to the filesystem.
1670 * Wake up possible waiters wanting to freeze the filesystem.
1672 static inline void sb_end_pagefault(struct super_block *sb)
1674 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1678 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1679 * @sb: the super we wrote to
1681 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1682 * waiters wanting to freeze the filesystem.
1684 static inline void sb_end_intwrite(struct super_block *sb)
1686 __sb_end_write(sb, SB_FREEZE_FS);
1690 * sb_start_write - get write access to a superblock
1691 * @sb: the super we write to
1693 * When a process wants to write data or metadata to a file system (i.e. dirty
1694 * a page or an inode), it should embed the operation in a sb_start_write() -
1695 * sb_end_write() pair to get exclusion against file system freezing. This
1696 * function increments number of writers preventing freezing. If the file
1697 * system is already frozen, the function waits until the file system is
1700 * Since freeze protection behaves as a lock, users have to preserve
1701 * ordering of freeze protection and other filesystem locks. Generally,
1702 * freeze protection should be the outermost lock. In particular, we have:
1705 * -> i_mutex (write path, truncate, directory ops, ...)
1706 * -> s_umount (freeze_super, thaw_super)
1708 static inline void sb_start_write(struct super_block *sb)
1710 __sb_start_write(sb, SB_FREEZE_WRITE);
1713 static inline bool sb_start_write_trylock(struct super_block *sb)
1715 return __sb_start_write_trylock(sb, SB_FREEZE_WRITE);
1719 * sb_start_pagefault - get write access to a superblock from a page fault
1720 * @sb: the super we write to
1722 * When a process starts handling write page fault, it should embed the
1723 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1724 * exclusion against file system freezing. This is needed since the page fault
1725 * is going to dirty a page. This function increments number of running page
1726 * faults preventing freezing. If the file system is already frozen, the
1727 * function waits until the file system is thawed.
1729 * Since page fault freeze protection behaves as a lock, users have to preserve
1730 * ordering of freeze protection and other filesystem locks. It is advised to
1731 * put sb_start_pagefault() close to mmap_lock in lock ordering. Page fault
1732 * handling code implies lock dependency:
1735 * -> sb_start_pagefault
1737 static inline void sb_start_pagefault(struct super_block *sb)
1739 __sb_start_write(sb, SB_FREEZE_PAGEFAULT);
1743 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1744 * @sb: the super we write to
1746 * This is the third level of protection against filesystem freezing. It is
1747 * free for use by a filesystem. The only requirement is that it must rank
1748 * below sb_start_pagefault.
1750 * For example filesystem can call sb_start_intwrite() when starting a
1751 * transaction which somewhat eases handling of freezing for internal sources
1752 * of filesystem changes (internal fs threads, discarding preallocation on file
1755 static inline void sb_start_intwrite(struct super_block *sb)
1757 __sb_start_write(sb, SB_FREEZE_FS);
1760 static inline bool sb_start_intwrite_trylock(struct super_block *sb)
1762 return __sb_start_write_trylock(sb, SB_FREEZE_FS);
1765 bool inode_owner_or_capable(struct user_namespace *mnt_userns,
1766 const struct inode *inode);
1769 * VFS helper functions..
1771 int vfs_create(struct user_namespace *, struct inode *,
1772 struct dentry *, umode_t, bool);
1773 int vfs_mkdir(struct user_namespace *, struct inode *,
1774 struct dentry *, umode_t);
1775 int vfs_mknod(struct user_namespace *, struct inode *, struct dentry *,
1777 int vfs_symlink(struct user_namespace *, struct inode *,
1778 struct dentry *, const char *);
1779 int vfs_link(struct dentry *, struct user_namespace *, struct inode *,
1780 struct dentry *, struct inode **);
1781 int vfs_rmdir(struct user_namespace *, struct inode *, struct dentry *);
1782 int vfs_unlink(struct user_namespace *, struct inode *, struct dentry *,
1786 struct user_namespace *old_mnt_userns;
1787 struct inode *old_dir;
1788 struct dentry *old_dentry;
1789 struct user_namespace *new_mnt_userns;
1790 struct inode *new_dir;
1791 struct dentry *new_dentry;
1792 struct inode **delegated_inode;
1794 } __randomize_layout;
1796 int vfs_rename(struct renamedata *);
1798 static inline int vfs_whiteout(struct user_namespace *mnt_userns,
1799 struct inode *dir, struct dentry *dentry)
1801 return vfs_mknod(mnt_userns, dir, dentry, S_IFCHR | WHITEOUT_MODE,
1805 struct dentry *vfs_tmpfile(struct user_namespace *mnt_userns,
1806 struct dentry *dentry, umode_t mode, int open_flag);
1808 int vfs_mkobj(struct dentry *, umode_t,
1809 int (*f)(struct dentry *, umode_t, void *),
1812 int vfs_fchown(struct file *file, uid_t user, gid_t group);
1813 int vfs_fchmod(struct file *file, umode_t mode);
1814 int vfs_utimes(const struct path *path, struct timespec64 *times);
1816 extern long vfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
1818 #ifdef CONFIG_COMPAT
1819 extern long compat_ptr_ioctl(struct file *file, unsigned int cmd,
1822 #define compat_ptr_ioctl NULL
1826 * VFS file helper functions.
1828 void inode_init_owner(struct user_namespace *mnt_userns, struct inode *inode,
1829 const struct inode *dir, umode_t mode);
1830 extern bool may_open_dev(const struct path *path);
1833 * This is the "filldir" function type, used by readdir() to let
1834 * the kernel specify what kind of dirent layout it wants to have.
1835 * This allows the kernel to read directories into kernel space or
1836 * to have different dirent layouts depending on the binary type.
1839 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1842 struct dir_context {
1848 * These flags let !MMU mmap() govern direct device mapping vs immediate
1849 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1851 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1852 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1853 * NOMMU_MAP_READ: Can be mapped for reading
1854 * NOMMU_MAP_WRITE: Can be mapped for writing
1855 * NOMMU_MAP_EXEC: Can be mapped for execution
1857 #define NOMMU_MAP_COPY 0x00000001
1858 #define NOMMU_MAP_DIRECT 0x00000008
1859 #define NOMMU_MAP_READ VM_MAYREAD
1860 #define NOMMU_MAP_WRITE VM_MAYWRITE
1861 #define NOMMU_MAP_EXEC VM_MAYEXEC
1863 #define NOMMU_VMFLAGS \
1864 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1867 * These flags control the behavior of the remap_file_range function pointer.
1868 * If it is called with len == 0 that means "remap to end of source file".
1869 * See Documentation/filesystems/vfs.rst for more details about this call.
1871 * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate)
1872 * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request
1874 #define REMAP_FILE_DEDUP (1 << 0)
1875 #define REMAP_FILE_CAN_SHORTEN (1 << 1)
1878 * These flags signal that the caller is ok with altering various aspects of
1879 * the behavior of the remap operation. The changes must be made by the
1880 * implementation; the vfs remap helper functions can take advantage of them.
1881 * Flags in this category exist to preserve the quirky behavior of the hoisted
1882 * btrfs clone/dedupe ioctls.
1884 #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN)
1888 struct file_operations {
1889 struct module *owner;
1890 loff_t (*llseek) (struct file *, loff_t, int);
1891 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1892 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1893 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1894 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1895 int (*iopoll)(struct kiocb *kiocb, bool spin);
1896 int (*iterate) (struct file *, struct dir_context *);
1897 int (*iterate_shared) (struct file *, struct dir_context *);
1898 __poll_t (*poll) (struct file *, struct poll_table_struct *);
1899 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1900 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1901 int (*mmap) (struct file *, struct vm_area_struct *);
1902 unsigned long mmap_supported_flags;
1903 int (*open) (struct inode *, struct file *);
1904 int (*flush) (struct file *, fl_owner_t id);
1905 int (*release) (struct inode *, struct file *);
1906 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1907 int (*fasync) (int, struct file *, int);
1908 int (*lock) (struct file *, int, struct file_lock *);
1909 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1910 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1911 int (*check_flags)(int);
1912 int (*flock) (struct file *, int, struct file_lock *);
1913 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1914 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1915 int (*setlease)(struct file *, long, struct file_lock **, void **);
1916 long (*fallocate)(struct file *file, int mode, loff_t offset,
1918 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1920 unsigned (*mmap_capabilities)(struct file *);
1922 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1923 loff_t, size_t, unsigned int);
1924 loff_t (*remap_file_range)(struct file *file_in, loff_t pos_in,
1925 struct file *file_out, loff_t pos_out,
1926 loff_t len, unsigned int remap_flags);
1927 int (*fadvise)(struct file *, loff_t, loff_t, int);
1928 } __randomize_layout;
1930 struct inode_operations {
1931 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1932 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
1933 int (*permission) (struct user_namespace *, struct inode *, int);
1934 struct posix_acl * (*get_acl)(struct inode *, int);
1936 int (*readlink) (struct dentry *, char __user *,int);
1938 int (*create) (struct user_namespace *, struct inode *,struct dentry *,
1940 int (*link) (struct dentry *,struct inode *,struct dentry *);
1941 int (*unlink) (struct inode *,struct dentry *);
1942 int (*symlink) (struct user_namespace *, struct inode *,struct dentry *,
1944 int (*mkdir) (struct user_namespace *, struct inode *,struct dentry *,
1946 int (*rmdir) (struct inode *,struct dentry *);
1947 int (*mknod) (struct user_namespace *, struct inode *,struct dentry *,
1949 int (*rename) (struct user_namespace *, struct inode *, struct dentry *,
1950 struct inode *, struct dentry *, unsigned int);
1951 int (*setattr) (struct user_namespace *, struct dentry *,
1953 int (*getattr) (struct user_namespace *, const struct path *,
1954 struct kstat *, u32, unsigned int);
1955 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1956 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1958 int (*update_time)(struct inode *, struct timespec64 *, int);
1959 int (*atomic_open)(struct inode *, struct dentry *,
1960 struct file *, unsigned open_flag,
1961 umode_t create_mode);
1962 int (*tmpfile) (struct user_namespace *, struct inode *,
1963 struct dentry *, umode_t);
1964 int (*set_acl)(struct user_namespace *, struct inode *,
1965 struct posix_acl *, int);
1966 } ____cacheline_aligned;
1968 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
1969 struct iov_iter *iter)
1971 return file->f_op->read_iter(kio, iter);
1974 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
1975 struct iov_iter *iter)
1977 return file->f_op->write_iter(kio, iter);
1980 static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
1982 return file->f_op->mmap(file, vma);
1985 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1986 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1987 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1988 loff_t, size_t, unsigned int);
1989 extern ssize_t generic_copy_file_range(struct file *file_in, loff_t pos_in,
1990 struct file *file_out, loff_t pos_out,
1991 size_t len, unsigned int flags);
1992 extern int generic_remap_file_range_prep(struct file *file_in, loff_t pos_in,
1993 struct file *file_out, loff_t pos_out,
1995 unsigned int remap_flags);
1996 extern loff_t do_clone_file_range(struct file *file_in, loff_t pos_in,
1997 struct file *file_out, loff_t pos_out,
1998 loff_t len, unsigned int remap_flags);
1999 extern loff_t vfs_clone_file_range(struct file *file_in, loff_t pos_in,
2000 struct file *file_out, loff_t pos_out,
2001 loff_t len, unsigned int remap_flags);
2002 extern int vfs_dedupe_file_range(struct file *file,
2003 struct file_dedupe_range *same);
2004 extern loff_t vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos,
2005 struct file *dst_file, loff_t dst_pos,
2006 loff_t len, unsigned int remap_flags);
2009 struct super_operations {
2010 struct inode *(*alloc_inode)(struct super_block *sb);
2011 void (*destroy_inode)(struct inode *);
2012 void (*free_inode)(struct inode *);
2014 void (*dirty_inode) (struct inode *, int flags);
2015 int (*write_inode) (struct inode *, struct writeback_control *wbc);
2016 int (*drop_inode) (struct inode *);
2017 void (*evict_inode) (struct inode *);
2018 void (*put_super) (struct super_block *);
2019 int (*sync_fs)(struct super_block *sb, int wait);
2020 int (*freeze_super) (struct super_block *);
2021 int (*freeze_fs) (struct super_block *);
2022 int (*thaw_super) (struct super_block *);
2023 int (*unfreeze_fs) (struct super_block *);
2024 int (*statfs) (struct dentry *, struct kstatfs *);
2025 int (*remount_fs) (struct super_block *, int *, char *);
2026 void (*umount_begin) (struct super_block *);
2028 int (*show_options)(struct seq_file *, struct dentry *);
2029 int (*show_devname)(struct seq_file *, struct dentry *);
2030 int (*show_path)(struct seq_file *, struct dentry *);
2031 int (*show_stats)(struct seq_file *, struct dentry *);
2033 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
2034 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
2035 struct dquot **(*get_dquots)(struct inode *);
2037 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
2038 long (*nr_cached_objects)(struct super_block *,
2039 struct shrink_control *);
2040 long (*free_cached_objects)(struct super_block *,
2041 struct shrink_control *);
2045 * Inode flags - they have no relation to superblock flags now
2047 #define S_SYNC (1 << 0) /* Writes are synced at once */
2048 #define S_NOATIME (1 << 1) /* Do not update access times */
2049 #define S_APPEND (1 << 2) /* Append-only file */
2050 #define S_IMMUTABLE (1 << 3) /* Immutable file */
2051 #define S_DEAD (1 << 4) /* removed, but still open directory */
2052 #define S_NOQUOTA (1 << 5) /* Inode is not counted to quota */
2053 #define S_DIRSYNC (1 << 6) /* Directory modifications are synchronous */
2054 #define S_NOCMTIME (1 << 7) /* Do not update file c/mtime */
2055 #define S_SWAPFILE (1 << 8) /* Do not truncate: swapon got its bmaps */
2056 #define S_PRIVATE (1 << 9) /* Inode is fs-internal */
2057 #define S_IMA (1 << 10) /* Inode has an associated IMA struct */
2058 #define S_AUTOMOUNT (1 << 11) /* Automount/referral quasi-directory */
2059 #define S_NOSEC (1 << 12) /* no suid or xattr security attributes */
2060 #ifdef CONFIG_FS_DAX
2061 #define S_DAX (1 << 13) /* Direct Access, avoiding the page cache */
2063 #define S_DAX 0 /* Make all the DAX code disappear */
2065 #define S_ENCRYPTED (1 << 14) /* Encrypted file (using fs/crypto/) */
2066 #define S_CASEFOLD (1 << 15) /* Casefolded file */
2067 #define S_VERITY (1 << 16) /* Verity file (using fs/verity/) */
2070 * Note that nosuid etc flags are inode-specific: setting some file-system
2071 * flags just means all the inodes inherit those flags by default. It might be
2072 * possible to override it selectively if you really wanted to with some
2073 * ioctl() that is not currently implemented.
2075 * Exception: SB_RDONLY is always applied to the entire file system.
2077 * Unfortunately, it is possible to change a filesystems flags with it mounted
2078 * with files in use. This means that all of the inodes will not have their
2079 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
2080 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
2082 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
2084 static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; }
2085 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
2086 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
2087 ((inode)->i_flags & S_SYNC))
2088 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
2089 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
2090 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
2091 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
2092 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
2094 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
2095 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
2096 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
2097 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
2099 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
2100 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
2101 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
2102 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
2103 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
2104 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
2105 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
2106 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
2107 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
2108 #define IS_CASEFOLDED(inode) ((inode)->i_flags & S_CASEFOLD)
2109 #define IS_VERITY(inode) ((inode)->i_flags & S_VERITY)
2111 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
2112 (inode)->i_rdev == WHITEOUT_DEV)
2114 static inline bool HAS_UNMAPPED_ID(struct user_namespace *mnt_userns,
2115 struct inode *inode)
2117 return !uid_valid(i_uid_into_mnt(mnt_userns, inode)) ||
2118 !gid_valid(i_gid_into_mnt(mnt_userns, inode));
2121 static inline enum rw_hint file_write_hint(struct file *file)
2123 if (file->f_write_hint != WRITE_LIFE_NOT_SET)
2124 return file->f_write_hint;
2126 return file_inode(file)->i_write_hint;
2129 static inline int iocb_flags(struct file *file);
2131 static inline u16 ki_hint_validate(enum rw_hint hint)
2133 typeof(((struct kiocb *)0)->ki_hint) max_hint = -1;
2135 if (hint <= max_hint)
2140 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
2142 *kiocb = (struct kiocb) {
2144 .ki_flags = iocb_flags(filp),
2145 .ki_hint = ki_hint_validate(file_write_hint(filp)),
2146 .ki_ioprio = get_current_ioprio(),
2150 static inline void kiocb_clone(struct kiocb *kiocb, struct kiocb *kiocb_src,
2153 *kiocb = (struct kiocb) {
2155 .ki_flags = kiocb_src->ki_flags,
2156 .ki_hint = kiocb_src->ki_hint,
2157 .ki_ioprio = kiocb_src->ki_ioprio,
2158 .ki_pos = kiocb_src->ki_pos,
2163 * Inode state bits. Protected by inode->i_lock
2165 * Four bits determine the dirty state of the inode: I_DIRTY_SYNC,
2166 * I_DIRTY_DATASYNC, I_DIRTY_PAGES, and I_DIRTY_TIME.
2168 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
2169 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
2170 * various stages of removing an inode.
2172 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2174 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
2175 * fdatasync() (unless I_DIRTY_DATASYNC is also set).
2176 * Timestamp updates are the usual cause.
2177 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
2178 * these changes separately from I_DIRTY_SYNC so that we
2179 * don't have to write inode on fdatasync() when only
2180 * e.g. the timestamps have changed.
2181 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
2182 * I_DIRTY_TIME The inode itself only has dirty timestamps, and the
2183 * lazytime mount option is enabled. We keep track of this
2184 * separately from I_DIRTY_SYNC in order to implement
2185 * lazytime. This gets cleared if I_DIRTY_INODE
2186 * (I_DIRTY_SYNC and/or I_DIRTY_DATASYNC) gets set. I.e.
2187 * either I_DIRTY_TIME *or* I_DIRTY_INODE can be set in
2188 * i_state, but not both. I_DIRTY_PAGES may still be set.
2189 * I_NEW Serves as both a mutex and completion notification.
2190 * New inodes set I_NEW. If two processes both create
2191 * the same inode, one of them will release its inode and
2192 * wait for I_NEW to be released before returning.
2193 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2194 * also cause waiting on I_NEW, without I_NEW actually
2195 * being set. find_inode() uses this to prevent returning
2196 * nearly-dead inodes.
2197 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2198 * is zero. I_FREEING must be set when I_WILL_FREE is
2200 * I_FREEING Set when inode is about to be freed but still has dirty
2201 * pages or buffers attached or the inode itself is still
2203 * I_CLEAR Added by clear_inode(). In this state the inode is
2204 * clean and can be destroyed. Inode keeps I_FREEING.
2206 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2207 * prohibited for many purposes. iget() must wait for
2208 * the inode to be completely released, then create it
2209 * anew. Other functions will just ignore such inodes,
2210 * if appropriate. I_NEW is used for waiting.
2212 * I_SYNC Writeback of inode is running. The bit is set during
2213 * data writeback, and cleared with a wakeup on the bit
2214 * address once it is done. The bit is also used to pin
2215 * the inode in memory for flusher thread.
2217 * I_REFERENCED Marks the inode as recently references on the LRU list.
2219 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2221 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2222 * synchronize competing switching instances and to tell
2223 * wb stat updates to grab the i_pages lock. See
2224 * inode_switch_wbs_work_fn() for details.
2226 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2227 * and work dirs among overlayfs mounts.
2229 * I_CREATING New object's inode in the middle of setting up.
2231 * I_DONTCACHE Evict inode as soon as it is not used anymore.
2233 * I_SYNC_QUEUED Inode is queued in b_io or b_more_io writeback lists.
2234 * Used to detect that mark_inode_dirty() should not move
2235 * inode between dirty lists.
2237 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2239 #define I_DIRTY_SYNC (1 << 0)
2240 #define I_DIRTY_DATASYNC (1 << 1)
2241 #define I_DIRTY_PAGES (1 << 2)
2243 #define I_NEW (1 << __I_NEW)
2244 #define I_WILL_FREE (1 << 4)
2245 #define I_FREEING (1 << 5)
2246 #define I_CLEAR (1 << 6)
2248 #define I_SYNC (1 << __I_SYNC)
2249 #define I_REFERENCED (1 << 8)
2250 #define __I_DIO_WAKEUP 9
2251 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2252 #define I_LINKABLE (1 << 10)
2253 #define I_DIRTY_TIME (1 << 11)
2254 #define I_WB_SWITCH (1 << 13)
2255 #define I_OVL_INUSE (1 << 14)
2256 #define I_CREATING (1 << 15)
2257 #define I_DONTCACHE (1 << 16)
2258 #define I_SYNC_QUEUED (1 << 17)
2260 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2261 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2262 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2264 extern void __mark_inode_dirty(struct inode *, int);
2265 static inline void mark_inode_dirty(struct inode *inode)
2267 __mark_inode_dirty(inode, I_DIRTY);
2270 static inline void mark_inode_dirty_sync(struct inode *inode)
2272 __mark_inode_dirty(inode, I_DIRTY_SYNC);
2276 * Returns true if the given inode itself only has dirty timestamps (its pages
2277 * may still be dirty) and isn't currently being allocated or freed.
2278 * Filesystems should call this if when writing an inode when lazytime is
2279 * enabled, they want to opportunistically write the timestamps of other inodes
2280 * located very nearby on-disk, e.g. in the same inode block. This returns true
2281 * if the given inode is in need of such an opportunistic update. Requires
2282 * i_lock, or at least later re-checking under i_lock.
2284 static inline bool inode_is_dirtytime_only(struct inode *inode)
2286 return (inode->i_state & (I_DIRTY_TIME | I_NEW |
2287 I_FREEING | I_WILL_FREE)) == I_DIRTY_TIME;
2290 extern void inc_nlink(struct inode *inode);
2291 extern void drop_nlink(struct inode *inode);
2292 extern void clear_nlink(struct inode *inode);
2293 extern void set_nlink(struct inode *inode, unsigned int nlink);
2295 static inline void inode_inc_link_count(struct inode *inode)
2298 mark_inode_dirty(inode);
2301 static inline void inode_dec_link_count(struct inode *inode)
2304 mark_inode_dirty(inode);
2307 enum file_time_flags {
2314 extern bool atime_needs_update(const struct path *, struct inode *);
2315 extern void touch_atime(const struct path *);
2316 static inline void file_accessed(struct file *file)
2318 if (!(file->f_flags & O_NOATIME))
2319 touch_atime(&file->f_path);
2322 extern int file_modified(struct file *file);
2324 int sync_inode(struct inode *inode, struct writeback_control *wbc);
2325 int sync_inode_metadata(struct inode *inode, int wait);
2327 struct file_system_type {
2330 #define FS_REQUIRES_DEV 1
2331 #define FS_BINARY_MOUNTDATA 2
2332 #define FS_HAS_SUBTYPE 4
2333 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2334 #define FS_DISALLOW_NOTIFY_PERM 16 /* Disable fanotify permission events */
2335 #define FS_ALLOW_IDMAP 32 /* FS has been updated to handle vfs idmappings. */
2336 #define FS_THP_SUPPORT 8192 /* Remove once all fs converted */
2337 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2338 int (*init_fs_context)(struct fs_context *);
2339 const struct fs_parameter_spec *parameters;
2340 struct dentry *(*mount) (struct file_system_type *, int,
2341 const char *, void *);
2342 void (*kill_sb) (struct super_block *);
2343 struct module *owner;
2344 struct file_system_type * next;
2345 struct hlist_head fs_supers;
2347 struct lock_class_key s_lock_key;
2348 struct lock_class_key s_umount_key;
2349 struct lock_class_key s_vfs_rename_key;
2350 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2352 struct lock_class_key i_lock_key;
2353 struct lock_class_key i_mutex_key;
2354 struct lock_class_key i_mutex_dir_key;
2357 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2359 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2360 int flags, const char *dev_name, void *data,
2361 int (*fill_super)(struct super_block *, void *, int));
2362 extern struct dentry *mount_single(struct file_system_type *fs_type,
2363 int flags, void *data,
2364 int (*fill_super)(struct super_block *, void *, int));
2365 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2366 int flags, void *data,
2367 int (*fill_super)(struct super_block *, void *, int));
2368 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2369 void generic_shutdown_super(struct super_block *sb);
2370 void kill_block_super(struct super_block *sb);
2371 void kill_anon_super(struct super_block *sb);
2372 void kill_litter_super(struct super_block *sb);
2373 void deactivate_super(struct super_block *sb);
2374 void deactivate_locked_super(struct super_block *sb);
2375 int set_anon_super(struct super_block *s, void *data);
2376 int set_anon_super_fc(struct super_block *s, struct fs_context *fc);
2377 int get_anon_bdev(dev_t *);
2378 void free_anon_bdev(dev_t);
2379 struct super_block *sget_fc(struct fs_context *fc,
2380 int (*test)(struct super_block *, struct fs_context *),
2381 int (*set)(struct super_block *, struct fs_context *));
2382 struct super_block *sget(struct file_system_type *type,
2383 int (*test)(struct super_block *,void *),
2384 int (*set)(struct super_block *,void *),
2385 int flags, void *data);
2387 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2388 #define fops_get(fops) \
2389 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2390 #define fops_put(fops) \
2391 do { if (fops) module_put((fops)->owner); } while(0)
2393 * This one is to be used *ONLY* from ->open() instances.
2394 * fops must be non-NULL, pinned down *and* module dependencies
2395 * should be sufficient to pin the caller down as well.
2397 #define replace_fops(f, fops) \
2399 struct file *__file = (f); \
2400 fops_put(__file->f_op); \
2401 BUG_ON(!(__file->f_op = (fops))); \
2404 extern int register_filesystem(struct file_system_type *);
2405 extern int unregister_filesystem(struct file_system_type *);
2406 extern struct vfsmount *kern_mount(struct file_system_type *);
2407 extern void kern_unmount(struct vfsmount *mnt);
2408 extern int may_umount_tree(struct vfsmount *);
2409 extern int may_umount(struct vfsmount *);
2410 extern long do_mount(const char *, const char __user *,
2411 const char *, unsigned long, void *);
2412 extern struct vfsmount *collect_mounts(const struct path *);
2413 extern void drop_collected_mounts(struct vfsmount *);
2414 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2416 extern int vfs_statfs(const struct path *, struct kstatfs *);
2417 extern int user_statfs(const char __user *, struct kstatfs *);
2418 extern int fd_statfs(int, struct kstatfs *);
2419 extern int freeze_super(struct super_block *super);
2420 extern int thaw_super(struct super_block *super);
2421 extern bool our_mnt(struct vfsmount *mnt);
2422 extern __printf(2, 3)
2423 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2424 extern int super_setup_bdi(struct super_block *sb);
2426 extern int current_umask(void);
2428 extern void ihold(struct inode * inode);
2429 extern void iput(struct inode *);
2430 extern int generic_update_time(struct inode *, struct timespec64 *, int);
2433 extern struct kobject *fs_kobj;
2435 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2437 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2438 extern int locks_mandatory_locked(struct file *);
2439 extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
2442 * Candidates for mandatory locking have the setgid bit set
2443 * but no group execute bit - an otherwise meaningless combination.
2446 static inline int __mandatory_lock(struct inode *ino)
2448 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2452 * ... and these candidates should be on SB_MANDLOCK mounted fs,
2453 * otherwise these will be advisory locks
2456 static inline int mandatory_lock(struct inode *ino)
2458 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2461 static inline int locks_verify_locked(struct file *file)
2463 if (mandatory_lock(locks_inode(file)))
2464 return locks_mandatory_locked(file);
2468 static inline int locks_verify_truncate(struct inode *inode,
2472 if (!inode->i_flctx || !mandatory_lock(inode))
2475 if (size < inode->i_size) {
2476 return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2479 return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2484 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2486 static inline int locks_mandatory_locked(struct file *file)
2491 static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2492 loff_t start, loff_t end, unsigned char type)
2497 static inline int __mandatory_lock(struct inode *inode)
2502 static inline int mandatory_lock(struct inode *inode)
2507 static inline int locks_verify_locked(struct file *file)
2512 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2518 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2521 #ifdef CONFIG_FILE_LOCKING
2522 static inline int break_lease(struct inode *inode, unsigned int mode)
2525 * Since this check is lockless, we must ensure that any refcounts
2526 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2527 * could end up racing with tasks trying to set a new lease on this
2531 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2532 return __break_lease(inode, mode, FL_LEASE);
2536 static inline int break_deleg(struct inode *inode, unsigned int mode)
2539 * Since this check is lockless, we must ensure that any refcounts
2540 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2541 * could end up racing with tasks trying to set a new lease on this
2545 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2546 return __break_lease(inode, mode, FL_DELEG);
2550 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2554 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2555 if (ret == -EWOULDBLOCK && delegated_inode) {
2556 *delegated_inode = inode;
2562 static inline int break_deleg_wait(struct inode **delegated_inode)
2566 ret = break_deleg(*delegated_inode, O_WRONLY);
2567 iput(*delegated_inode);
2568 *delegated_inode = NULL;
2572 static inline int break_layout(struct inode *inode, bool wait)
2575 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2576 return __break_lease(inode,
2577 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2582 #else /* !CONFIG_FILE_LOCKING */
2583 static inline int break_lease(struct inode *inode, unsigned int mode)
2588 static inline int break_deleg(struct inode *inode, unsigned int mode)
2593 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2598 static inline int break_deleg_wait(struct inode **delegated_inode)
2604 static inline int break_layout(struct inode *inode, bool wait)
2609 #endif /* CONFIG_FILE_LOCKING */
2614 const char *name; /* pointer to actual string */
2615 const __user char *uptr; /* original userland pointer */
2617 struct audit_names *aname;
2620 static_assert(offsetof(struct filename, iname) % sizeof(long) == 0);
2622 static inline struct user_namespace *file_mnt_user_ns(struct file *file)
2624 return mnt_user_ns(file->f_path.mnt);
2626 extern long vfs_truncate(const struct path *, loff_t);
2627 int do_truncate(struct user_namespace *, struct dentry *, loff_t start,
2628 unsigned int time_attrs, struct file *filp);
2629 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2631 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2633 extern struct file *file_open_name(struct filename *, int, umode_t);
2634 extern struct file *filp_open(const char *, int, umode_t);
2635 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2636 const char *, int, umode_t);
2637 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2638 extern struct file * open_with_fake_path(const struct path *, int,
2639 struct inode*, const struct cred *);
2640 static inline struct file *file_clone_open(struct file *file)
2642 return dentry_open(&file->f_path, file->f_flags, file->f_cred);
2644 extern int filp_close(struct file *, fl_owner_t id);
2646 extern struct filename *getname_flags(const char __user *, int, int *);
2647 extern struct filename *getname(const char __user *);
2648 extern struct filename *getname_kernel(const char *);
2649 extern void putname(struct filename *name);
2651 extern int finish_open(struct file *file, struct dentry *dentry,
2652 int (*open)(struct inode *, struct file *));
2653 extern int finish_no_open(struct file *file, struct dentry *dentry);
2656 extern void __init vfs_caches_init_early(void);
2657 extern void __init vfs_caches_init(void);
2659 extern struct kmem_cache *names_cachep;
2661 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2662 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2664 extern struct super_block *blockdev_superblock;
2665 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2667 return IS_ENABLED(CONFIG_BLOCK) && sb == blockdev_superblock;
2670 void emergency_thaw_all(void);
2671 extern int sync_filesystem(struct super_block *);
2672 extern const struct file_operations def_blk_fops;
2673 extern const struct file_operations def_chr_fops;
2676 #define CHRDEV_MAJOR_MAX 512
2677 /* Marks the bottom of the first segment of free char majors */
2678 #define CHRDEV_MAJOR_DYN_END 234
2679 /* Marks the top and bottom of the second segment of free char majors */
2680 #define CHRDEV_MAJOR_DYN_EXT_START 511
2681 #define CHRDEV_MAJOR_DYN_EXT_END 384
2683 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2684 extern int register_chrdev_region(dev_t, unsigned, const char *);
2685 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2686 unsigned int count, const char *name,
2687 const struct file_operations *fops);
2688 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2689 unsigned int count, const char *name);
2690 extern void unregister_chrdev_region(dev_t, unsigned);
2691 extern void chrdev_show(struct seq_file *,off_t);
2693 static inline int register_chrdev(unsigned int major, const char *name,
2694 const struct file_operations *fops)
2696 return __register_chrdev(major, 0, 256, name, fops);
2699 static inline void unregister_chrdev(unsigned int major, const char *name)
2701 __unregister_chrdev(major, 0, 256, name);
2704 extern void init_special_inode(struct inode *, umode_t, dev_t);
2706 /* Invalid inode operations -- fs/bad_inode.c */
2707 extern void make_bad_inode(struct inode *);
2708 extern bool is_bad_inode(struct inode *);
2710 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2711 pgoff_t start, pgoff_t end);
2713 void invalidate_mapping_pagevec(struct address_space *mapping,
2714 pgoff_t start, pgoff_t end,
2715 unsigned long *nr_pagevec);
2717 static inline void invalidate_remote_inode(struct inode *inode)
2719 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2720 S_ISLNK(inode->i_mode))
2721 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2723 extern int invalidate_inode_pages2(struct address_space *mapping);
2724 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2725 pgoff_t start, pgoff_t end);
2726 extern int write_inode_now(struct inode *, int);
2727 extern int filemap_fdatawrite(struct address_space *);
2728 extern int filemap_flush(struct address_space *);
2729 extern int filemap_fdatawait_keep_errors(struct address_space *mapping);
2730 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2732 extern int filemap_fdatawait_range_keep_errors(struct address_space *mapping,
2733 loff_t start_byte, loff_t end_byte);
2735 static inline int filemap_fdatawait(struct address_space *mapping)
2737 return filemap_fdatawait_range(mapping, 0, LLONG_MAX);
2740 extern bool filemap_range_has_page(struct address_space *, loff_t lstart,
2742 extern int filemap_write_and_wait_range(struct address_space *mapping,
2743 loff_t lstart, loff_t lend);
2744 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2745 loff_t start, loff_t end, int sync_mode);
2746 extern int filemap_fdatawrite_range(struct address_space *mapping,
2747 loff_t start, loff_t end);
2748 extern int filemap_check_errors(struct address_space *mapping);
2749 extern void __filemap_set_wb_err(struct address_space *mapping, int err);
2751 static inline int filemap_write_and_wait(struct address_space *mapping)
2753 return filemap_write_and_wait_range(mapping, 0, LLONG_MAX);
2756 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2758 extern int __must_check file_check_and_advance_wb_err(struct file *file);
2759 extern int __must_check file_write_and_wait_range(struct file *file,
2760 loff_t start, loff_t end);
2762 static inline int file_write_and_wait(struct file *file)
2764 return file_write_and_wait_range(file, 0, LLONG_MAX);
2768 * filemap_set_wb_err - set a writeback error on an address_space
2769 * @mapping: mapping in which to set writeback error
2770 * @err: error to be set in mapping
2772 * When writeback fails in some way, we must record that error so that
2773 * userspace can be informed when fsync and the like are called. We endeavor
2774 * to report errors on any file that was open at the time of the error. Some
2775 * internal callers also need to know when writeback errors have occurred.
2777 * When a writeback error occurs, most filesystems will want to call
2778 * filemap_set_wb_err to record the error in the mapping so that it will be
2779 * automatically reported whenever fsync is called on the file.
2781 static inline void filemap_set_wb_err(struct address_space *mapping, int err)
2783 /* Fastpath for common case of no error */
2785 __filemap_set_wb_err(mapping, err);
2789 * filemap_check_wb_err - has an error occurred since the mark was sampled?
2790 * @mapping: mapping to check for writeback errors
2791 * @since: previously-sampled errseq_t
2793 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2794 * the given value was sampled.
2796 * If it has then report the latest error set, otherwise return 0.
2798 static inline int filemap_check_wb_err(struct address_space *mapping,
2801 return errseq_check(&mapping->wb_err, since);
2805 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2806 * @mapping: mapping to be sampled
2808 * Writeback errors are always reported relative to a particular sample point
2809 * in the past. This function provides those sample points.
2811 static inline errseq_t filemap_sample_wb_err(struct address_space *mapping)
2813 return errseq_sample(&mapping->wb_err);
2817 * file_sample_sb_err - sample the current errseq_t to test for later errors
2818 * @file: file pointer to be sampled
2820 * Grab the most current superblock-level errseq_t value for the given
2823 static inline errseq_t file_sample_sb_err(struct file *file)
2825 return errseq_sample(&file->f_path.dentry->d_sb->s_wb_err);
2828 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2830 extern int vfs_fsync(struct file *file, int datasync);
2832 extern int sync_file_range(struct file *file, loff_t offset, loff_t nbytes,
2833 unsigned int flags);
2836 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2837 * to already be updated for the write, and will return either the amount
2838 * of bytes passed in, or an error if syncing the file failed.
2840 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2842 if (iocb->ki_flags & IOCB_DSYNC) {
2843 int ret = vfs_fsync_range(iocb->ki_filp,
2844 iocb->ki_pos - count, iocb->ki_pos - 1,
2845 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2853 extern void emergency_sync(void);
2854 extern void emergency_remount(void);
2857 extern int bmap(struct inode *inode, sector_t *block);
2859 static inline int bmap(struct inode *inode, sector_t *block)
2865 int notify_change(struct user_namespace *, struct dentry *,
2866 struct iattr *, struct inode **);
2867 int inode_permission(struct user_namespace *, struct inode *, int);
2868 int generic_permission(struct user_namespace *, struct inode *, int);
2869 static inline int file_permission(struct file *file, int mask)
2871 return inode_permission(file_mnt_user_ns(file),
2872 file_inode(file), mask);
2874 static inline int path_permission(const struct path *path, int mask)
2876 return inode_permission(mnt_user_ns(path->mnt),
2877 d_inode(path->dentry), mask);
2879 int __check_sticky(struct user_namespace *mnt_userns, struct inode *dir,
2880 struct inode *inode);
2882 static inline bool execute_ok(struct inode *inode)
2884 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2887 static inline void file_start_write(struct file *file)
2889 if (!S_ISREG(file_inode(file)->i_mode))
2891 sb_start_write(file_inode(file)->i_sb);
2894 static inline bool file_start_write_trylock(struct file *file)
2896 if (!S_ISREG(file_inode(file)->i_mode))
2898 return sb_start_write_trylock(file_inode(file)->i_sb);
2901 static inline void file_end_write(struct file *file)
2903 if (!S_ISREG(file_inode(file)->i_mode))
2905 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2909 * get_write_access() gets write permission for a file.
2910 * put_write_access() releases this write permission.
2911 * This is used for regular files.
2912 * We cannot support write (and maybe mmap read-write shared) accesses and
2913 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2914 * can have the following values:
2915 * 0: no writers, no VM_DENYWRITE mappings
2916 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2917 * > 0: (i_writecount) users are writing to the file.
2919 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2920 * except for the cases where we don't hold i_writecount yet. Then we need to
2921 * use {get,deny}_write_access() - these functions check the sign and refuse
2922 * to do the change if sign is wrong.
2924 static inline int get_write_access(struct inode *inode)
2926 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2928 static inline int deny_write_access(struct file *file)
2930 struct inode *inode = file_inode(file);
2931 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2933 static inline void put_write_access(struct inode * inode)
2935 atomic_dec(&inode->i_writecount);
2937 static inline void allow_write_access(struct file *file)
2940 atomic_inc(&file_inode(file)->i_writecount);
2942 static inline bool inode_is_open_for_write(const struct inode *inode)
2944 return atomic_read(&inode->i_writecount) > 0;
2947 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
2948 static inline void i_readcount_dec(struct inode *inode)
2950 BUG_ON(!atomic_read(&inode->i_readcount));
2951 atomic_dec(&inode->i_readcount);
2953 static inline void i_readcount_inc(struct inode *inode)
2955 atomic_inc(&inode->i_readcount);
2958 static inline void i_readcount_dec(struct inode *inode)
2962 static inline void i_readcount_inc(struct inode *inode)
2967 extern int do_pipe_flags(int *, int);
2969 extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
2970 ssize_t __kernel_read(struct file *file, void *buf, size_t count, loff_t *pos);
2971 extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
2972 extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
2973 extern struct file * open_exec(const char *);
2975 /* fs/dcache.c -- generic fs support functions */
2976 extern bool is_subdir(struct dentry *, struct dentry *);
2977 extern bool path_is_under(const struct path *, const struct path *);
2979 extern char *file_path(struct file *, char *, int);
2981 #include <linux/err.h>
2983 /* needed for stackable file system support */
2984 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2986 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2988 extern int inode_init_always(struct super_block *, struct inode *);
2989 extern void inode_init_once(struct inode *);
2990 extern void address_space_init_once(struct address_space *mapping);
2991 extern struct inode * igrab(struct inode *);
2992 extern ino_t iunique(struct super_block *, ino_t);
2993 extern int inode_needs_sync(struct inode *inode);
2994 extern int generic_delete_inode(struct inode *inode);
2995 static inline int generic_drop_inode(struct inode *inode)
2997 return !inode->i_nlink || inode_unhashed(inode);
2999 extern void d_mark_dontcache(struct inode *inode);
3001 extern struct inode *ilookup5_nowait(struct super_block *sb,
3002 unsigned long hashval, int (*test)(struct inode *, void *),
3004 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
3005 int (*test)(struct inode *, void *), void *data);
3006 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
3008 extern struct inode *inode_insert5(struct inode *inode, unsigned long hashval,
3009 int (*test)(struct inode *, void *),
3010 int (*set)(struct inode *, void *),
3012 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
3013 extern struct inode * iget_locked(struct super_block *, unsigned long);
3014 extern struct inode *find_inode_nowait(struct super_block *,
3016 int (*match)(struct inode *,
3017 unsigned long, void *),
3019 extern struct inode *find_inode_rcu(struct super_block *, unsigned long,
3020 int (*)(struct inode *, void *), void *);
3021 extern struct inode *find_inode_by_ino_rcu(struct super_block *, unsigned long);
3022 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
3023 extern int insert_inode_locked(struct inode *);
3024 #ifdef CONFIG_DEBUG_LOCK_ALLOC
3025 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
3027 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
3029 extern void unlock_new_inode(struct inode *);
3030 extern void discard_new_inode(struct inode *);
3031 extern unsigned int get_next_ino(void);
3032 extern void evict_inodes(struct super_block *sb);
3035 * Userspace may rely on the the inode number being non-zero. For example, glibc
3036 * simply ignores files with zero i_ino in unlink() and other places.
3038 * As an additional complication, if userspace was compiled with
3039 * _FILE_OFFSET_BITS=32 on a 64-bit kernel we'll only end up reading out the
3040 * lower 32 bits, so we need to check that those aren't zero explicitly. With
3041 * _FILE_OFFSET_BITS=64, this may cause some harmless false-negatives, but
3042 * better safe than sorry.
3044 static inline bool is_zero_ino(ino_t ino)
3046 return (u32)ino == 0;
3049 extern void __iget(struct inode * inode);
3050 extern void iget_failed(struct inode *);
3051 extern void clear_inode(struct inode *);
3052 extern void __destroy_inode(struct inode *);
3053 extern struct inode *new_inode_pseudo(struct super_block *sb);
3054 extern struct inode *new_inode(struct super_block *sb);
3055 extern void free_inode_nonrcu(struct inode *inode);
3056 extern int should_remove_suid(struct dentry *);
3057 extern int file_remove_privs(struct file *);
3059 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
3060 static inline void insert_inode_hash(struct inode *inode)
3062 __insert_inode_hash(inode, inode->i_ino);
3065 extern void __remove_inode_hash(struct inode *);
3066 static inline void remove_inode_hash(struct inode *inode)
3068 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
3069 __remove_inode_hash(inode);
3072 extern void inode_sb_list_add(struct inode *inode);
3074 extern int sb_set_blocksize(struct super_block *, int);
3075 extern int sb_min_blocksize(struct super_block *, int);
3077 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
3078 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
3079 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
3080 extern int generic_write_check_limits(struct file *file, loff_t pos,
3082 extern int generic_file_rw_checks(struct file *file_in, struct file *file_out);
3083 ssize_t filemap_read(struct kiocb *iocb, struct iov_iter *to,
3084 ssize_t already_read);
3085 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
3086 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
3087 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
3088 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
3089 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
3091 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
3093 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
3095 ssize_t vfs_iocb_iter_read(struct file *file, struct kiocb *iocb,
3096 struct iov_iter *iter);
3097 ssize_t vfs_iocb_iter_write(struct file *file, struct kiocb *iocb,
3098 struct iov_iter *iter);
3100 /* fs/block_dev.c */
3101 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
3102 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
3103 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
3105 extern void block_sync_page(struct page *page);
3108 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
3109 struct pipe_inode_info *, size_t, unsigned int);
3110 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
3111 struct file *, loff_t *, size_t, unsigned int);
3112 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
3113 struct file *out, loff_t *, size_t len, unsigned int flags);
3114 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
3115 loff_t *opos, size_t len, unsigned int flags);
3119 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
3120 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
3121 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
3122 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
3123 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
3124 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
3125 int whence, loff_t maxsize, loff_t eof);
3126 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
3127 int whence, loff_t size);
3128 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
3129 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
3130 extern int generic_file_open(struct inode * inode, struct file * filp);
3131 extern int nonseekable_open(struct inode * inode, struct file * filp);
3132 extern int stream_open(struct inode * inode, struct file * filp);
3135 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
3136 loff_t file_offset);
3139 /* need locking between buffered and direct access */
3142 /* filesystem does not support filling holes */
3143 DIO_SKIP_HOLES = 0x02,
3146 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
3147 struct block_device *bdev, struct iov_iter *iter,
3148 get_block_t get_block,
3149 dio_iodone_t end_io, dio_submit_t submit_io,
3152 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
3153 struct inode *inode,
3154 struct iov_iter *iter,
3155 get_block_t get_block)
3157 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
3158 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
3162 void inode_dio_wait(struct inode *inode);
3165 * inode_dio_begin - signal start of a direct I/O requests
3166 * @inode: inode the direct I/O happens on
3168 * This is called once we've finished processing a direct I/O request,
3169 * and is used to wake up callers waiting for direct I/O to be quiesced.
3171 static inline void inode_dio_begin(struct inode *inode)
3173 atomic_inc(&inode->i_dio_count);
3177 * inode_dio_end - signal finish of a direct I/O requests
3178 * @inode: inode the direct I/O happens on
3180 * This is called once we've finished processing a direct I/O request,
3181 * and is used to wake up callers waiting for direct I/O to be quiesced.
3183 static inline void inode_dio_end(struct inode *inode)
3185 if (atomic_dec_and_test(&inode->i_dio_count))
3186 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
3190 * Warn about a page cache invalidation failure diring a direct I/O write.
3192 void dio_warn_stale_pagecache(struct file *filp);
3194 extern void inode_set_flags(struct inode *inode, unsigned int flags,
3197 extern const struct file_operations generic_ro_fops;
3199 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3201 extern int readlink_copy(char __user *, int, const char *);
3202 extern int page_readlink(struct dentry *, char __user *, int);
3203 extern const char *page_get_link(struct dentry *, struct inode *,
3204 struct delayed_call *);
3205 extern void page_put_link(void *);
3206 extern int __page_symlink(struct inode *inode, const char *symname, int len,
3208 extern int page_symlink(struct inode *inode, const char *symname, int len);
3209 extern const struct inode_operations page_symlink_inode_operations;
3210 extern void kfree_link(void *);
3211 void generic_fillattr(struct user_namespace *, struct inode *, struct kstat *);
3212 extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
3213 extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
3214 void __inode_add_bytes(struct inode *inode, loff_t bytes);
3215 void inode_add_bytes(struct inode *inode, loff_t bytes);
3216 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
3217 void inode_sub_bytes(struct inode *inode, loff_t bytes);
3218 static inline loff_t __inode_get_bytes(struct inode *inode)
3220 return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes;
3222 loff_t inode_get_bytes(struct inode *inode);
3223 void inode_set_bytes(struct inode *inode, loff_t bytes);
3224 const char *simple_get_link(struct dentry *, struct inode *,
3225 struct delayed_call *);
3226 extern const struct inode_operations simple_symlink_inode_operations;
3228 extern int iterate_dir(struct file *, struct dir_context *);
3230 int vfs_fstatat(int dfd, const char __user *filename, struct kstat *stat,
3232 int vfs_fstat(int fd, struct kstat *stat);
3234 static inline int vfs_stat(const char __user *filename, struct kstat *stat)
3236 return vfs_fstatat(AT_FDCWD, filename, stat, 0);
3238 static inline int vfs_lstat(const char __user *name, struct kstat *stat)
3240 return vfs_fstatat(AT_FDCWD, name, stat, AT_SYMLINK_NOFOLLOW);
3243 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
3244 extern int vfs_readlink(struct dentry *, char __user *, int);
3246 extern struct file_system_type *get_filesystem(struct file_system_type *fs);
3247 extern void put_filesystem(struct file_system_type *fs);
3248 extern struct file_system_type *get_fs_type(const char *name);
3249 extern struct super_block *get_super(struct block_device *);
3250 extern struct super_block *get_active_super(struct block_device *bdev);
3251 extern void drop_super(struct super_block *sb);
3252 extern void drop_super_exclusive(struct super_block *sb);
3253 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
3254 extern void iterate_supers_type(struct file_system_type *,
3255 void (*)(struct super_block *, void *), void *);
3257 extern int dcache_dir_open(struct inode *, struct file *);
3258 extern int dcache_dir_close(struct inode *, struct file *);
3259 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
3260 extern int dcache_readdir(struct file *, struct dir_context *);
3261 extern int simple_setattr(struct user_namespace *, struct dentry *,
3263 extern int simple_getattr(struct user_namespace *, const struct path *,
3264 struct kstat *, u32, unsigned int);
3265 extern int simple_statfs(struct dentry *, struct kstatfs *);
3266 extern int simple_open(struct inode *inode, struct file *file);
3267 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3268 extern int simple_unlink(struct inode *, struct dentry *);
3269 extern int simple_rmdir(struct inode *, struct dentry *);
3270 extern int simple_rename(struct user_namespace *, struct inode *,
3271 struct dentry *, struct inode *, struct dentry *,
3273 extern void simple_recursive_removal(struct dentry *,
3274 void (*callback)(struct dentry *));
3275 extern int noop_fsync(struct file *, loff_t, loff_t, int);
3276 extern int noop_set_page_dirty(struct page *page);
3277 extern void noop_invalidatepage(struct page *page, unsigned int offset,
3278 unsigned int length);
3279 extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter);
3280 extern int simple_empty(struct dentry *);
3281 extern int simple_readpage(struct file *file, struct page *page);
3282 extern int simple_write_begin(struct file *file, struct address_space *mapping,
3283 loff_t pos, unsigned len, unsigned flags,
3284 struct page **pagep, void **fsdata);
3285 extern int simple_write_end(struct file *file, struct address_space *mapping,
3286 loff_t pos, unsigned len, unsigned copied,
3287 struct page *page, void *fsdata);
3288 extern int always_delete_dentry(const struct dentry *);
3289 extern struct inode *alloc_anon_inode(struct super_block *);
3290 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
3291 extern const struct dentry_operations simple_dentry_operations;
3293 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3294 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3295 extern const struct file_operations simple_dir_operations;
3296 extern const struct inode_operations simple_dir_inode_operations;
3297 extern void make_empty_dir_inode(struct inode *inode);
3298 extern bool is_empty_dir_inode(struct inode *inode);
3299 struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
3300 struct dentry *d_alloc_name(struct dentry *, const char *);
3301 extern int simple_fill_super(struct super_block *, unsigned long,
3302 const struct tree_descr *);
3303 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3304 extern void simple_release_fs(struct vfsmount **mount, int *count);
3306 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3307 loff_t *ppos, const void *from, size_t available);
3308 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3309 const void __user *from, size_t count);
3311 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3312 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3314 extern int generic_check_addressable(unsigned, u64);
3316 extern void generic_set_encrypted_ci_d_ops(struct dentry *dentry);
3318 #ifdef CONFIG_MIGRATION
3319 extern int buffer_migrate_page(struct address_space *,
3320 struct page *, struct page *,
3322 extern int buffer_migrate_page_norefs(struct address_space *,
3323 struct page *, struct page *,
3326 #define buffer_migrate_page NULL
3327 #define buffer_migrate_page_norefs NULL
3330 int setattr_prepare(struct user_namespace *, struct dentry *, struct iattr *);
3331 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3332 void setattr_copy(struct user_namespace *, struct inode *inode,
3333 const struct iattr *attr);
3335 extern int file_update_time(struct file *file);
3337 static inline bool vma_is_dax(const struct vm_area_struct *vma)
3339 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3342 static inline bool vma_is_fsdax(struct vm_area_struct *vma)
3344 struct inode *inode;
3346 if (!IS_ENABLED(CONFIG_FS_DAX) || !vma->vm_file)
3348 if (!vma_is_dax(vma))
3350 inode = file_inode(vma->vm_file);
3351 if (S_ISCHR(inode->i_mode))
3352 return false; /* device-dax */
3356 static inline int iocb_flags(struct file *file)
3359 if (file->f_flags & O_APPEND)
3361 if (file->f_flags & O_DIRECT)
3363 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3365 if (file->f_flags & __O_SYNC)
3370 static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
3372 int kiocb_flags = 0;
3374 /* make sure there's no overlap between RWF and private IOCB flags */
3375 BUILD_BUG_ON((__force int) RWF_SUPPORTED & IOCB_EVENTFD);
3379 if (unlikely(flags & ~RWF_SUPPORTED))
3382 if (flags & RWF_NOWAIT) {
3383 if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
3385 kiocb_flags |= IOCB_NOIO;
3387 kiocb_flags |= (__force int) (flags & RWF_SUPPORTED);
3388 if (flags & RWF_SYNC)
3389 kiocb_flags |= IOCB_DSYNC;
3391 ki->ki_flags |= kiocb_flags;
3395 static inline ino_t parent_ino(struct dentry *dentry)
3400 * Don't strictly need d_lock here? If the parent ino could change
3401 * then surely we'd have a deeper race in the caller?
3403 spin_lock(&dentry->d_lock);
3404 res = dentry->d_parent->d_inode->i_ino;
3405 spin_unlock(&dentry->d_lock);
3409 /* Transaction based IO helpers */
3412 * An argresp is stored in an allocated page and holds the
3413 * size of the argument or response, along with its content
3415 struct simple_transaction_argresp {
3420 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3422 char *simple_transaction_get(struct file *file, const char __user *buf,
3424 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3425 size_t size, loff_t *pos);
3426 int simple_transaction_release(struct inode *inode, struct file *file);
3428 void simple_transaction_set(struct file *file, size_t n);
3431 * simple attribute files
3433 * These attributes behave similar to those in sysfs:
3435 * Writing to an attribute immediately sets a value, an open file can be
3436 * written to multiple times.
3438 * Reading from an attribute creates a buffer from the value that might get
3439 * read with multiple read calls. When the attribute has been read
3440 * completely, no further read calls are possible until the file is opened
3443 * All attributes contain a text representation of a numeric value
3444 * that are accessed with the get() and set() functions.
3446 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3447 static int __fops ## _open(struct inode *inode, struct file *file) \
3449 __simple_attr_check_format(__fmt, 0ull); \
3450 return simple_attr_open(inode, file, __get, __set, __fmt); \
3452 static const struct file_operations __fops = { \
3453 .owner = THIS_MODULE, \
3454 .open = __fops ## _open, \
3455 .release = simple_attr_release, \
3456 .read = simple_attr_read, \
3457 .write = simple_attr_write, \
3458 .llseek = generic_file_llseek, \
3461 static inline __printf(1, 2)
3462 void __simple_attr_check_format(const char *fmt, ...)
3464 /* don't do anything, just let the compiler check the arguments; */
3467 int simple_attr_open(struct inode *inode, struct file *file,
3468 int (*get)(void *, u64 *), int (*set)(void *, u64),
3470 int simple_attr_release(struct inode *inode, struct file *file);
3471 ssize_t simple_attr_read(struct file *file, char __user *buf,
3472 size_t len, loff_t *ppos);
3473 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3474 size_t len, loff_t *ppos);
3477 int proc_nr_files(struct ctl_table *table, int write,
3478 void *buffer, size_t *lenp, loff_t *ppos);
3479 int proc_nr_dentry(struct ctl_table *table, int write,
3480 void *buffer, size_t *lenp, loff_t *ppos);
3481 int proc_nr_inodes(struct ctl_table *table, int write,
3482 void *buffer, size_t *lenp, loff_t *ppos);
3483 int __init get_filesystem_list(char *buf);
3485 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3486 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3488 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3489 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3490 (flag & __FMODE_NONOTIFY)))
3492 static inline bool is_sxid(umode_t mode)
3494 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3497 static inline int check_sticky(struct user_namespace *mnt_userns,
3498 struct inode *dir, struct inode *inode)
3500 if (!(dir->i_mode & S_ISVTX))
3503 return __check_sticky(mnt_userns, dir, inode);
3506 static inline void inode_has_no_xattr(struct inode *inode)
3508 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
3509 inode->i_flags |= S_NOSEC;
3512 static inline bool is_root_inode(struct inode *inode)
3514 return inode == inode->i_sb->s_root->d_inode;
3517 static inline bool dir_emit(struct dir_context *ctx,
3518 const char *name, int namelen,
3519 u64 ino, unsigned type)
3521 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3523 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3525 return ctx->actor(ctx, ".", 1, ctx->pos,
3526 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3528 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3530 return ctx->actor(ctx, "..", 2, ctx->pos,
3531 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3533 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3535 if (ctx->pos == 0) {
3536 if (!dir_emit_dot(file, ctx))
3540 if (ctx->pos == 1) {
3541 if (!dir_emit_dotdot(file, ctx))
3547 static inline bool dir_relax(struct inode *inode)
3549 inode_unlock(inode);
3551 return !IS_DEADDIR(inode);
3554 static inline bool dir_relax_shared(struct inode *inode)
3556 inode_unlock_shared(inode);
3557 inode_lock_shared(inode);
3558 return !IS_DEADDIR(inode);
3561 extern bool path_noexec(const struct path *path);
3562 extern void inode_nohighmem(struct inode *inode);
3565 extern int vfs_fadvise(struct file *file, loff_t offset, loff_t len,
3567 extern int generic_fadvise(struct file *file, loff_t offset, loff_t len,
3570 int vfs_ioc_setflags_prepare(struct inode *inode, unsigned int oldflags,
3571 unsigned int flags);
3573 int vfs_ioc_fssetxattr_check(struct inode *inode, const struct fsxattr *old_fa,
3574 struct fsxattr *fa);
3576 static inline void simple_fill_fsxattr(struct fsxattr *fa, __u32 xflags)
3578 memset(fa, 0, sizeof(*fa));
3579 fa->fsx_xflags = xflags;
3583 * Flush file data before changing attributes. Caller must hold any locks
3584 * required to prevent further writes to this file until we're done setting
3587 static inline int inode_drain_writes(struct inode *inode)
3589 inode_dio_wait(inode);
3590 return filemap_write_and_wait(inode->i_mapping);
3593 #endif /* _LINUX_FS_H */