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>
43 #include <asm/byteorder.h>
44 #include <uapi/linux/fs.h>
46 struct backing_dev_info;
49 struct export_operations;
50 struct fiemap_extent_info;
55 struct pipe_inode_info;
56 struct poll_table_struct;
58 struct vm_area_struct;
61 struct swap_info_struct;
63 struct workqueue_struct;
66 struct fscrypt_operations;
68 struct fsverity_operations;
70 struct fs_parameter_spec;
72 extern void __init inode_init(void);
73 extern void __init inode_init_early(void);
74 extern void __init files_init(void);
75 extern void __init files_maxfiles_init(void);
77 extern struct files_stat_struct files_stat;
78 extern unsigned long get_max_files(void);
79 extern unsigned int sysctl_nr_open;
80 extern struct inodes_stat_t inodes_stat;
81 extern int leases_enable, lease_break_time;
82 extern int sysctl_protected_symlinks;
83 extern int sysctl_protected_hardlinks;
84 extern int sysctl_protected_fifos;
85 extern int sysctl_protected_regular;
87 typedef __kernel_rwf_t rwf_t;
90 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
91 struct buffer_head *bh_result, int create);
92 typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
93 ssize_t bytes, void *private);
95 #define MAY_EXEC 0x00000001
96 #define MAY_WRITE 0x00000002
97 #define MAY_READ 0x00000004
98 #define MAY_APPEND 0x00000008
99 #define MAY_ACCESS 0x00000010
100 #define MAY_OPEN 0x00000020
101 #define MAY_CHDIR 0x00000040
102 /* called from RCU mode, don't block */
103 #define MAY_NOT_BLOCK 0x00000080
106 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
107 * to O_WRONLY and O_RDWR via the strange trick in do_dentry_open()
110 /* file is open for reading */
111 #define FMODE_READ ((__force fmode_t)0x1)
112 /* file is open for writing */
113 #define FMODE_WRITE ((__force fmode_t)0x2)
114 /* file is seekable */
115 #define FMODE_LSEEK ((__force fmode_t)0x4)
116 /* file can be accessed using pread */
117 #define FMODE_PREAD ((__force fmode_t)0x8)
118 /* file can be accessed using pwrite */
119 #define FMODE_PWRITE ((__force fmode_t)0x10)
120 /* File is opened for execution with sys_execve / sys_uselib */
121 #define FMODE_EXEC ((__force fmode_t)0x20)
122 /* File is opened with O_NDELAY (only set for block devices) */
123 #define FMODE_NDELAY ((__force fmode_t)0x40)
124 /* File is opened with O_EXCL (only set for block devices) */
125 #define FMODE_EXCL ((__force fmode_t)0x80)
126 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
127 (specialy hack for floppy.c) */
128 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
129 /* 32bit hashes as llseek() offset (for directories) */
130 #define FMODE_32BITHASH ((__force fmode_t)0x200)
131 /* 64bit hashes as llseek() offset (for directories) */
132 #define FMODE_64BITHASH ((__force fmode_t)0x400)
135 * Don't update ctime and mtime.
137 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
138 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
140 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
142 /* Expect random access pattern */
143 #define FMODE_RANDOM ((__force fmode_t)0x1000)
145 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
146 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
148 /* File is opened with O_PATH; almost nothing can be done with it */
149 #define FMODE_PATH ((__force fmode_t)0x4000)
151 /* File needs atomic accesses to f_pos */
152 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
153 /* Write access to underlying fs */
154 #define FMODE_WRITER ((__force fmode_t)0x10000)
155 /* Has read method(s) */
156 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
157 /* Has write method(s) */
158 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
160 #define FMODE_OPENED ((__force fmode_t)0x80000)
161 #define FMODE_CREATED ((__force fmode_t)0x100000)
163 /* File is stream-like */
164 #define FMODE_STREAM ((__force fmode_t)0x200000)
166 /* File was opened by fanotify and shouldn't generate fanotify events */
167 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
169 /* File is capable of returning -EAGAIN if I/O will block */
170 #define FMODE_NOWAIT ((__force fmode_t)0x8000000)
172 /* File represents mount that needs unmounting */
173 #define FMODE_NEED_UNMOUNT ((__force fmode_t)0x10000000)
175 /* File does not contribute to nr_files count */
176 #define FMODE_NOACCOUNT ((__force fmode_t)0x20000000)
178 /* File supports async buffered reads */
179 #define FMODE_BUF_RASYNC ((__force fmode_t)0x40000000)
182 * Attribute flags. These should be or-ed together to figure out what
185 #define ATTR_MODE (1 << 0)
186 #define ATTR_UID (1 << 1)
187 #define ATTR_GID (1 << 2)
188 #define ATTR_SIZE (1 << 3)
189 #define ATTR_ATIME (1 << 4)
190 #define ATTR_MTIME (1 << 5)
191 #define ATTR_CTIME (1 << 6)
192 #define ATTR_ATIME_SET (1 << 7)
193 #define ATTR_MTIME_SET (1 << 8)
194 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
195 #define ATTR_KILL_SUID (1 << 11)
196 #define ATTR_KILL_SGID (1 << 12)
197 #define ATTR_FILE (1 << 13)
198 #define ATTR_KILL_PRIV (1 << 14)
199 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
200 #define ATTR_TIMES_SET (1 << 16)
201 #define ATTR_TOUCH (1 << 17)
204 * Whiteout is represented by a char device. The following constants define the
205 * mode and device number to use.
207 #define WHITEOUT_MODE 0
208 #define WHITEOUT_DEV 0
211 * This is the Inode Attributes structure, used for notify_change(). It
212 * uses the above definitions as flags, to know which values have changed.
213 * Also, in this manner, a Filesystem can look at only the values it cares
214 * about. Basically, these are the attributes that the VFS layer can
215 * request to change from the FS layer.
217 * Derek Atkins <warlord@MIT.EDU> 94-10-20
220 unsigned int ia_valid;
225 struct timespec64 ia_atime;
226 struct timespec64 ia_mtime;
227 struct timespec64 ia_ctime;
230 * Not an attribute, but an auxiliary info for filesystems wanting to
231 * implement an ftruncate() like method. NOTE: filesystem should
232 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
234 struct file *ia_file;
238 * Includes for diskquotas.
240 #include <linux/quota.h>
243 * Maximum number of layers of fs stack. Needs to be limited to
244 * prevent kernel stack overflow
246 #define FILESYSTEM_MAX_STACK_DEPTH 2
249 * enum positive_aop_returns - aop return codes with specific semantics
251 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
252 * completed, that the page is still locked, and
253 * should be considered active. The VM uses this hint
254 * to return the page to the active list -- it won't
255 * be a candidate for writeback again in the near
256 * future. Other callers must be careful to unlock
257 * the page if they get this return. Returned by
260 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
261 * unlocked it and the page might have been truncated.
262 * The caller should back up to acquiring a new page and
263 * trying again. The aop will be taking reasonable
264 * precautions not to livelock. If the caller held a page
265 * reference, it should drop it before retrying. Returned
268 * address_space_operation functions return these large constants to indicate
269 * special semantics to the caller. These are much larger than the bytes in a
270 * page to allow for functions that return the number of bytes operated on in a
274 enum positive_aop_returns {
275 AOP_WRITEPAGE_ACTIVATE = 0x80000,
276 AOP_TRUNCATED_PAGE = 0x80001,
279 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
280 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
281 * helper code (eg buffer layer)
282 * to clear GFP_FS from alloc */
285 * oh the beauties of C type declarations.
288 struct address_space;
289 struct writeback_control;
290 struct readahead_control;
293 * Write life time hint values.
294 * Stored in struct inode as u8.
297 WRITE_LIFE_NOT_SET = 0,
298 WRITE_LIFE_NONE = RWH_WRITE_LIFE_NONE,
299 WRITE_LIFE_SHORT = RWH_WRITE_LIFE_SHORT,
300 WRITE_LIFE_MEDIUM = RWH_WRITE_LIFE_MEDIUM,
301 WRITE_LIFE_LONG = RWH_WRITE_LIFE_LONG,
302 WRITE_LIFE_EXTREME = RWH_WRITE_LIFE_EXTREME,
305 /* Match RWF_* bits to IOCB bits */
306 #define IOCB_HIPRI (__force int) RWF_HIPRI
307 #define IOCB_DSYNC (__force int) RWF_DSYNC
308 #define IOCB_SYNC (__force int) RWF_SYNC
309 #define IOCB_NOWAIT (__force int) RWF_NOWAIT
310 #define IOCB_APPEND (__force int) RWF_APPEND
312 /* non-RWF related bits - start at 16 */
313 #define IOCB_EVENTFD (1 << 16)
314 #define IOCB_DIRECT (1 << 17)
315 #define IOCB_WRITE (1 << 18)
316 /* iocb->ki_waitq is valid */
317 #define IOCB_WAITQ (1 << 19)
318 #define IOCB_NOIO (1 << 20)
321 struct file *ki_filp;
323 /* The 'ki_filp' pointer is shared in a union for aio */
324 randomized_struct_fields_start
327 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
331 u16 ki_ioprio; /* See linux/ioprio.h */
333 unsigned int ki_cookie; /* for ->iopoll */
334 struct wait_page_queue *ki_waitq; /* for async buffered IO */
337 randomized_struct_fields_end
340 static inline bool is_sync_kiocb(struct kiocb *kiocb)
342 return kiocb->ki_complete == NULL;
346 * "descriptor" for what we're up to with a read.
347 * This allows us to use the same read code yet
348 * have multiple different users of the data that
349 * we read from a file.
351 * The simplest case just copies the data to user
364 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
365 unsigned long, unsigned long);
367 struct address_space_operations {
368 int (*writepage)(struct page *page, struct writeback_control *wbc);
369 int (*readpage)(struct file *, struct page *);
371 /* Write back some dirty pages from this mapping. */
372 int (*writepages)(struct address_space *, struct writeback_control *);
374 /* Set a page dirty. Return true if this dirtied it */
375 int (*set_page_dirty)(struct page *page);
378 * Reads in the requested pages. Unlike ->readpage(), this is
379 * PURELY used for read-ahead!.
381 int (*readpages)(struct file *filp, struct address_space *mapping,
382 struct list_head *pages, unsigned nr_pages);
383 void (*readahead)(struct readahead_control *);
385 int (*write_begin)(struct file *, struct address_space *mapping,
386 loff_t pos, unsigned len, unsigned flags,
387 struct page **pagep, void **fsdata);
388 int (*write_end)(struct file *, struct address_space *mapping,
389 loff_t pos, unsigned len, unsigned copied,
390 struct page *page, void *fsdata);
392 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
393 sector_t (*bmap)(struct address_space *, sector_t);
394 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
395 int (*releasepage) (struct page *, gfp_t);
396 void (*freepage)(struct page *);
397 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
399 * migrate the contents of a page to the specified target. If
400 * migrate_mode is MIGRATE_ASYNC, it must not block.
402 int (*migratepage) (struct address_space *,
403 struct page *, struct page *, enum migrate_mode);
404 bool (*isolate_page)(struct page *, isolate_mode_t);
405 void (*putback_page)(struct page *);
406 int (*launder_page) (struct page *);
407 int (*is_partially_uptodate) (struct page *, unsigned long,
409 void (*is_dirty_writeback) (struct page *, bool *, bool *);
410 int (*error_remove_page)(struct address_space *, struct page *);
412 /* swapfile support */
413 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
415 void (*swap_deactivate)(struct file *file);
418 extern const struct address_space_operations empty_aops;
421 * pagecache_write_begin/pagecache_write_end must be used by general code
422 * to write into the pagecache.
424 int pagecache_write_begin(struct file *, struct address_space *mapping,
425 loff_t pos, unsigned len, unsigned flags,
426 struct page **pagep, void **fsdata);
428 int pagecache_write_end(struct file *, struct address_space *mapping,
429 loff_t pos, unsigned len, unsigned copied,
430 struct page *page, void *fsdata);
433 * struct address_space - Contents of a cacheable, mappable object.
434 * @host: Owner, either the inode or the block_device.
435 * @i_pages: Cached pages.
436 * @gfp_mask: Memory allocation flags to use for allocating pages.
437 * @i_mmap_writable: Number of VM_SHARED mappings.
438 * @nr_thps: Number of THPs in the pagecache (non-shmem only).
439 * @i_mmap: Tree of private and shared mappings.
440 * @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable.
441 * @nrpages: Number of page entries, protected by the i_pages lock.
442 * @nrexceptional: Shadow or DAX entries, protected by the i_pages lock.
443 * @writeback_index: Writeback starts here.
445 * @flags: Error bits and flags (AS_*).
446 * @wb_err: The most recent error which has occurred.
447 * @private_lock: For use by the owner of the address_space.
448 * @private_list: For use by the owner of the address_space.
449 * @private_data: For use by the owner of the address_space.
451 struct address_space {
453 struct xarray i_pages;
455 atomic_t i_mmap_writable;
456 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
457 /* number of thp, only for non-shmem files */
460 struct rb_root_cached i_mmap;
461 struct rw_semaphore i_mmap_rwsem;
462 unsigned long nrpages;
463 unsigned long nrexceptional;
464 pgoff_t writeback_index;
465 const struct address_space_operations *a_ops;
468 spinlock_t private_lock;
469 struct list_head private_list;
471 } __attribute__((aligned(sizeof(long)))) __randomize_layout;
473 * On most architectures that alignment is already the case; but
474 * must be enforced here for CRIS, to let the least significant bit
475 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
478 /* XArray tags, for tagging dirty and writeback pages in the pagecache. */
479 #define PAGECACHE_TAG_DIRTY XA_MARK_0
480 #define PAGECACHE_TAG_WRITEBACK XA_MARK_1
481 #define PAGECACHE_TAG_TOWRITE XA_MARK_2
484 * Returns true if any of the pages in the mapping are marked with the tag.
486 static inline bool mapping_tagged(struct address_space *mapping, xa_mark_t tag)
488 return xa_marked(&mapping->i_pages, tag);
491 static inline void i_mmap_lock_write(struct address_space *mapping)
493 down_write(&mapping->i_mmap_rwsem);
496 static inline int i_mmap_trylock_write(struct address_space *mapping)
498 return down_write_trylock(&mapping->i_mmap_rwsem);
501 static inline void i_mmap_unlock_write(struct address_space *mapping)
503 up_write(&mapping->i_mmap_rwsem);
506 static inline void i_mmap_lock_read(struct address_space *mapping)
508 down_read(&mapping->i_mmap_rwsem);
511 static inline void i_mmap_unlock_read(struct address_space *mapping)
513 up_read(&mapping->i_mmap_rwsem);
516 static inline void i_mmap_assert_locked(struct address_space *mapping)
518 lockdep_assert_held(&mapping->i_mmap_rwsem);
521 static inline void i_mmap_assert_write_locked(struct address_space *mapping)
523 lockdep_assert_held_write(&mapping->i_mmap_rwsem);
527 * Might pages of this file be mapped into userspace?
529 static inline int mapping_mapped(struct address_space *mapping)
531 return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root);
535 * Might pages of this file have been modified in userspace?
536 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap
537 * marks vma as VM_SHARED if it is shared, and the file was opened for
538 * writing i.e. vma may be mprotected writable even if now readonly.
540 * If i_mmap_writable is negative, no new writable mappings are allowed. You
541 * can only deny writable mappings, if none exists right now.
543 static inline int mapping_writably_mapped(struct address_space *mapping)
545 return atomic_read(&mapping->i_mmap_writable) > 0;
548 static inline int mapping_map_writable(struct address_space *mapping)
550 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
554 static inline void mapping_unmap_writable(struct address_space *mapping)
556 atomic_dec(&mapping->i_mmap_writable);
559 static inline int mapping_deny_writable(struct address_space *mapping)
561 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
565 static inline void mapping_allow_writable(struct address_space *mapping)
567 atomic_inc(&mapping->i_mmap_writable);
571 * Use sequence counter to get consistent i_size on 32-bit processors.
573 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
574 #include <linux/seqlock.h>
575 #define __NEED_I_SIZE_ORDERED
576 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
578 #define i_size_ordered_init(inode) do { } while (0)
582 #define ACL_NOT_CACHED ((void *)(-1))
583 #define ACL_DONT_CACHE ((void *)(-3))
585 static inline struct posix_acl *
586 uncached_acl_sentinel(struct task_struct *task)
588 return (void *)task + 1;
592 is_uncached_acl(struct posix_acl *acl)
594 return (long)acl & 1;
597 #define IOP_FASTPERM 0x0001
598 #define IOP_LOOKUP 0x0002
599 #define IOP_NOFOLLOW 0x0004
600 #define IOP_XATTR 0x0008
601 #define IOP_DEFAULT_READLINK 0x0010
603 struct fsnotify_mark_connector;
606 * Keep mostly read-only and often accessed (especially for
607 * the RCU path lookup and 'stat' data) fields at the beginning
608 * of the 'struct inode'
612 unsigned short i_opflags;
615 unsigned int i_flags;
617 #ifdef CONFIG_FS_POSIX_ACL
618 struct posix_acl *i_acl;
619 struct posix_acl *i_default_acl;
622 const struct inode_operations *i_op;
623 struct super_block *i_sb;
624 struct address_space *i_mapping;
626 #ifdef CONFIG_SECURITY
630 /* Stat data, not accessed from path walking */
633 * Filesystems may only read i_nlink directly. They shall use the
634 * following functions for modification:
636 * (set|clear|inc|drop)_nlink
637 * inode_(inc|dec)_link_count
640 const unsigned int i_nlink;
641 unsigned int __i_nlink;
645 struct timespec64 i_atime;
646 struct timespec64 i_mtime;
647 struct timespec64 i_ctime;
648 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
649 unsigned short i_bytes;
654 #ifdef __NEED_I_SIZE_ORDERED
655 seqcount_t i_size_seqcount;
659 unsigned long i_state;
660 struct rw_semaphore i_rwsem;
662 unsigned long dirtied_when; /* jiffies of first dirtying */
663 unsigned long dirtied_time_when;
665 struct hlist_node i_hash;
666 struct list_head i_io_list; /* backing dev IO list */
667 #ifdef CONFIG_CGROUP_WRITEBACK
668 struct bdi_writeback *i_wb; /* the associated cgroup wb */
670 /* foreign inode detection, see wbc_detach_inode() */
672 u16 i_wb_frn_avg_time;
673 u16 i_wb_frn_history;
675 struct list_head i_lru; /* inode LRU list */
676 struct list_head i_sb_list;
677 struct list_head i_wb_list; /* backing dev writeback list */
679 struct hlist_head i_dentry;
680 struct rcu_head i_rcu;
682 atomic64_t i_version;
683 atomic64_t i_sequence; /* see futex */
685 atomic_t i_dio_count;
686 atomic_t i_writecount;
687 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
688 atomic_t i_readcount; /* struct files open RO */
691 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
692 void (*free_inode)(struct inode *);
694 struct file_lock_context *i_flctx;
695 struct address_space i_data;
696 struct list_head i_devices;
698 struct pipe_inode_info *i_pipe;
706 #ifdef CONFIG_FSNOTIFY
707 __u32 i_fsnotify_mask; /* all events this inode cares about */
708 struct fsnotify_mark_connector __rcu *i_fsnotify_marks;
711 #ifdef CONFIG_FS_ENCRYPTION
712 struct fscrypt_info *i_crypt_info;
715 #ifdef CONFIG_FS_VERITY
716 struct fsverity_info *i_verity_info;
719 void *i_private; /* fs or device private pointer */
720 } __randomize_layout;
722 struct timespec64 timestamp_truncate(struct timespec64 t, struct inode *inode);
724 static inline unsigned int i_blocksize(const struct inode *node)
726 return (1 << node->i_blkbits);
729 static inline int inode_unhashed(struct inode *inode)
731 return hlist_unhashed(&inode->i_hash);
735 * __mark_inode_dirty expects inodes to be hashed. Since we don't
736 * want special inodes in the fileset inode space, we make them
737 * appear hashed, but do not put on any lists. hlist_del()
738 * will work fine and require no locking.
740 static inline void inode_fake_hash(struct inode *inode)
742 hlist_add_fake(&inode->i_hash);
746 * inode->i_mutex nesting subclasses for the lock validator:
748 * 0: the object of the current VFS operation
752 * 4: second non-directory
753 * 5: second parent (when locking independent directories in rename)
755 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
756 * non-directories at once.
758 * The locking order between these classes is
759 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
761 enum inode_i_mutex_lock_class
771 static inline void inode_lock(struct inode *inode)
773 down_write(&inode->i_rwsem);
776 static inline void inode_unlock(struct inode *inode)
778 up_write(&inode->i_rwsem);
781 static inline void inode_lock_shared(struct inode *inode)
783 down_read(&inode->i_rwsem);
786 static inline void inode_unlock_shared(struct inode *inode)
788 up_read(&inode->i_rwsem);
791 static inline int inode_trylock(struct inode *inode)
793 return down_write_trylock(&inode->i_rwsem);
796 static inline int inode_trylock_shared(struct inode *inode)
798 return down_read_trylock(&inode->i_rwsem);
801 static inline int inode_is_locked(struct inode *inode)
803 return rwsem_is_locked(&inode->i_rwsem);
806 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
808 down_write_nested(&inode->i_rwsem, subclass);
811 static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass)
813 down_read_nested(&inode->i_rwsem, subclass);
816 void lock_two_nondirectories(struct inode *, struct inode*);
817 void unlock_two_nondirectories(struct inode *, struct inode*);
820 * NOTE: in a 32bit arch with a preemptable kernel and
821 * an UP compile the i_size_read/write must be atomic
822 * with respect to the local cpu (unlike with preempt disabled),
823 * but they don't need to be atomic with respect to other cpus like in
824 * true SMP (so they need either to either locally disable irq around
825 * the read or for example on x86 they can be still implemented as a
826 * cmpxchg8b without the need of the lock prefix). For SMP compiles
827 * and 64bit archs it makes no difference if preempt is enabled or not.
829 static inline loff_t i_size_read(const struct inode *inode)
831 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
836 seq = read_seqcount_begin(&inode->i_size_seqcount);
837 i_size = inode->i_size;
838 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
840 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
844 i_size = inode->i_size;
848 return inode->i_size;
853 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
854 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
855 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
857 static inline void i_size_write(struct inode *inode, loff_t i_size)
859 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
861 write_seqcount_begin(&inode->i_size_seqcount);
862 inode->i_size = i_size;
863 write_seqcount_end(&inode->i_size_seqcount);
865 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
867 inode->i_size = i_size;
870 inode->i_size = i_size;
874 static inline unsigned iminor(const struct inode *inode)
876 return MINOR(inode->i_rdev);
879 static inline unsigned imajor(const struct inode *inode)
881 return MAJOR(inode->i_rdev);
885 rwlock_t lock; /* protects pid, uid, euid fields */
886 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
887 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
888 kuid_t uid, euid; /* uid/euid of process setting the owner */
889 int signum; /* posix.1b rt signal to be delivered on IO */
893 * Track a single file's readahead state
895 struct file_ra_state {
896 pgoff_t start; /* where readahead started */
897 unsigned int size; /* # of readahead pages */
898 unsigned int async_size; /* do asynchronous readahead when
899 there are only # of pages ahead */
901 unsigned int ra_pages; /* Maximum readahead window */
902 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
903 loff_t prev_pos; /* Cache last read() position */
907 * Check if @index falls in the readahead windows.
909 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
911 return (index >= ra->start &&
912 index < ra->start + ra->size);
917 struct llist_node fu_llist;
918 struct rcu_head fu_rcuhead;
921 struct inode *f_inode; /* cached value */
922 const struct file_operations *f_op;
925 * Protects f_ep, f_flags.
926 * Must not be taken from IRQ context.
929 enum rw_hint f_write_hint;
930 atomic_long_t f_count;
931 unsigned int f_flags;
933 struct mutex f_pos_lock;
935 struct fown_struct f_owner;
936 const struct cred *f_cred;
937 struct file_ra_state f_ra;
940 #ifdef CONFIG_SECURITY
943 /* needed for tty driver, and maybe others */
947 /* Used by fs/eventpoll.c to link all the hooks to this file */
948 struct hlist_head *f_ep;
949 #endif /* #ifdef CONFIG_EPOLL */
950 struct address_space *f_mapping;
952 errseq_t f_sb_err; /* for syncfs */
954 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
959 /* file identifier */
960 unsigned char f_handle[];
963 static inline struct file *get_file(struct file *f)
965 atomic_long_inc(&f->f_count);
968 #define get_file_rcu_many(x, cnt) \
969 atomic_long_add_unless(&(x)->f_count, (cnt), 0)
970 #define get_file_rcu(x) get_file_rcu_many((x), 1)
971 #define file_count(x) atomic_long_read(&(x)->f_count)
973 #define MAX_NON_LFS ((1UL<<31) - 1)
975 /* Page cache limit. The filesystems should put that into their s_maxbytes
976 limits, otherwise bad things can happen in VM. */
977 #if BITS_PER_LONG==32
978 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
979 #elif BITS_PER_LONG==64
980 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
985 #define FL_DELEG 4 /* NFSv4 delegation */
986 #define FL_ACCESS 8 /* not trying to lock, just looking */
987 #define FL_EXISTS 16 /* when unlocking, test for existence */
988 #define FL_LEASE 32 /* lease held on this file */
989 #define FL_CLOSE 64 /* unlock on close */
990 #define FL_SLEEP 128 /* A blocking lock */
991 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
992 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
993 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
994 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
996 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
999 * Special return value from posix_lock_file() and vfs_lock_file() for
1000 * asynchronous locking.
1002 #define FILE_LOCK_DEFERRED 1
1004 /* legacy typedef, should eventually be removed */
1005 typedef void *fl_owner_t;
1009 struct file_lock_operations {
1010 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1011 void (*fl_release_private)(struct file_lock *);
1014 struct lock_manager_operations {
1015 fl_owner_t (*lm_get_owner)(fl_owner_t);
1016 void (*lm_put_owner)(fl_owner_t);
1017 void (*lm_notify)(struct file_lock *); /* unblock callback */
1018 int (*lm_grant)(struct file_lock *, int);
1019 bool (*lm_break)(struct file_lock *);
1020 int (*lm_change)(struct file_lock *, int, struct list_head *);
1021 void (*lm_setup)(struct file_lock *, void **);
1022 bool (*lm_breaker_owns_lease)(struct file_lock *);
1025 struct lock_manager {
1026 struct list_head list;
1028 * NFSv4 and up also want opens blocked during the grace period;
1035 void locks_start_grace(struct net *, struct lock_manager *);
1036 void locks_end_grace(struct lock_manager *);
1037 bool locks_in_grace(struct net *);
1038 bool opens_in_grace(struct net *);
1040 /* that will die - we need it for nfs_lock_info */
1041 #include <linux/nfs_fs_i.h>
1044 * struct file_lock represents a generic "file lock". It's used to represent
1045 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1046 * note that the same struct is used to represent both a request for a lock and
1047 * the lock itself, but the same object is never used for both.
1049 * FIXME: should we create a separate "struct lock_request" to help distinguish
1052 * The varous i_flctx lists are ordered by:
1055 * 2) lock range start
1058 * Obviously, the last two criteria only matter for POSIX locks.
1061 struct file_lock *fl_blocker; /* The lock, that is blocking us */
1062 struct list_head fl_list; /* link into file_lock_context */
1063 struct hlist_node fl_link; /* node in global lists */
1064 struct list_head fl_blocked_requests; /* list of requests with
1065 * ->fl_blocker pointing here
1067 struct list_head fl_blocked_member; /* node in
1068 * ->fl_blocker->fl_blocked_requests
1070 fl_owner_t fl_owner;
1071 unsigned int fl_flags;
1072 unsigned char fl_type;
1073 unsigned int fl_pid;
1074 int fl_link_cpu; /* what cpu's list is this on? */
1075 wait_queue_head_t fl_wait;
1076 struct file *fl_file;
1080 struct fasync_struct * fl_fasync; /* for lease break notifications */
1081 /* for lease breaks: */
1082 unsigned long fl_break_time;
1083 unsigned long fl_downgrade_time;
1085 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1086 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1088 struct nfs_lock_info nfs_fl;
1089 struct nfs4_lock_info nfs4_fl;
1091 struct list_head link; /* link in AFS vnode's pending_locks list */
1092 int state; /* state of grant or error if -ve */
1093 unsigned int debug_id;
1096 } __randomize_layout;
1098 struct file_lock_context {
1099 spinlock_t flc_lock;
1100 struct list_head flc_flock;
1101 struct list_head flc_posix;
1102 struct list_head flc_lease;
1105 /* The following constant reflects the upper bound of the file/locking space */
1107 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1108 #define OFFSET_MAX INT_LIMIT(loff_t)
1109 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1112 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1114 #define locks_inode(f) file_inode(f)
1116 #ifdef CONFIG_FILE_LOCKING
1117 extern int fcntl_getlk(struct file *, unsigned int, struct flock *);
1118 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1121 #if BITS_PER_LONG == 32
1122 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *);
1123 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1127 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1128 extern int fcntl_getlease(struct file *filp);
1131 void locks_free_lock_context(struct inode *inode);
1132 void locks_free_lock(struct file_lock *fl);
1133 extern void locks_init_lock(struct file_lock *);
1134 extern struct file_lock * locks_alloc_lock(void);
1135 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1136 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1137 extern void locks_remove_posix(struct file *, fl_owner_t);
1138 extern void locks_remove_file(struct file *);
1139 extern void locks_release_private(struct file_lock *);
1140 extern void posix_test_lock(struct file *, struct file_lock *);
1141 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1142 extern int locks_delete_block(struct file_lock *);
1143 extern int vfs_test_lock(struct file *, struct file_lock *);
1144 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1145 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1146 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1147 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1148 extern void lease_get_mtime(struct inode *, struct timespec64 *time);
1149 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1150 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1151 extern int lease_modify(struct file_lock *, int, struct list_head *);
1153 struct notifier_block;
1154 extern int lease_register_notifier(struct notifier_block *);
1155 extern void lease_unregister_notifier(struct notifier_block *);
1157 struct files_struct;
1158 extern void show_fd_locks(struct seq_file *f,
1159 struct file *filp, struct files_struct *files);
1160 #else /* !CONFIG_FILE_LOCKING */
1161 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1162 struct flock __user *user)
1167 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1168 unsigned int cmd, struct flock __user *user)
1173 #if BITS_PER_LONG == 32
1174 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1175 struct flock64 __user *user)
1180 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1181 unsigned int cmd, struct flock64 __user *user)
1186 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1191 static inline int fcntl_getlease(struct file *filp)
1197 locks_free_lock_context(struct inode *inode)
1201 static inline void locks_init_lock(struct file_lock *fl)
1206 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1211 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1216 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1221 static inline void locks_remove_file(struct file *filp)
1226 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1231 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1232 struct file_lock *conflock)
1237 static inline int locks_delete_block(struct file_lock *waiter)
1242 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1247 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1248 struct file_lock *fl, struct file_lock *conf)
1253 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1258 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1263 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1268 static inline void lease_get_mtime(struct inode *inode,
1269 struct timespec64 *time)
1274 static inline int generic_setlease(struct file *filp, long arg,
1275 struct file_lock **flp, void **priv)
1280 static inline int vfs_setlease(struct file *filp, long arg,
1281 struct file_lock **lease, void **priv)
1286 static inline int lease_modify(struct file_lock *fl, int arg,
1287 struct list_head *dispose)
1292 struct files_struct;
1293 static inline void show_fd_locks(struct seq_file *f,
1294 struct file *filp, struct files_struct *files) {}
1295 #endif /* !CONFIG_FILE_LOCKING */
1297 static inline struct inode *file_inode(const struct file *f)
1302 static inline struct dentry *file_dentry(const struct file *file)
1304 return d_real(file->f_path.dentry, file_inode(file));
1307 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1309 return locks_lock_inode_wait(locks_inode(filp), fl);
1312 struct fasync_struct {
1316 struct fasync_struct *fa_next; /* singly linked list */
1317 struct file *fa_file;
1318 struct rcu_head fa_rcu;
1321 #define FASYNC_MAGIC 0x4601
1323 /* SMP safe fasync helpers: */
1324 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1325 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1326 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1327 extern struct fasync_struct *fasync_alloc(void);
1328 extern void fasync_free(struct fasync_struct *);
1330 /* can be called from interrupts */
1331 extern void kill_fasync(struct fasync_struct **, int, int);
1333 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1334 extern int f_setown(struct file *filp, unsigned long arg, int force);
1335 extern void f_delown(struct file *filp);
1336 extern pid_t f_getown(struct file *filp);
1337 extern int send_sigurg(struct fown_struct *fown);
1340 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1341 * represented in both.
1343 #define SB_RDONLY 1 /* Mount read-only */
1344 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1345 #define SB_NODEV 4 /* Disallow access to device special files */
1346 #define SB_NOEXEC 8 /* Disallow program execution */
1347 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1348 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1349 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1350 #define SB_NOATIME 1024 /* Do not update access times. */
1351 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1352 #define SB_SILENT 32768
1353 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1354 #define SB_INLINECRYPT (1<<17) /* Use blk-crypto for encrypted files */
1355 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1356 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1357 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1359 /* These sb flags are internal to the kernel */
1360 #define SB_SUBMOUNT (1<<26)
1361 #define SB_FORCE (1<<27)
1362 #define SB_NOSEC (1<<28)
1363 #define SB_BORN (1<<29)
1364 #define SB_ACTIVE (1<<30)
1365 #define SB_NOUSER (1<<31)
1367 /* These flags relate to encoding and casefolding */
1368 #define SB_ENC_STRICT_MODE_FL (1 << 0)
1370 #define sb_has_strict_encoding(sb) \
1371 (sb->s_encoding_flags & SB_ENC_STRICT_MODE_FL)
1377 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1378 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1379 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1380 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1381 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1384 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1385 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1386 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1387 #define SB_I_STABLE_WRITES 0x00000008 /* don't modify blks until WB is done */
1389 /* sb->s_iflags to limit user namespace mounts */
1390 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1391 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1392 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1394 #define SB_I_SKIP_SYNC 0x00000100 /* Skip superblock at global sync */
1396 /* Possible states of 'frozen' field */
1398 SB_UNFROZEN = 0, /* FS is unfrozen */
1399 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1400 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1401 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1402 * internal threads if needed) */
1403 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1406 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1409 int frozen; /* Is sb frozen? */
1410 wait_queue_head_t wait_unfrozen; /* wait for thaw */
1411 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1414 struct super_block {
1415 struct list_head s_list; /* Keep this first */
1416 dev_t s_dev; /* search index; _not_ kdev_t */
1417 unsigned char s_blocksize_bits;
1418 unsigned long s_blocksize;
1419 loff_t s_maxbytes; /* Max file size */
1420 struct file_system_type *s_type;
1421 const struct super_operations *s_op;
1422 const struct dquot_operations *dq_op;
1423 const struct quotactl_ops *s_qcop;
1424 const struct export_operations *s_export_op;
1425 unsigned long s_flags;
1426 unsigned long s_iflags; /* internal SB_I_* flags */
1427 unsigned long s_magic;
1428 struct dentry *s_root;
1429 struct rw_semaphore s_umount;
1432 #ifdef CONFIG_SECURITY
1435 const struct xattr_handler **s_xattr;
1436 #ifdef CONFIG_FS_ENCRYPTION
1437 const struct fscrypt_operations *s_cop;
1438 struct key *s_master_keys; /* master crypto keys in use */
1440 #ifdef CONFIG_FS_VERITY
1441 const struct fsverity_operations *s_vop;
1443 #ifdef CONFIG_UNICODE
1444 struct unicode_map *s_encoding;
1445 __u16 s_encoding_flags;
1447 struct hlist_bl_head s_roots; /* alternate root dentries for NFS */
1448 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1449 struct block_device *s_bdev;
1450 struct backing_dev_info *s_bdi;
1451 struct mtd_info *s_mtd;
1452 struct hlist_node s_instances;
1453 unsigned int s_quota_types; /* Bitmask of supported quota types */
1454 struct quota_info s_dquot; /* Diskquota specific options */
1456 struct sb_writers s_writers;
1459 * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and
1460 * s_fsnotify_marks together for cache efficiency. They are frequently
1461 * accessed and rarely modified.
1463 void *s_fs_info; /* Filesystem private info */
1465 /* Granularity of c/m/atime in ns (cannot be worse than a second) */
1467 /* Time limits for c/m/atime in seconds */
1468 time64_t s_time_min;
1469 time64_t s_time_max;
1470 #ifdef CONFIG_FSNOTIFY
1471 __u32 s_fsnotify_mask;
1472 struct fsnotify_mark_connector __rcu *s_fsnotify_marks;
1475 char s_id[32]; /* Informational name */
1476 uuid_t s_uuid; /* UUID */
1478 unsigned int s_max_links;
1482 * The next field is for VFS *only*. No filesystems have any business
1483 * even looking at it. You had been warned.
1485 struct mutex s_vfs_rename_mutex; /* Kludge */
1488 * Filesystem subtype. If non-empty the filesystem type field
1489 * in /proc/mounts will be "type.subtype"
1491 const char *s_subtype;
1493 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1496 * Saved pool identifier for cleancache (-1 means none)
1498 int cleancache_poolid;
1500 struct shrinker s_shrink; /* per-sb shrinker handle */
1502 /* Number of inodes with nlink == 0 but still referenced */
1503 atomic_long_t s_remove_count;
1505 /* Pending fsnotify inode refs */
1506 atomic_long_t s_fsnotify_inode_refs;
1508 /* Being remounted read-only */
1509 int s_readonly_remount;
1511 /* per-sb errseq_t for reporting writeback errors via syncfs */
1514 /* AIO completions deferred from interrupt context */
1515 struct workqueue_struct *s_dio_done_wq;
1516 struct hlist_head s_pins;
1519 * Owning user namespace and default context in which to
1520 * interpret filesystem uids, gids, quotas, device nodes,
1521 * xattrs and security labels.
1523 struct user_namespace *s_user_ns;
1526 * The list_lru structure is essentially just a pointer to a table
1527 * of per-node lru lists, each of which has its own spinlock.
1528 * There is no need to put them into separate cachelines.
1530 struct list_lru s_dentry_lru;
1531 struct list_lru s_inode_lru;
1532 struct rcu_head rcu;
1533 struct work_struct destroy_work;
1535 struct mutex s_sync_lock; /* sync serialisation lock */
1538 * Indicates how deep in a filesystem stack this SB is
1542 /* s_inode_list_lock protects s_inodes */
1543 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1544 struct list_head s_inodes; /* all inodes */
1546 spinlock_t s_inode_wblist_lock;
1547 struct list_head s_inodes_wb; /* writeback inodes */
1548 } __randomize_layout;
1550 /* Helper functions so that in most cases filesystems will
1551 * not need to deal directly with kuid_t and kgid_t and can
1552 * instead deal with the raw numeric values that are stored
1553 * in the filesystem.
1555 static inline uid_t i_uid_read(const struct inode *inode)
1557 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid);
1560 static inline gid_t i_gid_read(const struct inode *inode)
1562 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid);
1565 static inline void i_uid_write(struct inode *inode, uid_t uid)
1567 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid);
1570 static inline void i_gid_write(struct inode *inode, gid_t gid)
1572 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid);
1575 extern struct timespec64 current_time(struct inode *inode);
1578 * Snapshotting support.
1582 * These are internal functions, please use sb_start_{write,pagefault,intwrite}
1585 static inline void __sb_end_write(struct super_block *sb, int level)
1587 percpu_up_read(sb->s_writers.rw_sem + level-1);
1590 static inline void __sb_start_write(struct super_block *sb, int level)
1592 percpu_down_read(sb->s_writers.rw_sem + level - 1);
1595 static inline bool __sb_start_write_trylock(struct super_block *sb, int level)
1597 return percpu_down_read_trylock(sb->s_writers.rw_sem + level - 1);
1600 #define __sb_writers_acquired(sb, lev) \
1601 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1602 #define __sb_writers_release(sb, lev) \
1603 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1606 * sb_end_write - drop write access to a superblock
1607 * @sb: the super we wrote to
1609 * Decrement number of writers to the filesystem. Wake up possible waiters
1610 * wanting to freeze the filesystem.
1612 static inline void sb_end_write(struct super_block *sb)
1614 __sb_end_write(sb, SB_FREEZE_WRITE);
1618 * sb_end_pagefault - drop write access to a superblock from a page fault
1619 * @sb: the super we wrote to
1621 * Decrement number of processes handling write page fault to the filesystem.
1622 * Wake up possible waiters wanting to freeze the filesystem.
1624 static inline void sb_end_pagefault(struct super_block *sb)
1626 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1630 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1631 * @sb: the super we wrote to
1633 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1634 * waiters wanting to freeze the filesystem.
1636 static inline void sb_end_intwrite(struct super_block *sb)
1638 __sb_end_write(sb, SB_FREEZE_FS);
1642 * sb_start_write - get write access to a superblock
1643 * @sb: the super we write to
1645 * When a process wants to write data or metadata to a file system (i.e. dirty
1646 * a page or an inode), it should embed the operation in a sb_start_write() -
1647 * sb_end_write() pair to get exclusion against file system freezing. This
1648 * function increments number of writers preventing freezing. If the file
1649 * system is already frozen, the function waits until the file system is
1652 * Since freeze protection behaves as a lock, users have to preserve
1653 * ordering of freeze protection and other filesystem locks. Generally,
1654 * freeze protection should be the outermost lock. In particular, we have:
1657 * -> i_mutex (write path, truncate, directory ops, ...)
1658 * -> s_umount (freeze_super, thaw_super)
1660 static inline void sb_start_write(struct super_block *sb)
1662 __sb_start_write(sb, SB_FREEZE_WRITE);
1665 static inline bool sb_start_write_trylock(struct super_block *sb)
1667 return __sb_start_write_trylock(sb, SB_FREEZE_WRITE);
1671 * sb_start_pagefault - get write access to a superblock from a page fault
1672 * @sb: the super we write to
1674 * When a process starts handling write page fault, it should embed the
1675 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1676 * exclusion against file system freezing. This is needed since the page fault
1677 * is going to dirty a page. This function increments number of running page
1678 * faults preventing freezing. If the file system is already frozen, the
1679 * function waits until the file system is thawed.
1681 * Since page fault freeze protection behaves as a lock, users have to preserve
1682 * ordering of freeze protection and other filesystem locks. It is advised to
1683 * put sb_start_pagefault() close to mmap_lock in lock ordering. Page fault
1684 * handling code implies lock dependency:
1687 * -> sb_start_pagefault
1689 static inline void sb_start_pagefault(struct super_block *sb)
1691 __sb_start_write(sb, SB_FREEZE_PAGEFAULT);
1695 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1696 * @sb: the super we write to
1698 * This is the third level of protection against filesystem freezing. It is
1699 * free for use by a filesystem. The only requirement is that it must rank
1700 * below sb_start_pagefault.
1702 * For example filesystem can call sb_start_intwrite() when starting a
1703 * transaction which somewhat eases handling of freezing for internal sources
1704 * of filesystem changes (internal fs threads, discarding preallocation on file
1707 static inline void sb_start_intwrite(struct super_block *sb)
1709 __sb_start_write(sb, SB_FREEZE_FS);
1712 static inline bool sb_start_intwrite_trylock(struct super_block *sb)
1714 return __sb_start_write_trylock(sb, SB_FREEZE_FS);
1718 extern bool inode_owner_or_capable(const struct inode *inode);
1721 * VFS helper functions..
1723 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1724 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1725 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1726 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1727 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1728 extern int vfs_rmdir(struct inode *, struct dentry *);
1729 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1730 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1732 static inline int vfs_whiteout(struct inode *dir, struct dentry *dentry)
1734 return vfs_mknod(dir, dentry, S_IFCHR | WHITEOUT_MODE, WHITEOUT_DEV);
1737 extern struct dentry *vfs_tmpfile(struct dentry *dentry, umode_t mode,
1740 int vfs_mkobj(struct dentry *, umode_t,
1741 int (*f)(struct dentry *, umode_t, void *),
1744 int vfs_fchown(struct file *file, uid_t user, gid_t group);
1745 int vfs_fchmod(struct file *file, umode_t mode);
1746 int vfs_utimes(const struct path *path, struct timespec64 *times);
1748 extern long vfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
1750 #ifdef CONFIG_COMPAT
1751 extern long compat_ptr_ioctl(struct file *file, unsigned int cmd,
1754 #define compat_ptr_ioctl NULL
1758 * VFS file helper functions.
1760 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1762 extern bool may_open_dev(const struct path *path);
1765 * This is the "filldir" function type, used by readdir() to let
1766 * the kernel specify what kind of dirent layout it wants to have.
1767 * This allows the kernel to read directories into kernel space or
1768 * to have different dirent layouts depending on the binary type.
1771 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1774 struct dir_context {
1780 * These flags let !MMU mmap() govern direct device mapping vs immediate
1781 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1783 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1784 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1785 * NOMMU_MAP_READ: Can be mapped for reading
1786 * NOMMU_MAP_WRITE: Can be mapped for writing
1787 * NOMMU_MAP_EXEC: Can be mapped for execution
1789 #define NOMMU_MAP_COPY 0x00000001
1790 #define NOMMU_MAP_DIRECT 0x00000008
1791 #define NOMMU_MAP_READ VM_MAYREAD
1792 #define NOMMU_MAP_WRITE VM_MAYWRITE
1793 #define NOMMU_MAP_EXEC VM_MAYEXEC
1795 #define NOMMU_VMFLAGS \
1796 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1799 * These flags control the behavior of the remap_file_range function pointer.
1800 * If it is called with len == 0 that means "remap to end of source file".
1801 * See Documentation/filesystems/vfs.rst for more details about this call.
1803 * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate)
1804 * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request
1806 #define REMAP_FILE_DEDUP (1 << 0)
1807 #define REMAP_FILE_CAN_SHORTEN (1 << 1)
1810 * These flags signal that the caller is ok with altering various aspects of
1811 * the behavior of the remap operation. The changes must be made by the
1812 * implementation; the vfs remap helper functions can take advantage of them.
1813 * Flags in this category exist to preserve the quirky behavior of the hoisted
1814 * btrfs clone/dedupe ioctls.
1816 #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN)
1820 struct file_operations {
1821 struct module *owner;
1822 loff_t (*llseek) (struct file *, loff_t, int);
1823 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1824 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1825 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1826 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1827 int (*iopoll)(struct kiocb *kiocb, bool spin);
1828 int (*iterate) (struct file *, struct dir_context *);
1829 int (*iterate_shared) (struct file *, struct dir_context *);
1830 __poll_t (*poll) (struct file *, struct poll_table_struct *);
1831 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1832 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1833 int (*mmap) (struct file *, struct vm_area_struct *);
1834 unsigned long mmap_supported_flags;
1835 int (*open) (struct inode *, struct file *);
1836 int (*flush) (struct file *, fl_owner_t id);
1837 int (*release) (struct inode *, struct file *);
1838 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1839 int (*fasync) (int, struct file *, int);
1840 int (*lock) (struct file *, int, struct file_lock *);
1841 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1842 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1843 int (*check_flags)(int);
1844 int (*flock) (struct file *, int, struct file_lock *);
1845 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1846 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1847 int (*setlease)(struct file *, long, struct file_lock **, void **);
1848 long (*fallocate)(struct file *file, int mode, loff_t offset,
1850 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1852 unsigned (*mmap_capabilities)(struct file *);
1854 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1855 loff_t, size_t, unsigned int);
1856 loff_t (*remap_file_range)(struct file *file_in, loff_t pos_in,
1857 struct file *file_out, loff_t pos_out,
1858 loff_t len, unsigned int remap_flags);
1859 int (*fadvise)(struct file *, loff_t, loff_t, int);
1860 } __randomize_layout;
1862 struct inode_operations {
1863 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1864 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
1865 int (*permission) (struct inode *, int);
1866 struct posix_acl * (*get_acl)(struct inode *, int);
1868 int (*readlink) (struct dentry *, char __user *,int);
1870 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1871 int (*link) (struct dentry *,struct inode *,struct dentry *);
1872 int (*unlink) (struct inode *,struct dentry *);
1873 int (*symlink) (struct inode *,struct dentry *,const char *);
1874 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1875 int (*rmdir) (struct inode *,struct dentry *);
1876 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1877 int (*rename) (struct inode *, struct dentry *,
1878 struct inode *, struct dentry *, unsigned int);
1879 int (*setattr) (struct dentry *, struct iattr *);
1880 int (*getattr) (const struct path *, struct kstat *, u32, unsigned int);
1881 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1882 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1884 int (*update_time)(struct inode *, struct timespec64 *, int);
1885 int (*atomic_open)(struct inode *, struct dentry *,
1886 struct file *, unsigned open_flag,
1887 umode_t create_mode);
1888 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1889 int (*set_acl)(struct inode *, struct posix_acl *, int);
1890 } ____cacheline_aligned;
1892 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
1893 struct iov_iter *iter)
1895 return file->f_op->read_iter(kio, iter);
1898 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
1899 struct iov_iter *iter)
1901 return file->f_op->write_iter(kio, iter);
1904 static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
1906 return file->f_op->mmap(file, vma);
1909 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1910 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1911 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1912 loff_t, size_t, unsigned int);
1913 extern ssize_t generic_copy_file_range(struct file *file_in, loff_t pos_in,
1914 struct file *file_out, loff_t pos_out,
1915 size_t len, unsigned int flags);
1916 extern int generic_remap_file_range_prep(struct file *file_in, loff_t pos_in,
1917 struct file *file_out, loff_t pos_out,
1919 unsigned int remap_flags);
1920 extern loff_t do_clone_file_range(struct file *file_in, loff_t pos_in,
1921 struct file *file_out, loff_t pos_out,
1922 loff_t len, unsigned int remap_flags);
1923 extern loff_t vfs_clone_file_range(struct file *file_in, loff_t pos_in,
1924 struct file *file_out, loff_t pos_out,
1925 loff_t len, unsigned int remap_flags);
1926 extern int vfs_dedupe_file_range(struct file *file,
1927 struct file_dedupe_range *same);
1928 extern loff_t vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos,
1929 struct file *dst_file, loff_t dst_pos,
1930 loff_t len, unsigned int remap_flags);
1933 struct super_operations {
1934 struct inode *(*alloc_inode)(struct super_block *sb);
1935 void (*destroy_inode)(struct inode *);
1936 void (*free_inode)(struct inode *);
1938 void (*dirty_inode) (struct inode *, int flags);
1939 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1940 int (*drop_inode) (struct inode *);
1941 void (*evict_inode) (struct inode *);
1942 void (*put_super) (struct super_block *);
1943 int (*sync_fs)(struct super_block *sb, int wait);
1944 int (*freeze_super) (struct super_block *);
1945 int (*freeze_fs) (struct super_block *);
1946 int (*thaw_super) (struct super_block *);
1947 int (*unfreeze_fs) (struct super_block *);
1948 int (*statfs) (struct dentry *, struct kstatfs *);
1949 int (*remount_fs) (struct super_block *, int *, char *);
1950 void (*umount_begin) (struct super_block *);
1952 int (*show_options)(struct seq_file *, struct dentry *);
1953 int (*show_devname)(struct seq_file *, struct dentry *);
1954 int (*show_path)(struct seq_file *, struct dentry *);
1955 int (*show_stats)(struct seq_file *, struct dentry *);
1957 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1958 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1959 struct dquot **(*get_dquots)(struct inode *);
1961 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1962 long (*nr_cached_objects)(struct super_block *,
1963 struct shrink_control *);
1964 long (*free_cached_objects)(struct super_block *,
1965 struct shrink_control *);
1969 * Inode flags - they have no relation to superblock flags now
1971 #define S_SYNC (1 << 0) /* Writes are synced at once */
1972 #define S_NOATIME (1 << 1) /* Do not update access times */
1973 #define S_APPEND (1 << 2) /* Append-only file */
1974 #define S_IMMUTABLE (1 << 3) /* Immutable file */
1975 #define S_DEAD (1 << 4) /* removed, but still open directory */
1976 #define S_NOQUOTA (1 << 5) /* Inode is not counted to quota */
1977 #define S_DIRSYNC (1 << 6) /* Directory modifications are synchronous */
1978 #define S_NOCMTIME (1 << 7) /* Do not update file c/mtime */
1979 #define S_SWAPFILE (1 << 8) /* Do not truncate: swapon got its bmaps */
1980 #define S_PRIVATE (1 << 9) /* Inode is fs-internal */
1981 #define S_IMA (1 << 10) /* Inode has an associated IMA struct */
1982 #define S_AUTOMOUNT (1 << 11) /* Automount/referral quasi-directory */
1983 #define S_NOSEC (1 << 12) /* no suid or xattr security attributes */
1984 #ifdef CONFIG_FS_DAX
1985 #define S_DAX (1 << 13) /* Direct Access, avoiding the page cache */
1987 #define S_DAX 0 /* Make all the DAX code disappear */
1989 #define S_ENCRYPTED (1 << 14) /* Encrypted file (using fs/crypto/) */
1990 #define S_CASEFOLD (1 << 15) /* Casefolded file */
1991 #define S_VERITY (1 << 16) /* Verity file (using fs/verity/) */
1994 * Note that nosuid etc flags are inode-specific: setting some file-system
1995 * flags just means all the inodes inherit those flags by default. It might be
1996 * possible to override it selectively if you really wanted to with some
1997 * ioctl() that is not currently implemented.
1999 * Exception: SB_RDONLY is always applied to the entire file system.
2001 * Unfortunately, it is possible to change a filesystems flags with it mounted
2002 * with files in use. This means that all of the inodes will not have their
2003 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
2004 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
2006 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
2008 static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; }
2009 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
2010 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
2011 ((inode)->i_flags & S_SYNC))
2012 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
2013 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
2014 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
2015 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
2016 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
2018 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
2019 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
2020 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
2021 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
2023 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
2024 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
2025 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
2026 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
2027 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
2028 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
2029 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
2030 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
2031 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
2032 #define IS_CASEFOLDED(inode) ((inode)->i_flags & S_CASEFOLD)
2033 #define IS_VERITY(inode) ((inode)->i_flags & S_VERITY)
2035 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
2036 (inode)->i_rdev == WHITEOUT_DEV)
2038 static inline bool HAS_UNMAPPED_ID(struct inode *inode)
2040 return !uid_valid(inode->i_uid) || !gid_valid(inode->i_gid);
2043 static inline enum rw_hint file_write_hint(struct file *file)
2045 if (file->f_write_hint != WRITE_LIFE_NOT_SET)
2046 return file->f_write_hint;
2048 return file_inode(file)->i_write_hint;
2051 static inline int iocb_flags(struct file *file);
2053 static inline u16 ki_hint_validate(enum rw_hint hint)
2055 typeof(((struct kiocb *)0)->ki_hint) max_hint = -1;
2057 if (hint <= max_hint)
2062 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
2064 *kiocb = (struct kiocb) {
2066 .ki_flags = iocb_flags(filp),
2067 .ki_hint = ki_hint_validate(file_write_hint(filp)),
2068 .ki_ioprio = get_current_ioprio(),
2072 static inline void kiocb_clone(struct kiocb *kiocb, struct kiocb *kiocb_src,
2075 *kiocb = (struct kiocb) {
2077 .ki_flags = kiocb_src->ki_flags,
2078 .ki_hint = kiocb_src->ki_hint,
2079 .ki_ioprio = kiocb_src->ki_ioprio,
2080 .ki_pos = kiocb_src->ki_pos,
2085 * Inode state bits. Protected by inode->i_lock
2087 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
2088 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
2090 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
2091 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
2092 * various stages of removing an inode.
2094 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2096 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
2097 * fdatasync(). i_atime is the usual cause.
2098 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
2099 * these changes separately from I_DIRTY_SYNC so that we
2100 * don't have to write inode on fdatasync() when only
2101 * mtime has changed in it.
2102 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
2103 * I_NEW Serves as both a mutex and completion notification.
2104 * New inodes set I_NEW. If two processes both create
2105 * the same inode, one of them will release its inode and
2106 * wait for I_NEW to be released before returning.
2107 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2108 * also cause waiting on I_NEW, without I_NEW actually
2109 * being set. find_inode() uses this to prevent returning
2110 * nearly-dead inodes.
2111 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2112 * is zero. I_FREEING must be set when I_WILL_FREE is
2114 * I_FREEING Set when inode is about to be freed but still has dirty
2115 * pages or buffers attached or the inode itself is still
2117 * I_CLEAR Added by clear_inode(). In this state the inode is
2118 * clean and can be destroyed. Inode keeps I_FREEING.
2120 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2121 * prohibited for many purposes. iget() must wait for
2122 * the inode to be completely released, then create it
2123 * anew. Other functions will just ignore such inodes,
2124 * if appropriate. I_NEW is used for waiting.
2126 * I_SYNC Writeback of inode is running. The bit is set during
2127 * data writeback, and cleared with a wakeup on the bit
2128 * address once it is done. The bit is also used to pin
2129 * the inode in memory for flusher thread.
2131 * I_REFERENCED Marks the inode as recently references on the LRU list.
2133 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2135 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2136 * synchronize competing switching instances and to tell
2137 * wb stat updates to grab the i_pages lock. See
2138 * inode_switch_wbs_work_fn() for details.
2140 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2141 * and work dirs among overlayfs mounts.
2143 * I_CREATING New object's inode in the middle of setting up.
2145 * I_DONTCACHE Evict inode as soon as it is not used anymore.
2147 * I_SYNC_QUEUED Inode is queued in b_io or b_more_io writeback lists.
2148 * Used to detect that mark_inode_dirty() should not move
2149 * inode between dirty lists.
2151 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2153 #define I_DIRTY_SYNC (1 << 0)
2154 #define I_DIRTY_DATASYNC (1 << 1)
2155 #define I_DIRTY_PAGES (1 << 2)
2157 #define I_NEW (1 << __I_NEW)
2158 #define I_WILL_FREE (1 << 4)
2159 #define I_FREEING (1 << 5)
2160 #define I_CLEAR (1 << 6)
2162 #define I_SYNC (1 << __I_SYNC)
2163 #define I_REFERENCED (1 << 8)
2164 #define __I_DIO_WAKEUP 9
2165 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2166 #define I_LINKABLE (1 << 10)
2167 #define I_DIRTY_TIME (1 << 11)
2168 #define I_WB_SWITCH (1 << 13)
2169 #define I_OVL_INUSE (1 << 14)
2170 #define I_CREATING (1 << 15)
2171 #define I_DONTCACHE (1 << 16)
2172 #define I_SYNC_QUEUED (1 << 17)
2174 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2175 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2176 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2178 extern void __mark_inode_dirty(struct inode *, int);
2179 static inline void mark_inode_dirty(struct inode *inode)
2181 __mark_inode_dirty(inode, I_DIRTY);
2184 static inline void mark_inode_dirty_sync(struct inode *inode)
2186 __mark_inode_dirty(inode, I_DIRTY_SYNC);
2189 extern void inc_nlink(struct inode *inode);
2190 extern void drop_nlink(struct inode *inode);
2191 extern void clear_nlink(struct inode *inode);
2192 extern void set_nlink(struct inode *inode, unsigned int nlink);
2194 static inline void inode_inc_link_count(struct inode *inode)
2197 mark_inode_dirty(inode);
2200 static inline void inode_dec_link_count(struct inode *inode)
2203 mark_inode_dirty(inode);
2206 enum file_time_flags {
2213 extern bool atime_needs_update(const struct path *, struct inode *);
2214 extern void touch_atime(const struct path *);
2215 static inline void file_accessed(struct file *file)
2217 if (!(file->f_flags & O_NOATIME))
2218 touch_atime(&file->f_path);
2221 extern int file_modified(struct file *file);
2223 int sync_inode(struct inode *inode, struct writeback_control *wbc);
2224 int sync_inode_metadata(struct inode *inode, int wait);
2226 struct file_system_type {
2229 #define FS_REQUIRES_DEV 1
2230 #define FS_BINARY_MOUNTDATA 2
2231 #define FS_HAS_SUBTYPE 4
2232 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2233 #define FS_DISALLOW_NOTIFY_PERM 16 /* Disable fanotify permission events */
2234 #define FS_THP_SUPPORT 8192 /* Remove once all fs converted */
2235 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2236 int (*init_fs_context)(struct fs_context *);
2237 const struct fs_parameter_spec *parameters;
2238 struct dentry *(*mount) (struct file_system_type *, int,
2239 const char *, void *);
2240 void (*kill_sb) (struct super_block *);
2241 struct module *owner;
2242 struct file_system_type * next;
2243 struct hlist_head fs_supers;
2245 struct lock_class_key s_lock_key;
2246 struct lock_class_key s_umount_key;
2247 struct lock_class_key s_vfs_rename_key;
2248 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2250 struct lock_class_key i_lock_key;
2251 struct lock_class_key i_mutex_key;
2252 struct lock_class_key i_mutex_dir_key;
2255 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2257 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2258 int flags, const char *dev_name, void *data,
2259 int (*fill_super)(struct super_block *, void *, int));
2260 extern struct dentry *mount_single(struct file_system_type *fs_type,
2261 int flags, void *data,
2262 int (*fill_super)(struct super_block *, void *, int));
2263 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2264 int flags, void *data,
2265 int (*fill_super)(struct super_block *, void *, int));
2266 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2267 void generic_shutdown_super(struct super_block *sb);
2268 void kill_block_super(struct super_block *sb);
2269 void kill_anon_super(struct super_block *sb);
2270 void kill_litter_super(struct super_block *sb);
2271 void deactivate_super(struct super_block *sb);
2272 void deactivate_locked_super(struct super_block *sb);
2273 int set_anon_super(struct super_block *s, void *data);
2274 int set_anon_super_fc(struct super_block *s, struct fs_context *fc);
2275 int get_anon_bdev(dev_t *);
2276 void free_anon_bdev(dev_t);
2277 struct super_block *sget_fc(struct fs_context *fc,
2278 int (*test)(struct super_block *, struct fs_context *),
2279 int (*set)(struct super_block *, struct fs_context *));
2280 struct super_block *sget(struct file_system_type *type,
2281 int (*test)(struct super_block *,void *),
2282 int (*set)(struct super_block *,void *),
2283 int flags, void *data);
2285 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2286 #define fops_get(fops) \
2287 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2288 #define fops_put(fops) \
2289 do { if (fops) module_put((fops)->owner); } while(0)
2291 * This one is to be used *ONLY* from ->open() instances.
2292 * fops must be non-NULL, pinned down *and* module dependencies
2293 * should be sufficient to pin the caller down as well.
2295 #define replace_fops(f, fops) \
2297 struct file *__file = (f); \
2298 fops_put(__file->f_op); \
2299 BUG_ON(!(__file->f_op = (fops))); \
2302 extern int register_filesystem(struct file_system_type *);
2303 extern int unregister_filesystem(struct file_system_type *);
2304 extern struct vfsmount *kern_mount(struct file_system_type *);
2305 extern void kern_unmount(struct vfsmount *mnt);
2306 extern int may_umount_tree(struct vfsmount *);
2307 extern int may_umount(struct vfsmount *);
2308 extern long do_mount(const char *, const char __user *,
2309 const char *, unsigned long, void *);
2310 extern struct vfsmount *collect_mounts(const struct path *);
2311 extern void drop_collected_mounts(struct vfsmount *);
2312 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2314 extern int vfs_statfs(const struct path *, struct kstatfs *);
2315 extern int user_statfs(const char __user *, struct kstatfs *);
2316 extern int fd_statfs(int, struct kstatfs *);
2317 extern int freeze_super(struct super_block *super);
2318 extern int thaw_super(struct super_block *super);
2319 extern bool our_mnt(struct vfsmount *mnt);
2320 extern __printf(2, 3)
2321 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2322 extern int super_setup_bdi(struct super_block *sb);
2324 extern int current_umask(void);
2326 extern void ihold(struct inode * inode);
2327 extern void iput(struct inode *);
2328 extern int generic_update_time(struct inode *, struct timespec64 *, int);
2331 extern struct kobject *fs_kobj;
2333 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2335 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2336 extern int locks_mandatory_locked(struct file *);
2337 extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
2340 * Candidates for mandatory locking have the setgid bit set
2341 * but no group execute bit - an otherwise meaningless combination.
2344 static inline int __mandatory_lock(struct inode *ino)
2346 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2350 * ... and these candidates should be on SB_MANDLOCK mounted fs,
2351 * otherwise these will be advisory locks
2354 static inline int mandatory_lock(struct inode *ino)
2356 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2359 static inline int locks_verify_locked(struct file *file)
2361 if (mandatory_lock(locks_inode(file)))
2362 return locks_mandatory_locked(file);
2366 static inline int locks_verify_truncate(struct inode *inode,
2370 if (!inode->i_flctx || !mandatory_lock(inode))
2373 if (size < inode->i_size) {
2374 return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2377 return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2382 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2384 static inline int locks_mandatory_locked(struct file *file)
2389 static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2390 loff_t start, loff_t end, unsigned char type)
2395 static inline int __mandatory_lock(struct inode *inode)
2400 static inline int mandatory_lock(struct inode *inode)
2405 static inline int locks_verify_locked(struct file *file)
2410 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2416 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2419 #ifdef CONFIG_FILE_LOCKING
2420 static inline int break_lease(struct inode *inode, unsigned int mode)
2423 * Since this check is lockless, we must ensure that any refcounts
2424 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2425 * could end up racing with tasks trying to set a new lease on this
2429 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2430 return __break_lease(inode, mode, FL_LEASE);
2434 static inline int break_deleg(struct inode *inode, unsigned int mode)
2437 * Since this check is lockless, we must ensure that any refcounts
2438 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2439 * could end up racing with tasks trying to set a new lease on this
2443 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2444 return __break_lease(inode, mode, FL_DELEG);
2448 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2452 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2453 if (ret == -EWOULDBLOCK && delegated_inode) {
2454 *delegated_inode = inode;
2460 static inline int break_deleg_wait(struct inode **delegated_inode)
2464 ret = break_deleg(*delegated_inode, O_WRONLY);
2465 iput(*delegated_inode);
2466 *delegated_inode = NULL;
2470 static inline int break_layout(struct inode *inode, bool wait)
2473 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2474 return __break_lease(inode,
2475 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2480 #else /* !CONFIG_FILE_LOCKING */
2481 static inline int break_lease(struct inode *inode, unsigned int mode)
2486 static inline int break_deleg(struct inode *inode, unsigned int mode)
2491 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2496 static inline int break_deleg_wait(struct inode **delegated_inode)
2502 static inline int break_layout(struct inode *inode, bool wait)
2507 #endif /* CONFIG_FILE_LOCKING */
2512 const char *name; /* pointer to actual string */
2513 const __user char *uptr; /* original userland pointer */
2515 struct audit_names *aname;
2518 static_assert(offsetof(struct filename, iname) % sizeof(long) == 0);
2520 extern long vfs_truncate(const struct path *, loff_t);
2521 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2523 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2525 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2527 extern struct file *file_open_name(struct filename *, int, umode_t);
2528 extern struct file *filp_open(const char *, int, umode_t);
2529 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2530 const char *, int, umode_t);
2531 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2532 extern struct file * open_with_fake_path(const struct path *, int,
2533 struct inode*, const struct cred *);
2534 static inline struct file *file_clone_open(struct file *file)
2536 return dentry_open(&file->f_path, file->f_flags, file->f_cred);
2538 extern int filp_close(struct file *, fl_owner_t id);
2540 extern struct filename *getname_flags(const char __user *, int, int *);
2541 extern struct filename *getname(const char __user *);
2542 extern struct filename *getname_kernel(const char *);
2543 extern void putname(struct filename *name);
2545 extern int finish_open(struct file *file, struct dentry *dentry,
2546 int (*open)(struct inode *, struct file *));
2547 extern int finish_no_open(struct file *file, struct dentry *dentry);
2550 extern void __init vfs_caches_init_early(void);
2551 extern void __init vfs_caches_init(void);
2553 extern struct kmem_cache *names_cachep;
2555 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2556 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2558 extern struct super_block *blockdev_superblock;
2559 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2561 return IS_ENABLED(CONFIG_BLOCK) && sb == blockdev_superblock;
2564 void emergency_thaw_all(void);
2565 extern int sync_filesystem(struct super_block *);
2566 extern const struct file_operations def_blk_fops;
2567 extern const struct file_operations def_chr_fops;
2570 #define CHRDEV_MAJOR_MAX 512
2571 /* Marks the bottom of the first segment of free char majors */
2572 #define CHRDEV_MAJOR_DYN_END 234
2573 /* Marks the top and bottom of the second segment of free char majors */
2574 #define CHRDEV_MAJOR_DYN_EXT_START 511
2575 #define CHRDEV_MAJOR_DYN_EXT_END 384
2577 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2578 extern int register_chrdev_region(dev_t, unsigned, const char *);
2579 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2580 unsigned int count, const char *name,
2581 const struct file_operations *fops);
2582 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2583 unsigned int count, const char *name);
2584 extern void unregister_chrdev_region(dev_t, unsigned);
2585 extern void chrdev_show(struct seq_file *,off_t);
2587 static inline int register_chrdev(unsigned int major, const char *name,
2588 const struct file_operations *fops)
2590 return __register_chrdev(major, 0, 256, name, fops);
2593 static inline void unregister_chrdev(unsigned int major, const char *name)
2595 __unregister_chrdev(major, 0, 256, name);
2598 extern void init_special_inode(struct inode *, umode_t, dev_t);
2600 /* Invalid inode operations -- fs/bad_inode.c */
2601 extern void make_bad_inode(struct inode *);
2602 extern bool is_bad_inode(struct inode *);
2604 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2605 pgoff_t start, pgoff_t end);
2607 void invalidate_mapping_pagevec(struct address_space *mapping,
2608 pgoff_t start, pgoff_t end,
2609 unsigned long *nr_pagevec);
2611 static inline void invalidate_remote_inode(struct inode *inode)
2613 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2614 S_ISLNK(inode->i_mode))
2615 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2617 extern int invalidate_inode_pages2(struct address_space *mapping);
2618 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2619 pgoff_t start, pgoff_t end);
2620 extern int write_inode_now(struct inode *, int);
2621 extern int filemap_fdatawrite(struct address_space *);
2622 extern int filemap_flush(struct address_space *);
2623 extern int filemap_fdatawait_keep_errors(struct address_space *mapping);
2624 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2626 extern int filemap_fdatawait_range_keep_errors(struct address_space *mapping,
2627 loff_t start_byte, loff_t end_byte);
2629 static inline int filemap_fdatawait(struct address_space *mapping)
2631 return filemap_fdatawait_range(mapping, 0, LLONG_MAX);
2634 extern bool filemap_range_has_page(struct address_space *, loff_t lstart,
2636 extern int filemap_write_and_wait_range(struct address_space *mapping,
2637 loff_t lstart, loff_t lend);
2638 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2639 loff_t start, loff_t end, int sync_mode);
2640 extern int filemap_fdatawrite_range(struct address_space *mapping,
2641 loff_t start, loff_t end);
2642 extern int filemap_check_errors(struct address_space *mapping);
2643 extern void __filemap_set_wb_err(struct address_space *mapping, int err);
2645 static inline int filemap_write_and_wait(struct address_space *mapping)
2647 return filemap_write_and_wait_range(mapping, 0, LLONG_MAX);
2650 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2652 extern int __must_check file_check_and_advance_wb_err(struct file *file);
2653 extern int __must_check file_write_and_wait_range(struct file *file,
2654 loff_t start, loff_t end);
2656 static inline int file_write_and_wait(struct file *file)
2658 return file_write_and_wait_range(file, 0, LLONG_MAX);
2662 * filemap_set_wb_err - set a writeback error on an address_space
2663 * @mapping: mapping in which to set writeback error
2664 * @err: error to be set in mapping
2666 * When writeback fails in some way, we must record that error so that
2667 * userspace can be informed when fsync and the like are called. We endeavor
2668 * to report errors on any file that was open at the time of the error. Some
2669 * internal callers also need to know when writeback errors have occurred.
2671 * When a writeback error occurs, most filesystems will want to call
2672 * filemap_set_wb_err to record the error in the mapping so that it will be
2673 * automatically reported whenever fsync is called on the file.
2675 static inline void filemap_set_wb_err(struct address_space *mapping, int err)
2677 /* Fastpath for common case of no error */
2679 __filemap_set_wb_err(mapping, err);
2683 * filemap_check_wb_err - has an error occurred since the mark was sampled?
2684 * @mapping: mapping to check for writeback errors
2685 * @since: previously-sampled errseq_t
2687 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2688 * the given value was sampled.
2690 * If it has then report the latest error set, otherwise return 0.
2692 static inline int filemap_check_wb_err(struct address_space *mapping,
2695 return errseq_check(&mapping->wb_err, since);
2699 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2700 * @mapping: mapping to be sampled
2702 * Writeback errors are always reported relative to a particular sample point
2703 * in the past. This function provides those sample points.
2705 static inline errseq_t filemap_sample_wb_err(struct address_space *mapping)
2707 return errseq_sample(&mapping->wb_err);
2711 * file_sample_sb_err - sample the current errseq_t to test for later errors
2712 * @file: file pointer to be sampled
2714 * Grab the most current superblock-level errseq_t value for the given
2717 static inline errseq_t file_sample_sb_err(struct file *file)
2719 return errseq_sample(&file->f_path.dentry->d_sb->s_wb_err);
2722 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2724 extern int vfs_fsync(struct file *file, int datasync);
2726 extern int sync_file_range(struct file *file, loff_t offset, loff_t nbytes,
2727 unsigned int flags);
2730 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2731 * to already be updated for the write, and will return either the amount
2732 * of bytes passed in, or an error if syncing the file failed.
2734 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2736 if (iocb->ki_flags & IOCB_DSYNC) {
2737 int ret = vfs_fsync_range(iocb->ki_filp,
2738 iocb->ki_pos - count, iocb->ki_pos - 1,
2739 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2747 extern void emergency_sync(void);
2748 extern void emergency_remount(void);
2751 extern int bmap(struct inode *inode, sector_t *block);
2753 static inline int bmap(struct inode *inode, sector_t *block)
2759 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2760 extern int inode_permission(struct inode *, int);
2761 extern int generic_permission(struct inode *, int);
2762 extern int __check_sticky(struct inode *dir, struct inode *inode);
2764 static inline bool execute_ok(struct inode *inode)
2766 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2769 static inline void file_start_write(struct file *file)
2771 if (!S_ISREG(file_inode(file)->i_mode))
2773 sb_start_write(file_inode(file)->i_sb);
2776 static inline bool file_start_write_trylock(struct file *file)
2778 if (!S_ISREG(file_inode(file)->i_mode))
2780 return sb_start_write_trylock(file_inode(file)->i_sb);
2783 static inline void file_end_write(struct file *file)
2785 if (!S_ISREG(file_inode(file)->i_mode))
2787 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2791 * get_write_access() gets write permission for a file.
2792 * put_write_access() releases this write permission.
2793 * This is used for regular files.
2794 * We cannot support write (and maybe mmap read-write shared) accesses and
2795 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2796 * can have the following values:
2797 * 0: no writers, no VM_DENYWRITE mappings
2798 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2799 * > 0: (i_writecount) users are writing to the file.
2801 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2802 * except for the cases where we don't hold i_writecount yet. Then we need to
2803 * use {get,deny}_write_access() - these functions check the sign and refuse
2804 * to do the change if sign is wrong.
2806 static inline int get_write_access(struct inode *inode)
2808 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2810 static inline int deny_write_access(struct file *file)
2812 struct inode *inode = file_inode(file);
2813 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2815 static inline void put_write_access(struct inode * inode)
2817 atomic_dec(&inode->i_writecount);
2819 static inline void allow_write_access(struct file *file)
2822 atomic_inc(&file_inode(file)->i_writecount);
2824 static inline bool inode_is_open_for_write(const struct inode *inode)
2826 return atomic_read(&inode->i_writecount) > 0;
2829 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
2830 static inline void i_readcount_dec(struct inode *inode)
2832 BUG_ON(!atomic_read(&inode->i_readcount));
2833 atomic_dec(&inode->i_readcount);
2835 static inline void i_readcount_inc(struct inode *inode)
2837 atomic_inc(&inode->i_readcount);
2840 static inline void i_readcount_dec(struct inode *inode)
2844 static inline void i_readcount_inc(struct inode *inode)
2849 extern int do_pipe_flags(int *, int);
2851 extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
2852 ssize_t __kernel_read(struct file *file, void *buf, size_t count, loff_t *pos);
2853 extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
2854 extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
2855 extern struct file * open_exec(const char *);
2857 /* fs/dcache.c -- generic fs support functions */
2858 extern bool is_subdir(struct dentry *, struct dentry *);
2859 extern bool path_is_under(const struct path *, const struct path *);
2861 extern char *file_path(struct file *, char *, int);
2863 #include <linux/err.h>
2865 /* needed for stackable file system support */
2866 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2868 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2870 extern int inode_init_always(struct super_block *, struct inode *);
2871 extern void inode_init_once(struct inode *);
2872 extern void address_space_init_once(struct address_space *mapping);
2873 extern struct inode * igrab(struct inode *);
2874 extern ino_t iunique(struct super_block *, ino_t);
2875 extern int inode_needs_sync(struct inode *inode);
2876 extern int generic_delete_inode(struct inode *inode);
2877 static inline int generic_drop_inode(struct inode *inode)
2879 return !inode->i_nlink || inode_unhashed(inode) ||
2880 (inode->i_state & I_DONTCACHE);
2882 extern void d_mark_dontcache(struct inode *inode);
2884 extern struct inode *ilookup5_nowait(struct super_block *sb,
2885 unsigned long hashval, int (*test)(struct inode *, void *),
2887 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2888 int (*test)(struct inode *, void *), void *data);
2889 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2891 extern struct inode *inode_insert5(struct inode *inode, unsigned long hashval,
2892 int (*test)(struct inode *, void *),
2893 int (*set)(struct inode *, void *),
2895 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2896 extern struct inode * iget_locked(struct super_block *, unsigned long);
2897 extern struct inode *find_inode_nowait(struct super_block *,
2899 int (*match)(struct inode *,
2900 unsigned long, void *),
2902 extern struct inode *find_inode_rcu(struct super_block *, unsigned long,
2903 int (*)(struct inode *, void *), void *);
2904 extern struct inode *find_inode_by_ino_rcu(struct super_block *, unsigned long);
2905 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2906 extern int insert_inode_locked(struct inode *);
2907 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2908 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2910 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2912 extern void unlock_new_inode(struct inode *);
2913 extern void discard_new_inode(struct inode *);
2914 extern unsigned int get_next_ino(void);
2915 extern void evict_inodes(struct super_block *sb);
2918 * Userspace may rely on the the inode number being non-zero. For example, glibc
2919 * simply ignores files with zero i_ino in unlink() and other places.
2921 * As an additional complication, if userspace was compiled with
2922 * _FILE_OFFSET_BITS=32 on a 64-bit kernel we'll only end up reading out the
2923 * lower 32 bits, so we need to check that those aren't zero explicitly. With
2924 * _FILE_OFFSET_BITS=64, this may cause some harmless false-negatives, but
2925 * better safe than sorry.
2927 static inline bool is_zero_ino(ino_t ino)
2929 return (u32)ino == 0;
2932 extern void __iget(struct inode * inode);
2933 extern void iget_failed(struct inode *);
2934 extern void clear_inode(struct inode *);
2935 extern void __destroy_inode(struct inode *);
2936 extern struct inode *new_inode_pseudo(struct super_block *sb);
2937 extern struct inode *new_inode(struct super_block *sb);
2938 extern void free_inode_nonrcu(struct inode *inode);
2939 extern int should_remove_suid(struct dentry *);
2940 extern int file_remove_privs(struct file *);
2942 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2943 static inline void insert_inode_hash(struct inode *inode)
2945 __insert_inode_hash(inode, inode->i_ino);
2948 extern void __remove_inode_hash(struct inode *);
2949 static inline void remove_inode_hash(struct inode *inode)
2951 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
2952 __remove_inode_hash(inode);
2955 extern void inode_sb_list_add(struct inode *inode);
2957 extern int sb_set_blocksize(struct super_block *, int);
2958 extern int sb_min_blocksize(struct super_block *, int);
2960 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2961 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2962 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2963 extern int generic_write_check_limits(struct file *file, loff_t pos,
2965 extern int generic_file_rw_checks(struct file *file_in, struct file *file_out);
2966 extern ssize_t generic_file_buffered_read(struct kiocb *iocb,
2967 struct iov_iter *to, ssize_t already_read);
2968 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2969 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2970 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2971 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
2972 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2974 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
2976 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
2978 ssize_t vfs_iocb_iter_read(struct file *file, struct kiocb *iocb,
2979 struct iov_iter *iter);
2980 ssize_t vfs_iocb_iter_write(struct file *file, struct kiocb *iocb,
2981 struct iov_iter *iter);
2983 /* fs/block_dev.c */
2984 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
2985 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2986 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2988 extern void block_sync_page(struct page *page);
2991 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2992 struct pipe_inode_info *, size_t, unsigned int);
2993 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2994 struct file *, loff_t *, size_t, unsigned int);
2995 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2996 struct file *out, loff_t *, size_t len, unsigned int flags);
2997 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2998 loff_t *opos, size_t len, unsigned int flags);
3002 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
3003 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
3004 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
3005 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
3006 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
3007 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
3008 int whence, loff_t maxsize, loff_t eof);
3009 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
3010 int whence, loff_t size);
3011 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
3012 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
3013 extern int generic_file_open(struct inode * inode, struct file * filp);
3014 extern int nonseekable_open(struct inode * inode, struct file * filp);
3015 extern int stream_open(struct inode * inode, struct file * filp);
3018 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
3019 loff_t file_offset);
3022 /* need locking between buffered and direct access */
3025 /* filesystem does not support filling holes */
3026 DIO_SKIP_HOLES = 0x02,
3029 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
3030 struct block_device *bdev, struct iov_iter *iter,
3031 get_block_t get_block,
3032 dio_iodone_t end_io, dio_submit_t submit_io,
3035 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
3036 struct inode *inode,
3037 struct iov_iter *iter,
3038 get_block_t get_block)
3040 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
3041 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
3045 void inode_dio_wait(struct inode *inode);
3048 * inode_dio_begin - signal start of a direct I/O requests
3049 * @inode: inode the direct I/O happens on
3051 * This is called once we've finished processing a direct I/O request,
3052 * and is used to wake up callers waiting for direct I/O to be quiesced.
3054 static inline void inode_dio_begin(struct inode *inode)
3056 atomic_inc(&inode->i_dio_count);
3060 * inode_dio_end - signal finish of a direct I/O requests
3061 * @inode: inode the direct I/O happens on
3063 * This is called once we've finished processing a direct I/O request,
3064 * and is used to wake up callers waiting for direct I/O to be quiesced.
3066 static inline void inode_dio_end(struct inode *inode)
3068 if (atomic_dec_and_test(&inode->i_dio_count))
3069 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
3073 * Warn about a page cache invalidation failure diring a direct I/O write.
3075 void dio_warn_stale_pagecache(struct file *filp);
3077 extern void inode_set_flags(struct inode *inode, unsigned int flags,
3080 extern const struct file_operations generic_ro_fops;
3082 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3084 extern int readlink_copy(char __user *, int, const char *);
3085 extern int page_readlink(struct dentry *, char __user *, int);
3086 extern const char *page_get_link(struct dentry *, struct inode *,
3087 struct delayed_call *);
3088 extern void page_put_link(void *);
3089 extern int __page_symlink(struct inode *inode, const char *symname, int len,
3091 extern int page_symlink(struct inode *inode, const char *symname, int len);
3092 extern const struct inode_operations page_symlink_inode_operations;
3093 extern void kfree_link(void *);
3094 extern void generic_fillattr(struct inode *, struct kstat *);
3095 extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
3096 extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
3097 void __inode_add_bytes(struct inode *inode, loff_t bytes);
3098 void inode_add_bytes(struct inode *inode, loff_t bytes);
3099 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
3100 void inode_sub_bytes(struct inode *inode, loff_t bytes);
3101 static inline loff_t __inode_get_bytes(struct inode *inode)
3103 return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes;
3105 loff_t inode_get_bytes(struct inode *inode);
3106 void inode_set_bytes(struct inode *inode, loff_t bytes);
3107 const char *simple_get_link(struct dentry *, struct inode *,
3108 struct delayed_call *);
3109 extern const struct inode_operations simple_symlink_inode_operations;
3111 extern int iterate_dir(struct file *, struct dir_context *);
3113 int vfs_fstatat(int dfd, const char __user *filename, struct kstat *stat,
3115 int vfs_fstat(int fd, struct kstat *stat);
3117 static inline int vfs_stat(const char __user *filename, struct kstat *stat)
3119 return vfs_fstatat(AT_FDCWD, filename, stat, 0);
3121 static inline int vfs_lstat(const char __user *name, struct kstat *stat)
3123 return vfs_fstatat(AT_FDCWD, name, stat, AT_SYMLINK_NOFOLLOW);
3126 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
3127 extern int vfs_readlink(struct dentry *, char __user *, int);
3129 extern struct file_system_type *get_filesystem(struct file_system_type *fs);
3130 extern void put_filesystem(struct file_system_type *fs);
3131 extern struct file_system_type *get_fs_type(const char *name);
3132 extern struct super_block *get_super(struct block_device *);
3133 extern struct super_block *get_active_super(struct block_device *bdev);
3134 extern void drop_super(struct super_block *sb);
3135 extern void drop_super_exclusive(struct super_block *sb);
3136 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
3137 extern void iterate_supers_type(struct file_system_type *,
3138 void (*)(struct super_block *, void *), void *);
3140 extern int dcache_dir_open(struct inode *, struct file *);
3141 extern int dcache_dir_close(struct inode *, struct file *);
3142 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
3143 extern int dcache_readdir(struct file *, struct dir_context *);
3144 extern int simple_setattr(struct dentry *, struct iattr *);
3145 extern int simple_getattr(const struct path *, struct kstat *, u32, unsigned int);
3146 extern int simple_statfs(struct dentry *, struct kstatfs *);
3147 extern int simple_open(struct inode *inode, struct file *file);
3148 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3149 extern int simple_unlink(struct inode *, struct dentry *);
3150 extern int simple_rmdir(struct inode *, struct dentry *);
3151 extern int simple_rename(struct inode *, struct dentry *,
3152 struct inode *, struct dentry *, unsigned int);
3153 extern void simple_recursive_removal(struct dentry *,
3154 void (*callback)(struct dentry *));
3155 extern int noop_fsync(struct file *, loff_t, loff_t, int);
3156 extern int noop_set_page_dirty(struct page *page);
3157 extern void noop_invalidatepage(struct page *page, unsigned int offset,
3158 unsigned int length);
3159 extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter);
3160 extern int simple_empty(struct dentry *);
3161 extern int simple_readpage(struct file *file, struct page *page);
3162 extern int simple_write_begin(struct file *file, struct address_space *mapping,
3163 loff_t pos, unsigned len, unsigned flags,
3164 struct page **pagep, void **fsdata);
3165 extern int simple_write_end(struct file *file, struct address_space *mapping,
3166 loff_t pos, unsigned len, unsigned copied,
3167 struct page *page, void *fsdata);
3168 extern int always_delete_dentry(const struct dentry *);
3169 extern struct inode *alloc_anon_inode(struct super_block *);
3170 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
3171 extern const struct dentry_operations simple_dentry_operations;
3173 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3174 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3175 extern const struct file_operations simple_dir_operations;
3176 extern const struct inode_operations simple_dir_inode_operations;
3177 extern void make_empty_dir_inode(struct inode *inode);
3178 extern bool is_empty_dir_inode(struct inode *inode);
3179 struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
3180 struct dentry *d_alloc_name(struct dentry *, const char *);
3181 extern int simple_fill_super(struct super_block *, unsigned long,
3182 const struct tree_descr *);
3183 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3184 extern void simple_release_fs(struct vfsmount **mount, int *count);
3186 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3187 loff_t *ppos, const void *from, size_t available);
3188 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3189 const void __user *from, size_t count);
3191 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3192 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3194 extern int generic_check_addressable(unsigned, u64);
3196 #ifdef CONFIG_UNICODE
3197 extern int generic_ci_d_hash(const struct dentry *dentry, struct qstr *str);
3198 extern int generic_ci_d_compare(const struct dentry *dentry, unsigned int len,
3199 const char *str, const struct qstr *name);
3201 extern void generic_set_encrypted_ci_d_ops(struct dentry *dentry);
3203 #ifdef CONFIG_MIGRATION
3204 extern int buffer_migrate_page(struct address_space *,
3205 struct page *, struct page *,
3207 extern int buffer_migrate_page_norefs(struct address_space *,
3208 struct page *, struct page *,
3211 #define buffer_migrate_page NULL
3212 #define buffer_migrate_page_norefs NULL
3215 extern int setattr_prepare(struct dentry *, struct iattr *);
3216 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3217 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
3219 extern int file_update_time(struct file *file);
3221 static inline bool vma_is_dax(const struct vm_area_struct *vma)
3223 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3226 static inline bool vma_is_fsdax(struct vm_area_struct *vma)
3228 struct inode *inode;
3230 if (!IS_ENABLED(CONFIG_FS_DAX) || !vma->vm_file)
3232 if (!vma_is_dax(vma))
3234 inode = file_inode(vma->vm_file);
3235 if (S_ISCHR(inode->i_mode))
3236 return false; /* device-dax */
3240 static inline int iocb_flags(struct file *file)
3243 if (file->f_flags & O_APPEND)
3245 if (file->f_flags & O_DIRECT)
3247 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3249 if (file->f_flags & __O_SYNC)
3254 static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
3256 int kiocb_flags = 0;
3258 /* make sure there's no overlap between RWF and private IOCB flags */
3259 BUILD_BUG_ON((__force int) RWF_SUPPORTED & IOCB_EVENTFD);
3263 if (unlikely(flags & ~RWF_SUPPORTED))
3266 if (flags & RWF_NOWAIT) {
3267 if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
3269 kiocb_flags |= IOCB_NOIO;
3271 kiocb_flags |= (__force int) (flags & RWF_SUPPORTED);
3272 if (flags & RWF_SYNC)
3273 kiocb_flags |= IOCB_DSYNC;
3275 ki->ki_flags |= kiocb_flags;
3279 static inline ino_t parent_ino(struct dentry *dentry)
3284 * Don't strictly need d_lock here? If the parent ino could change
3285 * then surely we'd have a deeper race in the caller?
3287 spin_lock(&dentry->d_lock);
3288 res = dentry->d_parent->d_inode->i_ino;
3289 spin_unlock(&dentry->d_lock);
3293 /* Transaction based IO helpers */
3296 * An argresp is stored in an allocated page and holds the
3297 * size of the argument or response, along with its content
3299 struct simple_transaction_argresp {
3304 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3306 char *simple_transaction_get(struct file *file, const char __user *buf,
3308 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3309 size_t size, loff_t *pos);
3310 int simple_transaction_release(struct inode *inode, struct file *file);
3312 void simple_transaction_set(struct file *file, size_t n);
3315 * simple attribute files
3317 * These attributes behave similar to those in sysfs:
3319 * Writing to an attribute immediately sets a value, an open file can be
3320 * written to multiple times.
3322 * Reading from an attribute creates a buffer from the value that might get
3323 * read with multiple read calls. When the attribute has been read
3324 * completely, no further read calls are possible until the file is opened
3327 * All attributes contain a text representation of a numeric value
3328 * that are accessed with the get() and set() functions.
3330 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3331 static int __fops ## _open(struct inode *inode, struct file *file) \
3333 __simple_attr_check_format(__fmt, 0ull); \
3334 return simple_attr_open(inode, file, __get, __set, __fmt); \
3336 static const struct file_operations __fops = { \
3337 .owner = THIS_MODULE, \
3338 .open = __fops ## _open, \
3339 .release = simple_attr_release, \
3340 .read = simple_attr_read, \
3341 .write = simple_attr_write, \
3342 .llseek = generic_file_llseek, \
3345 static inline __printf(1, 2)
3346 void __simple_attr_check_format(const char *fmt, ...)
3348 /* don't do anything, just let the compiler check the arguments; */
3351 int simple_attr_open(struct inode *inode, struct file *file,
3352 int (*get)(void *, u64 *), int (*set)(void *, u64),
3354 int simple_attr_release(struct inode *inode, struct file *file);
3355 ssize_t simple_attr_read(struct file *file, char __user *buf,
3356 size_t len, loff_t *ppos);
3357 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3358 size_t len, loff_t *ppos);
3361 int proc_nr_files(struct ctl_table *table, int write,
3362 void *buffer, size_t *lenp, loff_t *ppos);
3363 int proc_nr_dentry(struct ctl_table *table, int write,
3364 void *buffer, size_t *lenp, loff_t *ppos);
3365 int proc_nr_inodes(struct ctl_table *table, int write,
3366 void *buffer, size_t *lenp, loff_t *ppos);
3367 int __init get_filesystem_list(char *buf);
3369 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3370 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3372 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3373 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3374 (flag & __FMODE_NONOTIFY)))
3376 static inline bool is_sxid(umode_t mode)
3378 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3381 static inline int check_sticky(struct inode *dir, struct inode *inode)
3383 if (!(dir->i_mode & S_ISVTX))
3386 return __check_sticky(dir, inode);
3389 static inline void inode_has_no_xattr(struct inode *inode)
3391 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
3392 inode->i_flags |= S_NOSEC;
3395 static inline bool is_root_inode(struct inode *inode)
3397 return inode == inode->i_sb->s_root->d_inode;
3400 static inline bool dir_emit(struct dir_context *ctx,
3401 const char *name, int namelen,
3402 u64 ino, unsigned type)
3404 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3406 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3408 return ctx->actor(ctx, ".", 1, ctx->pos,
3409 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3411 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3413 return ctx->actor(ctx, "..", 2, ctx->pos,
3414 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3416 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3418 if (ctx->pos == 0) {
3419 if (!dir_emit_dot(file, ctx))
3423 if (ctx->pos == 1) {
3424 if (!dir_emit_dotdot(file, ctx))
3430 static inline bool dir_relax(struct inode *inode)
3432 inode_unlock(inode);
3434 return !IS_DEADDIR(inode);
3437 static inline bool dir_relax_shared(struct inode *inode)
3439 inode_unlock_shared(inode);
3440 inode_lock_shared(inode);
3441 return !IS_DEADDIR(inode);
3444 extern bool path_noexec(const struct path *path);
3445 extern void inode_nohighmem(struct inode *inode);
3448 extern int vfs_fadvise(struct file *file, loff_t offset, loff_t len,
3450 extern int generic_fadvise(struct file *file, loff_t offset, loff_t len,
3453 int vfs_ioc_setflags_prepare(struct inode *inode, unsigned int oldflags,
3454 unsigned int flags);
3456 int vfs_ioc_fssetxattr_check(struct inode *inode, const struct fsxattr *old_fa,
3457 struct fsxattr *fa);
3459 static inline void simple_fill_fsxattr(struct fsxattr *fa, __u32 xflags)
3461 memset(fa, 0, sizeof(*fa));
3462 fa->fsx_xflags = xflags;
3466 * Flush file data before changing attributes. Caller must hold any locks
3467 * required to prevent further writes to this file until we're done setting
3470 static inline int inode_drain_writes(struct inode *inode)
3472 inode_dio_wait(inode);
3473 return filemap_write_and_wait(inode->i_mapping);
3476 #endif /* _LINUX_FS_H */