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)
319 /* kiocb is a read or write operation submitted by fs/aio.c. */
320 #define IOCB_AIO_RW (1 << 23)
323 struct file *ki_filp;
325 /* The 'ki_filp' pointer is shared in a union for aio */
326 randomized_struct_fields_start
329 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
333 u16 ki_ioprio; /* See linux/ioprio.h */
335 unsigned int ki_cookie; /* for ->iopoll */
336 struct wait_page_queue *ki_waitq; /* for async buffered IO */
339 randomized_struct_fields_end
342 static inline bool is_sync_kiocb(struct kiocb *kiocb)
344 return kiocb->ki_complete == NULL;
348 * "descriptor" for what we're up to with a read.
349 * This allows us to use the same read code yet
350 * have multiple different users of the data that
351 * we read from a file.
353 * The simplest case just copies the data to user
366 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
367 unsigned long, unsigned long);
369 struct address_space_operations {
370 int (*writepage)(struct page *page, struct writeback_control *wbc);
371 int (*readpage)(struct file *, struct page *);
373 /* Write back some dirty pages from this mapping. */
374 int (*writepages)(struct address_space *, struct writeback_control *);
376 /* Set a page dirty. Return true if this dirtied it */
377 int (*set_page_dirty)(struct page *page);
380 * Reads in the requested pages. Unlike ->readpage(), this is
381 * PURELY used for read-ahead!.
383 int (*readpages)(struct file *filp, struct address_space *mapping,
384 struct list_head *pages, unsigned nr_pages);
385 void (*readahead)(struct readahead_control *);
387 int (*write_begin)(struct file *, struct address_space *mapping,
388 loff_t pos, unsigned len, unsigned flags,
389 struct page **pagep, void **fsdata);
390 int (*write_end)(struct file *, struct address_space *mapping,
391 loff_t pos, unsigned len, unsigned copied,
392 struct page *page, void *fsdata);
394 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
395 sector_t (*bmap)(struct address_space *, sector_t);
396 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
397 int (*releasepage) (struct page *, gfp_t);
398 void (*freepage)(struct page *);
399 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
401 * migrate the contents of a page to the specified target. If
402 * migrate_mode is MIGRATE_ASYNC, it must not block.
404 int (*migratepage) (struct address_space *,
405 struct page *, struct page *, enum migrate_mode);
406 bool (*isolate_page)(struct page *, isolate_mode_t);
407 void (*putback_page)(struct page *);
408 int (*launder_page) (struct page *);
409 int (*is_partially_uptodate) (struct page *, unsigned long,
411 void (*is_dirty_writeback) (struct page *, bool *, bool *);
412 int (*error_remove_page)(struct address_space *, struct page *);
414 /* swapfile support */
415 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
417 void (*swap_deactivate)(struct file *file);
420 extern const struct address_space_operations empty_aops;
423 * pagecache_write_begin/pagecache_write_end must be used by general code
424 * to write into the pagecache.
426 int pagecache_write_begin(struct file *, struct address_space *mapping,
427 loff_t pos, unsigned len, unsigned flags,
428 struct page **pagep, void **fsdata);
430 int pagecache_write_end(struct file *, struct address_space *mapping,
431 loff_t pos, unsigned len, unsigned copied,
432 struct page *page, void *fsdata);
435 * struct address_space - Contents of a cacheable, mappable object.
436 * @host: Owner, either the inode or the block_device.
437 * @i_pages: Cached pages.
438 * @gfp_mask: Memory allocation flags to use for allocating pages.
439 * @i_mmap_writable: Number of VM_SHARED mappings.
440 * @nr_thps: Number of THPs in the pagecache (non-shmem only).
441 * @i_mmap: Tree of private and shared mappings.
442 * @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable.
443 * @nrpages: Number of page entries, protected by the i_pages lock.
444 * @nrexceptional: Shadow or DAX entries, protected by the i_pages lock.
445 * @writeback_index: Writeback starts here.
447 * @flags: Error bits and flags (AS_*).
448 * @wb_err: The most recent error which has occurred.
449 * @private_lock: For use by the owner of the address_space.
450 * @private_list: For use by the owner of the address_space.
451 * @private_data: For use by the owner of the address_space.
453 struct address_space {
455 struct xarray i_pages;
457 atomic_t i_mmap_writable;
458 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
459 /* number of thp, only for non-shmem files */
462 struct rb_root_cached i_mmap;
463 struct rw_semaphore i_mmap_rwsem;
464 unsigned long nrpages;
465 unsigned long nrexceptional;
466 pgoff_t writeback_index;
467 const struct address_space_operations *a_ops;
470 spinlock_t private_lock;
471 struct list_head private_list;
473 } __attribute__((aligned(sizeof(long)))) __randomize_layout;
475 * On most architectures that alignment is already the case; but
476 * must be enforced here for CRIS, to let the least significant bit
477 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
480 /* XArray tags, for tagging dirty and writeback pages in the pagecache. */
481 #define PAGECACHE_TAG_DIRTY XA_MARK_0
482 #define PAGECACHE_TAG_WRITEBACK XA_MARK_1
483 #define PAGECACHE_TAG_TOWRITE XA_MARK_2
486 * Returns true if any of the pages in the mapping are marked with the tag.
488 static inline bool mapping_tagged(struct address_space *mapping, xa_mark_t tag)
490 return xa_marked(&mapping->i_pages, tag);
493 static inline void i_mmap_lock_write(struct address_space *mapping)
495 down_write(&mapping->i_mmap_rwsem);
498 static inline int i_mmap_trylock_write(struct address_space *mapping)
500 return down_write_trylock(&mapping->i_mmap_rwsem);
503 static inline void i_mmap_unlock_write(struct address_space *mapping)
505 up_write(&mapping->i_mmap_rwsem);
508 static inline void i_mmap_lock_read(struct address_space *mapping)
510 down_read(&mapping->i_mmap_rwsem);
513 static inline void i_mmap_unlock_read(struct address_space *mapping)
515 up_read(&mapping->i_mmap_rwsem);
518 static inline void i_mmap_assert_locked(struct address_space *mapping)
520 lockdep_assert_held(&mapping->i_mmap_rwsem);
523 static inline void i_mmap_assert_write_locked(struct address_space *mapping)
525 lockdep_assert_held_write(&mapping->i_mmap_rwsem);
529 * Might pages of this file be mapped into userspace?
531 static inline int mapping_mapped(struct address_space *mapping)
533 return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root);
537 * Might pages of this file have been modified in userspace?
538 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap
539 * marks vma as VM_SHARED if it is shared, and the file was opened for
540 * writing i.e. vma may be mprotected writable even if now readonly.
542 * If i_mmap_writable is negative, no new writable mappings are allowed. You
543 * can only deny writable mappings, if none exists right now.
545 static inline int mapping_writably_mapped(struct address_space *mapping)
547 return atomic_read(&mapping->i_mmap_writable) > 0;
550 static inline int mapping_map_writable(struct address_space *mapping)
552 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
556 static inline void mapping_unmap_writable(struct address_space *mapping)
558 atomic_dec(&mapping->i_mmap_writable);
561 static inline int mapping_deny_writable(struct address_space *mapping)
563 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
567 static inline void mapping_allow_writable(struct address_space *mapping)
569 atomic_inc(&mapping->i_mmap_writable);
573 * Use sequence counter to get consistent i_size on 32-bit processors.
575 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
576 #include <linux/seqlock.h>
577 #define __NEED_I_SIZE_ORDERED
578 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
580 #define i_size_ordered_init(inode) do { } while (0)
584 #define ACL_NOT_CACHED ((void *)(-1))
585 #define ACL_DONT_CACHE ((void *)(-3))
587 static inline struct posix_acl *
588 uncached_acl_sentinel(struct task_struct *task)
590 return (void *)task + 1;
594 is_uncached_acl(struct posix_acl *acl)
596 return (long)acl & 1;
599 #define IOP_FASTPERM 0x0001
600 #define IOP_LOOKUP 0x0002
601 #define IOP_NOFOLLOW 0x0004
602 #define IOP_XATTR 0x0008
603 #define IOP_DEFAULT_READLINK 0x0010
605 struct fsnotify_mark_connector;
608 * Keep mostly read-only and often accessed (especially for
609 * the RCU path lookup and 'stat' data) fields at the beginning
610 * of the 'struct inode'
614 unsigned short i_opflags;
617 unsigned int i_flags;
619 #ifdef CONFIG_FS_POSIX_ACL
620 struct posix_acl *i_acl;
621 struct posix_acl *i_default_acl;
624 const struct inode_operations *i_op;
625 struct super_block *i_sb;
626 struct address_space *i_mapping;
628 #ifdef CONFIG_SECURITY
632 /* Stat data, not accessed from path walking */
635 * Filesystems may only read i_nlink directly. They shall use the
636 * following functions for modification:
638 * (set|clear|inc|drop)_nlink
639 * inode_(inc|dec)_link_count
642 const unsigned int i_nlink;
643 unsigned int __i_nlink;
647 struct timespec64 i_atime;
648 struct timespec64 i_mtime;
649 struct timespec64 i_ctime;
650 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
651 unsigned short i_bytes;
656 #ifdef __NEED_I_SIZE_ORDERED
657 seqcount_t i_size_seqcount;
661 unsigned long i_state;
662 struct rw_semaphore i_rwsem;
664 unsigned long dirtied_when; /* jiffies of first dirtying */
665 unsigned long dirtied_time_when;
667 struct hlist_node i_hash;
668 struct list_head i_io_list; /* backing dev IO list */
669 #ifdef CONFIG_CGROUP_WRITEBACK
670 struct bdi_writeback *i_wb; /* the associated cgroup wb */
672 /* foreign inode detection, see wbc_detach_inode() */
674 u16 i_wb_frn_avg_time;
675 u16 i_wb_frn_history;
677 struct list_head i_lru; /* inode LRU list */
678 struct list_head i_sb_list;
679 struct list_head i_wb_list; /* backing dev writeback list */
681 struct hlist_head i_dentry;
682 struct rcu_head i_rcu;
684 atomic64_t i_version;
685 atomic64_t i_sequence; /* see futex */
687 atomic_t i_dio_count;
688 atomic_t i_writecount;
689 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
690 atomic_t i_readcount; /* struct files open RO */
693 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
694 void (*free_inode)(struct inode *);
696 struct file_lock_context *i_flctx;
697 struct address_space i_data;
698 struct list_head i_devices;
700 struct pipe_inode_info *i_pipe;
701 struct block_device *i_bdev;
709 #ifdef CONFIG_FSNOTIFY
710 __u32 i_fsnotify_mask; /* all events this inode cares about */
711 struct fsnotify_mark_connector __rcu *i_fsnotify_marks;
714 #ifdef CONFIG_FS_ENCRYPTION
715 struct fscrypt_info *i_crypt_info;
718 #ifdef CONFIG_FS_VERITY
719 struct fsverity_info *i_verity_info;
722 void *i_private; /* fs or device private pointer */
723 } __randomize_layout;
725 struct timespec64 timestamp_truncate(struct timespec64 t, struct inode *inode);
727 static inline unsigned int i_blocksize(const struct inode *node)
729 return (1 << node->i_blkbits);
732 static inline int inode_unhashed(struct inode *inode)
734 return hlist_unhashed(&inode->i_hash);
738 * __mark_inode_dirty expects inodes to be hashed. Since we don't
739 * want special inodes in the fileset inode space, we make them
740 * appear hashed, but do not put on any lists. hlist_del()
741 * will work fine and require no locking.
743 static inline void inode_fake_hash(struct inode *inode)
745 hlist_add_fake(&inode->i_hash);
749 * inode->i_mutex nesting subclasses for the lock validator:
751 * 0: the object of the current VFS operation
755 * 4: second non-directory
756 * 5: second parent (when locking independent directories in rename)
758 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
759 * non-directories at once.
761 * The locking order between these classes is
762 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
764 enum inode_i_mutex_lock_class
774 static inline void inode_lock(struct inode *inode)
776 down_write(&inode->i_rwsem);
779 static inline void inode_unlock(struct inode *inode)
781 up_write(&inode->i_rwsem);
784 static inline void inode_lock_shared(struct inode *inode)
786 down_read(&inode->i_rwsem);
789 static inline void inode_unlock_shared(struct inode *inode)
791 up_read(&inode->i_rwsem);
794 static inline int inode_trylock(struct inode *inode)
796 return down_write_trylock(&inode->i_rwsem);
799 static inline int inode_trylock_shared(struct inode *inode)
801 return down_read_trylock(&inode->i_rwsem);
804 static inline int inode_is_locked(struct inode *inode)
806 return rwsem_is_locked(&inode->i_rwsem);
809 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
811 down_write_nested(&inode->i_rwsem, subclass);
814 static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass)
816 down_read_nested(&inode->i_rwsem, subclass);
819 void lock_two_nondirectories(struct inode *, struct inode*);
820 void unlock_two_nondirectories(struct inode *, struct inode*);
823 * NOTE: in a 32bit arch with a preemptable kernel and
824 * an UP compile the i_size_read/write must be atomic
825 * with respect to the local cpu (unlike with preempt disabled),
826 * but they don't need to be atomic with respect to other cpus like in
827 * true SMP (so they need either to either locally disable irq around
828 * the read or for example on x86 they can be still implemented as a
829 * cmpxchg8b without the need of the lock prefix). For SMP compiles
830 * and 64bit archs it makes no difference if preempt is enabled or not.
832 static inline loff_t i_size_read(const struct inode *inode)
834 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
839 seq = read_seqcount_begin(&inode->i_size_seqcount);
840 i_size = inode->i_size;
841 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
843 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
847 i_size = inode->i_size;
851 return inode->i_size;
856 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
857 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
858 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
860 static inline void i_size_write(struct inode *inode, loff_t i_size)
862 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
864 write_seqcount_begin(&inode->i_size_seqcount);
865 inode->i_size = i_size;
866 write_seqcount_end(&inode->i_size_seqcount);
868 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
870 inode->i_size = i_size;
873 inode->i_size = i_size;
877 static inline unsigned iminor(const struct inode *inode)
879 return MINOR(inode->i_rdev);
882 static inline unsigned imajor(const struct inode *inode)
884 return MAJOR(inode->i_rdev);
888 rwlock_t lock; /* protects pid, uid, euid fields */
889 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
890 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
891 kuid_t uid, euid; /* uid/euid of process setting the owner */
892 int signum; /* posix.1b rt signal to be delivered on IO */
896 * Track a single file's readahead state
898 struct file_ra_state {
899 pgoff_t start; /* where readahead started */
900 unsigned int size; /* # of readahead pages */
901 unsigned int async_size; /* do asynchronous readahead when
902 there are only # of pages ahead */
904 unsigned int ra_pages; /* Maximum readahead window */
905 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
906 loff_t prev_pos; /* Cache last read() position */
910 * Check if @index falls in the readahead windows.
912 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
914 return (index >= ra->start &&
915 index < ra->start + ra->size);
920 struct llist_node fu_llist;
921 struct rcu_head fu_rcuhead;
924 struct inode *f_inode; /* cached value */
925 const struct file_operations *f_op;
928 * Protects f_ep_links, f_flags.
929 * Must not be taken from IRQ context.
932 enum rw_hint f_write_hint;
933 atomic_long_t f_count;
934 unsigned int f_flags;
936 struct mutex f_pos_lock;
938 struct fown_struct f_owner;
939 const struct cred *f_cred;
940 struct file_ra_state f_ra;
943 #ifdef CONFIG_SECURITY
946 /* needed for tty driver, and maybe others */
950 /* Used by fs/eventpoll.c to link all the hooks to this file */
951 struct list_head f_ep_links;
952 struct list_head f_tfile_llink;
953 #endif /* #ifdef CONFIG_EPOLL */
954 struct address_space *f_mapping;
956 errseq_t f_sb_err; /* for syncfs */
958 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
963 /* file identifier */
964 unsigned char f_handle[];
967 static inline struct file *get_file(struct file *f)
969 atomic_long_inc(&f->f_count);
972 #define get_file_rcu_many(x, cnt) \
973 atomic_long_add_unless(&(x)->f_count, (cnt), 0)
974 #define get_file_rcu(x) get_file_rcu_many((x), 1)
975 #define file_count(x) atomic_long_read(&(x)->f_count)
977 #define MAX_NON_LFS ((1UL<<31) - 1)
979 /* Page cache limit. The filesystems should put that into their s_maxbytes
980 limits, otherwise bad things can happen in VM. */
981 #if BITS_PER_LONG==32
982 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
983 #elif BITS_PER_LONG==64
984 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
989 #define FL_DELEG 4 /* NFSv4 delegation */
990 #define FL_ACCESS 8 /* not trying to lock, just looking */
991 #define FL_EXISTS 16 /* when unlocking, test for existence */
992 #define FL_LEASE 32 /* lease held on this file */
993 #define FL_CLOSE 64 /* unlock on close */
994 #define FL_SLEEP 128 /* A blocking lock */
995 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
996 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
997 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
998 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
1000 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
1003 * Special return value from posix_lock_file() and vfs_lock_file() for
1004 * asynchronous locking.
1006 #define FILE_LOCK_DEFERRED 1
1008 /* legacy typedef, should eventually be removed */
1009 typedef void *fl_owner_t;
1013 struct file_lock_operations {
1014 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1015 void (*fl_release_private)(struct file_lock *);
1018 struct lock_manager_operations {
1019 fl_owner_t (*lm_get_owner)(fl_owner_t);
1020 void (*lm_put_owner)(fl_owner_t);
1021 void (*lm_notify)(struct file_lock *); /* unblock callback */
1022 int (*lm_grant)(struct file_lock *, int);
1023 bool (*lm_break)(struct file_lock *);
1024 int (*lm_change)(struct file_lock *, int, struct list_head *);
1025 void (*lm_setup)(struct file_lock *, void **);
1026 bool (*lm_breaker_owns_lease)(struct file_lock *);
1029 struct lock_manager {
1030 struct list_head list;
1032 * NFSv4 and up also want opens blocked during the grace period;
1039 void locks_start_grace(struct net *, struct lock_manager *);
1040 void locks_end_grace(struct lock_manager *);
1041 bool locks_in_grace(struct net *);
1042 bool opens_in_grace(struct net *);
1044 /* that will die - we need it for nfs_lock_info */
1045 #include <linux/nfs_fs_i.h>
1048 * struct file_lock represents a generic "file lock". It's used to represent
1049 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1050 * note that the same struct is used to represent both a request for a lock and
1051 * the lock itself, but the same object is never used for both.
1053 * FIXME: should we create a separate "struct lock_request" to help distinguish
1056 * The varous i_flctx lists are ordered by:
1059 * 2) lock range start
1062 * Obviously, the last two criteria only matter for POSIX locks.
1065 struct file_lock *fl_blocker; /* The lock, that is blocking us */
1066 struct list_head fl_list; /* link into file_lock_context */
1067 struct hlist_node fl_link; /* node in global lists */
1068 struct list_head fl_blocked_requests; /* list of requests with
1069 * ->fl_blocker pointing here
1071 struct list_head fl_blocked_member; /* node in
1072 * ->fl_blocker->fl_blocked_requests
1074 fl_owner_t fl_owner;
1075 unsigned int fl_flags;
1076 unsigned char fl_type;
1077 unsigned int fl_pid;
1078 int fl_link_cpu; /* what cpu's list is this on? */
1079 wait_queue_head_t fl_wait;
1080 struct file *fl_file;
1084 struct fasync_struct * fl_fasync; /* for lease break notifications */
1085 /* for lease breaks: */
1086 unsigned long fl_break_time;
1087 unsigned long fl_downgrade_time;
1089 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1090 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1092 struct nfs_lock_info nfs_fl;
1093 struct nfs4_lock_info nfs4_fl;
1095 struct list_head link; /* link in AFS vnode's pending_locks list */
1096 int state; /* state of grant or error if -ve */
1097 unsigned int debug_id;
1100 } __randomize_layout;
1102 struct file_lock_context {
1103 spinlock_t flc_lock;
1104 struct list_head flc_flock;
1105 struct list_head flc_posix;
1106 struct list_head flc_lease;
1109 /* The following constant reflects the upper bound of the file/locking space */
1111 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1112 #define OFFSET_MAX INT_LIMIT(loff_t)
1113 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1116 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1118 #define locks_inode(f) file_inode(f)
1120 #ifdef CONFIG_FILE_LOCKING
1121 extern int fcntl_getlk(struct file *, unsigned int, struct flock *);
1122 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1125 #if BITS_PER_LONG == 32
1126 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *);
1127 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1131 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1132 extern int fcntl_getlease(struct file *filp);
1135 void locks_free_lock_context(struct inode *inode);
1136 void locks_free_lock(struct file_lock *fl);
1137 extern void locks_init_lock(struct file_lock *);
1138 extern struct file_lock * locks_alloc_lock(void);
1139 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1140 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1141 extern void locks_remove_posix(struct file *, fl_owner_t);
1142 extern void locks_remove_file(struct file *);
1143 extern void locks_release_private(struct file_lock *);
1144 extern void posix_test_lock(struct file *, struct file_lock *);
1145 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1146 extern int locks_delete_block(struct file_lock *);
1147 extern int vfs_test_lock(struct file *, struct file_lock *);
1148 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1149 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1150 bool vfs_inode_has_locks(struct inode *inode);
1151 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1152 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1153 extern void lease_get_mtime(struct inode *, struct timespec64 *time);
1154 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1155 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1156 extern int lease_modify(struct file_lock *, int, struct list_head *);
1158 struct notifier_block;
1159 extern int lease_register_notifier(struct notifier_block *);
1160 extern void lease_unregister_notifier(struct notifier_block *);
1162 struct files_struct;
1163 extern void show_fd_locks(struct seq_file *f,
1164 struct file *filp, struct files_struct *files);
1165 #else /* !CONFIG_FILE_LOCKING */
1166 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1167 struct flock __user *user)
1172 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1173 unsigned int cmd, struct flock __user *user)
1178 #if BITS_PER_LONG == 32
1179 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1180 struct flock64 __user *user)
1185 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1186 unsigned int cmd, struct flock64 __user *user)
1191 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1196 static inline int fcntl_getlease(struct file *filp)
1202 locks_free_lock_context(struct inode *inode)
1206 static inline void locks_init_lock(struct file_lock *fl)
1211 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1216 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1221 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1226 static inline void locks_remove_file(struct file *filp)
1231 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1236 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1237 struct file_lock *conflock)
1242 static inline int locks_delete_block(struct file_lock *waiter)
1247 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1252 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1253 struct file_lock *fl, struct file_lock *conf)
1258 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1263 static inline bool vfs_inode_has_locks(struct inode *inode)
1268 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1273 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1278 static inline void lease_get_mtime(struct inode *inode,
1279 struct timespec64 *time)
1284 static inline int generic_setlease(struct file *filp, long arg,
1285 struct file_lock **flp, void **priv)
1290 static inline int vfs_setlease(struct file *filp, long arg,
1291 struct file_lock **lease, void **priv)
1296 static inline int lease_modify(struct file_lock *fl, int arg,
1297 struct list_head *dispose)
1302 struct files_struct;
1303 static inline void show_fd_locks(struct seq_file *f,
1304 struct file *filp, struct files_struct *files) {}
1305 #endif /* !CONFIG_FILE_LOCKING */
1307 static inline struct inode *file_inode(const struct file *f)
1312 static inline struct dentry *file_dentry(const struct file *file)
1314 return d_real(file->f_path.dentry, file_inode(file));
1317 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1319 return locks_lock_inode_wait(locks_inode(filp), fl);
1322 struct fasync_struct {
1326 struct fasync_struct *fa_next; /* singly linked list */
1327 struct file *fa_file;
1328 struct rcu_head fa_rcu;
1331 #define FASYNC_MAGIC 0x4601
1333 /* SMP safe fasync helpers: */
1334 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1335 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1336 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1337 extern struct fasync_struct *fasync_alloc(void);
1338 extern void fasync_free(struct fasync_struct *);
1340 /* can be called from interrupts */
1341 extern void kill_fasync(struct fasync_struct **, int, int);
1343 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1344 extern int f_setown(struct file *filp, unsigned long arg, int force);
1345 extern void f_delown(struct file *filp);
1346 extern pid_t f_getown(struct file *filp);
1347 extern int send_sigurg(struct fown_struct *fown);
1350 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1351 * represented in both.
1353 #define SB_RDONLY BIT(0) /* Mount read-only */
1354 #define SB_NOSUID BIT(1) /* Ignore suid and sgid bits */
1355 #define SB_NODEV BIT(2) /* Disallow access to device special files */
1356 #define SB_NOEXEC BIT(3) /* Disallow program execution */
1357 #define SB_SYNCHRONOUS BIT(4) /* Writes are synced at once */
1358 #define SB_MANDLOCK BIT(6) /* Allow mandatory locks on an FS */
1359 #define SB_DIRSYNC BIT(7) /* Directory modifications are synchronous */
1360 #define SB_NOATIME BIT(10) /* Do not update access times. */
1361 #define SB_NODIRATIME BIT(11) /* Do not update directory access times */
1362 #define SB_SILENT BIT(15)
1363 #define SB_POSIXACL BIT(16) /* VFS does not apply the umask */
1364 #define SB_INLINECRYPT BIT(17) /* Use blk-crypto for encrypted files */
1365 #define SB_KERNMOUNT BIT(22) /* this is a kern_mount call */
1366 #define SB_I_VERSION BIT(23) /* Update inode I_version field */
1367 #define SB_LAZYTIME BIT(25) /* Update the on-disk [acm]times lazily */
1369 /* These sb flags are internal to the kernel */
1370 #define SB_SUBMOUNT BIT(26)
1371 #define SB_FORCE BIT(27)
1372 #define SB_NOSEC BIT(28)
1373 #define SB_BORN BIT(29)
1374 #define SB_ACTIVE BIT(30)
1375 #define SB_NOUSER BIT(31)
1377 /* These flags relate to encoding and casefolding */
1378 #define SB_ENC_STRICT_MODE_FL (1 << 0)
1380 #define sb_has_strict_encoding(sb) \
1381 (sb->s_encoding_flags & SB_ENC_STRICT_MODE_FL)
1387 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1388 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1389 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1390 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1391 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1394 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1395 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1396 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1397 #define SB_I_STABLE_WRITES 0x00000008 /* don't modify blks until WB is done */
1399 /* sb->s_iflags to limit user namespace mounts */
1400 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1401 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1402 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1404 #define SB_I_SKIP_SYNC 0x00000100 /* Skip superblock at global sync */
1406 /* Possible states of 'frozen' field */
1408 SB_UNFROZEN = 0, /* FS is unfrozen */
1409 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1410 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1411 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1412 * internal threads if needed) */
1413 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1416 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1419 int frozen; /* Is sb frozen? */
1420 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1421 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1424 struct super_block {
1425 struct list_head s_list; /* Keep this first */
1426 dev_t s_dev; /* search index; _not_ kdev_t */
1427 unsigned char s_blocksize_bits;
1428 unsigned long s_blocksize;
1429 loff_t s_maxbytes; /* Max file size */
1430 struct file_system_type *s_type;
1431 const struct super_operations *s_op;
1432 const struct dquot_operations *dq_op;
1433 const struct quotactl_ops *s_qcop;
1434 const struct export_operations *s_export_op;
1435 unsigned long s_flags;
1436 unsigned long s_iflags; /* internal SB_I_* flags */
1437 unsigned long s_magic;
1438 struct dentry *s_root;
1439 struct rw_semaphore s_umount;
1442 #ifdef CONFIG_SECURITY
1445 const struct xattr_handler **s_xattr;
1446 #ifdef CONFIG_FS_ENCRYPTION
1447 const struct fscrypt_operations *s_cop;
1448 struct fscrypt_keyring *s_master_keys; /* master crypto keys in use */
1450 #ifdef CONFIG_FS_VERITY
1451 const struct fsverity_operations *s_vop;
1453 #ifdef CONFIG_UNICODE
1454 struct unicode_map *s_encoding;
1455 __u16 s_encoding_flags;
1457 struct hlist_bl_head s_roots; /* alternate root dentries for NFS */
1458 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1459 struct block_device *s_bdev;
1460 struct backing_dev_info *s_bdi;
1461 struct mtd_info *s_mtd;
1462 struct hlist_node s_instances;
1463 unsigned int s_quota_types; /* Bitmask of supported quota types */
1464 struct quota_info s_dquot; /* Diskquota specific options */
1466 struct sb_writers s_writers;
1469 * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and
1470 * s_fsnotify_marks together for cache efficiency. They are frequently
1471 * accessed and rarely modified.
1473 void *s_fs_info; /* Filesystem private info */
1475 /* Granularity of c/m/atime in ns (cannot be worse than a second) */
1477 /* Time limits for c/m/atime in seconds */
1478 time64_t s_time_min;
1479 time64_t s_time_max;
1480 #ifdef CONFIG_FSNOTIFY
1481 __u32 s_fsnotify_mask;
1482 struct fsnotify_mark_connector __rcu *s_fsnotify_marks;
1485 char s_id[32]; /* Informational name */
1486 uuid_t s_uuid; /* UUID */
1488 unsigned int s_max_links;
1492 * The next field is for VFS *only*. No filesystems have any business
1493 * even looking at it. You had been warned.
1495 struct mutex s_vfs_rename_mutex; /* Kludge */
1498 * Filesystem subtype. If non-empty the filesystem type field
1499 * in /proc/mounts will be "type.subtype"
1501 const char *s_subtype;
1503 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1506 * Saved pool identifier for cleancache (-1 means none)
1508 int cleancache_poolid;
1510 struct shrinker s_shrink; /* per-sb shrinker handle */
1512 /* Number of inodes with nlink == 0 but still referenced */
1513 atomic_long_t s_remove_count;
1515 /* Pending fsnotify inode refs */
1516 atomic_long_t s_fsnotify_inode_refs;
1518 /* Being remounted read-only */
1519 int s_readonly_remount;
1521 /* per-sb errseq_t for reporting writeback errors via syncfs */
1524 /* AIO completions deferred from interrupt context */
1525 struct workqueue_struct *s_dio_done_wq;
1526 struct hlist_head s_pins;
1529 * Owning user namespace and default context in which to
1530 * interpret filesystem uids, gids, quotas, device nodes,
1531 * xattrs and security labels.
1533 struct user_namespace *s_user_ns;
1536 * The list_lru structure is essentially just a pointer to a table
1537 * of per-node lru lists, each of which has its own spinlock.
1538 * There is no need to put them into separate cachelines.
1540 struct list_lru s_dentry_lru;
1541 struct list_lru s_inode_lru;
1542 struct rcu_head rcu;
1543 struct work_struct destroy_work;
1545 struct mutex s_sync_lock; /* sync serialisation lock */
1548 * Indicates how deep in a filesystem stack this SB is
1552 /* s_inode_list_lock protects s_inodes */
1553 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1554 struct list_head s_inodes; /* all inodes */
1556 spinlock_t s_inode_wblist_lock;
1557 struct list_head s_inodes_wb; /* writeback inodes */
1558 } __randomize_layout;
1560 /* Helper functions so that in most cases filesystems will
1561 * not need to deal directly with kuid_t and kgid_t and can
1562 * instead deal with the raw numeric values that are stored
1563 * in the filesystem.
1565 static inline uid_t i_uid_read(const struct inode *inode)
1567 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid);
1570 static inline gid_t i_gid_read(const struct inode *inode)
1572 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid);
1575 static inline void i_uid_write(struct inode *inode, uid_t uid)
1577 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid);
1580 static inline void i_gid_write(struct inode *inode, gid_t gid)
1582 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid);
1585 struct timespec64 current_time(struct inode *inode);
1586 struct timespec64 inode_set_ctime_current(struct inode *inode);
1589 * inode_get_ctime - fetch the current ctime from the inode
1590 * @inode: inode from which to fetch ctime
1592 * Grab the current ctime from the inode and return it.
1594 static inline struct timespec64 inode_get_ctime(const struct inode *inode)
1596 return inode->i_ctime;
1600 * inode_set_ctime_to_ts - set the ctime in the inode
1601 * @inode: inode in which to set the ctime
1602 * @ts: value to set in the ctime field
1604 * Set the ctime in @inode to @ts
1606 static inline struct timespec64 inode_set_ctime_to_ts(struct inode *inode,
1607 struct timespec64 ts)
1609 inode->i_ctime = ts;
1614 * inode_set_ctime - set the ctime in the inode
1615 * @inode: inode in which to set the ctime
1616 * @sec: tv_sec value to set
1617 * @nsec: tv_nsec value to set
1619 * Set the ctime in @inode to { @sec, @nsec }
1621 static inline struct timespec64 inode_set_ctime(struct inode *inode,
1622 time64_t sec, long nsec)
1624 struct timespec64 ts = { .tv_sec = sec,
1627 return inode_set_ctime_to_ts(inode, ts);
1631 * Snapshotting support.
1635 * These are internal functions, please use sb_start_{write,pagefault,intwrite}
1638 static inline void __sb_end_write(struct super_block *sb, int level)
1640 percpu_up_read(sb->s_writers.rw_sem + level-1);
1643 static inline void __sb_start_write(struct super_block *sb, int level)
1645 percpu_down_read(sb->s_writers.rw_sem + level - 1);
1648 static inline bool __sb_start_write_trylock(struct super_block *sb, int level)
1650 return percpu_down_read_trylock(sb->s_writers.rw_sem + level - 1);
1653 #define __sb_writers_acquired(sb, lev) \
1654 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1655 #define __sb_writers_release(sb, lev) \
1656 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1659 * sb_end_write - drop write access to a superblock
1660 * @sb: the super we wrote to
1662 * Decrement number of writers to the filesystem. Wake up possible waiters
1663 * wanting to freeze the filesystem.
1665 static inline void sb_end_write(struct super_block *sb)
1667 __sb_end_write(sb, SB_FREEZE_WRITE);
1671 * sb_end_pagefault - drop write access to a superblock from a page fault
1672 * @sb: the super we wrote to
1674 * Decrement number of processes handling write page fault to the filesystem.
1675 * Wake up possible waiters wanting to freeze the filesystem.
1677 static inline void sb_end_pagefault(struct super_block *sb)
1679 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1683 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1684 * @sb: the super we wrote to
1686 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1687 * waiters wanting to freeze the filesystem.
1689 static inline void sb_end_intwrite(struct super_block *sb)
1691 __sb_end_write(sb, SB_FREEZE_FS);
1695 * sb_start_write - get write access to a superblock
1696 * @sb: the super we write to
1698 * When a process wants to write data or metadata to a file system (i.e. dirty
1699 * a page or an inode), it should embed the operation in a sb_start_write() -
1700 * sb_end_write() pair to get exclusion against file system freezing. This
1701 * function increments number of writers preventing freezing. If the file
1702 * system is already frozen, the function waits until the file system is
1705 * Since freeze protection behaves as a lock, users have to preserve
1706 * ordering of freeze protection and other filesystem locks. Generally,
1707 * freeze protection should be the outermost lock. In particular, we have:
1710 * -> i_mutex (write path, truncate, directory ops, ...)
1711 * -> s_umount (freeze_super, thaw_super)
1713 static inline void sb_start_write(struct super_block *sb)
1715 __sb_start_write(sb, SB_FREEZE_WRITE);
1718 static inline bool sb_start_write_trylock(struct super_block *sb)
1720 return __sb_start_write_trylock(sb, SB_FREEZE_WRITE);
1724 * sb_start_pagefault - get write access to a superblock from a page fault
1725 * @sb: the super we write to
1727 * When a process starts handling write page fault, it should embed the
1728 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1729 * exclusion against file system freezing. This is needed since the page fault
1730 * is going to dirty a page. This function increments number of running page
1731 * faults preventing freezing. If the file system is already frozen, the
1732 * function waits until the file system is thawed.
1734 * Since page fault freeze protection behaves as a lock, users have to preserve
1735 * ordering of freeze protection and other filesystem locks. It is advised to
1736 * put sb_start_pagefault() close to mmap_lock in lock ordering. Page fault
1737 * handling code implies lock dependency:
1740 * -> sb_start_pagefault
1742 static inline void sb_start_pagefault(struct super_block *sb)
1744 __sb_start_write(sb, SB_FREEZE_PAGEFAULT);
1748 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1749 * @sb: the super we write to
1751 * This is the third level of protection against filesystem freezing. It is
1752 * free for use by a filesystem. The only requirement is that it must rank
1753 * below sb_start_pagefault.
1755 * For example filesystem can call sb_start_intwrite() when starting a
1756 * transaction which somewhat eases handling of freezing for internal sources
1757 * of filesystem changes (internal fs threads, discarding preallocation on file
1760 static inline void sb_start_intwrite(struct super_block *sb)
1762 __sb_start_write(sb, SB_FREEZE_FS);
1765 static inline bool sb_start_intwrite_trylock(struct super_block *sb)
1767 return __sb_start_write_trylock(sb, SB_FREEZE_FS);
1771 extern bool inode_owner_or_capable(const struct inode *inode);
1774 * VFS helper functions..
1776 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1777 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1778 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1779 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1780 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1781 extern int vfs_rmdir(struct inode *, struct dentry *);
1782 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1783 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1785 static inline int vfs_whiteout(struct inode *dir, struct dentry *dentry)
1787 return vfs_mknod(dir, dentry, S_IFCHR | WHITEOUT_MODE, WHITEOUT_DEV);
1790 extern struct dentry *vfs_tmpfile(struct dentry *dentry, umode_t mode,
1793 int vfs_mkobj(struct dentry *, umode_t,
1794 int (*f)(struct dentry *, umode_t, void *),
1797 int vfs_fchown(struct file *file, uid_t user, gid_t group);
1798 int vfs_fchmod(struct file *file, umode_t mode);
1799 int vfs_utimes(const struct path *path, struct timespec64 *times);
1801 extern long vfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
1803 #ifdef CONFIG_COMPAT
1804 extern long compat_ptr_ioctl(struct file *file, unsigned int cmd,
1807 #define compat_ptr_ioctl NULL
1811 * VFS file helper functions.
1813 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1815 extern bool may_open_dev(const struct path *path);
1816 umode_t mode_strip_sgid(const struct inode *dir, umode_t mode);
1819 * This is the "filldir" function type, used by readdir() to let
1820 * the kernel specify what kind of dirent layout it wants to have.
1821 * This allows the kernel to read directories into kernel space or
1822 * to have different dirent layouts depending on the binary type.
1825 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1828 struct dir_context {
1834 * These flags let !MMU mmap() govern direct device mapping vs immediate
1835 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1837 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1838 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1839 * NOMMU_MAP_READ: Can be mapped for reading
1840 * NOMMU_MAP_WRITE: Can be mapped for writing
1841 * NOMMU_MAP_EXEC: Can be mapped for execution
1843 #define NOMMU_MAP_COPY 0x00000001
1844 #define NOMMU_MAP_DIRECT 0x00000008
1845 #define NOMMU_MAP_READ VM_MAYREAD
1846 #define NOMMU_MAP_WRITE VM_MAYWRITE
1847 #define NOMMU_MAP_EXEC VM_MAYEXEC
1849 #define NOMMU_VMFLAGS \
1850 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1853 * These flags control the behavior of the remap_file_range function pointer.
1854 * If it is called with len == 0 that means "remap to end of source file".
1855 * See Documentation/filesystems/vfs.rst for more details about this call.
1857 * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate)
1858 * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request
1860 #define REMAP_FILE_DEDUP (1 << 0)
1861 #define REMAP_FILE_CAN_SHORTEN (1 << 1)
1864 * These flags signal that the caller is ok with altering various aspects of
1865 * the behavior of the remap operation. The changes must be made by the
1866 * implementation; the vfs remap helper functions can take advantage of them.
1867 * Flags in this category exist to preserve the quirky behavior of the hoisted
1868 * btrfs clone/dedupe ioctls.
1870 #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN)
1873 * These flags control the behavior of vfs_copy_file_range().
1874 * They are not available to the user via syscall.
1876 * COPY_FILE_SPLICE: call splice direct instead of fs clone/copy ops
1878 #define COPY_FILE_SPLICE (1 << 0)
1882 struct file_operations {
1883 struct module *owner;
1884 loff_t (*llseek) (struct file *, loff_t, int);
1885 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1886 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1887 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1888 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1889 int (*iopoll)(struct kiocb *kiocb, bool spin);
1890 int (*iterate) (struct file *, struct dir_context *);
1891 int (*iterate_shared) (struct file *, struct dir_context *);
1892 __poll_t (*poll) (struct file *, struct poll_table_struct *);
1893 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1894 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1895 int (*mmap) (struct file *, struct vm_area_struct *);
1896 unsigned long mmap_supported_flags;
1897 int (*open) (struct inode *, struct file *);
1898 int (*flush) (struct file *, fl_owner_t id);
1899 int (*release) (struct inode *, struct file *);
1900 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1901 int (*fasync) (int, struct file *, int);
1902 int (*lock) (struct file *, int, struct file_lock *);
1903 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1904 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1905 int (*check_flags)(int);
1906 int (*flock) (struct file *, int, struct file_lock *);
1907 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1908 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1909 int (*setlease)(struct file *, long, struct file_lock **, void **);
1910 long (*fallocate)(struct file *file, int mode, loff_t offset,
1912 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1914 unsigned (*mmap_capabilities)(struct file *);
1916 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1917 loff_t, size_t, unsigned int);
1918 loff_t (*remap_file_range)(struct file *file_in, loff_t pos_in,
1919 struct file *file_out, loff_t pos_out,
1920 loff_t len, unsigned int remap_flags);
1921 int (*fadvise)(struct file *, loff_t, loff_t, int);
1923 } __randomize_layout;
1925 struct inode_operations {
1926 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1927 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
1928 int (*permission) (struct inode *, int);
1929 struct posix_acl * (*get_acl)(struct inode *, int);
1931 int (*readlink) (struct dentry *, char __user *,int);
1933 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1934 int (*link) (struct dentry *,struct inode *,struct dentry *);
1935 int (*unlink) (struct inode *,struct dentry *);
1936 int (*symlink) (struct inode *,struct dentry *,const char *);
1937 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1938 int (*rmdir) (struct inode *,struct dentry *);
1939 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1940 int (*rename) (struct inode *, struct dentry *,
1941 struct inode *, struct dentry *, unsigned int);
1942 int (*setattr) (struct dentry *, struct iattr *);
1943 int (*getattr) (const struct path *, struct kstat *, u32, unsigned int);
1944 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1945 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1947 int (*update_time)(struct inode *, struct timespec64 *, int);
1948 int (*atomic_open)(struct inode *, struct dentry *,
1949 struct file *, unsigned open_flag,
1950 umode_t create_mode);
1951 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1952 int (*set_acl)(struct inode *, struct posix_acl *, int);
1953 } ____cacheline_aligned;
1955 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
1956 struct iov_iter *iter)
1958 return file->f_op->read_iter(kio, iter);
1961 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
1962 struct iov_iter *iter)
1964 return file->f_op->write_iter(kio, iter);
1967 static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
1969 return file->f_op->mmap(file, vma);
1972 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1973 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1974 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1975 loff_t, size_t, unsigned int);
1976 extern ssize_t generic_copy_file_range(struct file *file_in, loff_t pos_in,
1977 struct file *file_out, loff_t pos_out,
1978 size_t len, unsigned int flags);
1979 extern int generic_remap_file_range_prep(struct file *file_in, loff_t pos_in,
1980 struct file *file_out, loff_t pos_out,
1982 unsigned int remap_flags);
1983 extern loff_t do_clone_file_range(struct file *file_in, loff_t pos_in,
1984 struct file *file_out, loff_t pos_out,
1985 loff_t len, unsigned int remap_flags);
1986 extern loff_t vfs_clone_file_range(struct file *file_in, loff_t pos_in,
1987 struct file *file_out, loff_t pos_out,
1988 loff_t len, unsigned int remap_flags);
1989 extern int vfs_dedupe_file_range(struct file *file,
1990 struct file_dedupe_range *same);
1991 extern loff_t vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos,
1992 struct file *dst_file, loff_t dst_pos,
1993 loff_t len, unsigned int remap_flags);
1996 struct super_operations {
1997 struct inode *(*alloc_inode)(struct super_block *sb);
1998 void (*destroy_inode)(struct inode *);
1999 void (*free_inode)(struct inode *);
2001 void (*dirty_inode) (struct inode *, int flags);
2002 int (*write_inode) (struct inode *, struct writeback_control *wbc);
2003 int (*drop_inode) (struct inode *);
2004 void (*evict_inode) (struct inode *);
2005 void (*put_super) (struct super_block *);
2006 int (*sync_fs)(struct super_block *sb, int wait);
2007 int (*freeze_super) (struct super_block *);
2008 int (*freeze_fs) (struct super_block *);
2009 int (*thaw_super) (struct super_block *);
2010 int (*unfreeze_fs) (struct super_block *);
2011 int (*statfs) (struct dentry *, struct kstatfs *);
2012 int (*remount_fs) (struct super_block *, int *, char *);
2013 void (*umount_begin) (struct super_block *);
2015 int (*show_options)(struct seq_file *, struct dentry *);
2016 int (*show_devname)(struct seq_file *, struct dentry *);
2017 int (*show_path)(struct seq_file *, struct dentry *);
2018 int (*show_stats)(struct seq_file *, struct dentry *);
2020 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
2021 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
2022 struct dquot **(*get_dquots)(struct inode *);
2024 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
2025 long (*nr_cached_objects)(struct super_block *,
2026 struct shrink_control *);
2027 long (*free_cached_objects)(struct super_block *,
2028 struct shrink_control *);
2032 * Inode flags - they have no relation to superblock flags now
2034 #define S_SYNC (1 << 0) /* Writes are synced at once */
2035 #define S_NOATIME (1 << 1) /* Do not update access times */
2036 #define S_APPEND (1 << 2) /* Append-only file */
2037 #define S_IMMUTABLE (1 << 3) /* Immutable file */
2038 #define S_DEAD (1 << 4) /* removed, but still open directory */
2039 #define S_NOQUOTA (1 << 5) /* Inode is not counted to quota */
2040 #define S_DIRSYNC (1 << 6) /* Directory modifications are synchronous */
2041 #define S_NOCMTIME (1 << 7) /* Do not update file c/mtime */
2042 #define S_SWAPFILE (1 << 8) /* Do not truncate: swapon got its bmaps */
2043 #define S_PRIVATE (1 << 9) /* Inode is fs-internal */
2044 #define S_IMA (1 << 10) /* Inode has an associated IMA struct */
2045 #define S_AUTOMOUNT (1 << 11) /* Automount/referral quasi-directory */
2046 #define S_NOSEC (1 << 12) /* no suid or xattr security attributes */
2047 #ifdef CONFIG_FS_DAX
2048 #define S_DAX (1 << 13) /* Direct Access, avoiding the page cache */
2050 #define S_DAX 0 /* Make all the DAX code disappear */
2052 #define S_ENCRYPTED (1 << 14) /* Encrypted file (using fs/crypto/) */
2053 #define S_CASEFOLD (1 << 15) /* Casefolded file */
2054 #define S_VERITY (1 << 16) /* Verity file (using fs/verity/) */
2057 * Note that nosuid etc flags are inode-specific: setting some file-system
2058 * flags just means all the inodes inherit those flags by default. It might be
2059 * possible to override it selectively if you really wanted to with some
2060 * ioctl() that is not currently implemented.
2062 * Exception: SB_RDONLY is always applied to the entire file system.
2064 * Unfortunately, it is possible to change a filesystems flags with it mounted
2065 * with files in use. This means that all of the inodes will not have their
2066 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
2067 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
2069 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
2071 static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; }
2072 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
2073 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
2074 ((inode)->i_flags & S_SYNC))
2075 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
2076 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
2077 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
2078 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
2079 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
2081 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
2082 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
2083 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
2084 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
2086 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
2087 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
2088 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
2089 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
2090 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
2091 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
2092 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
2093 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
2094 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
2095 #define IS_CASEFOLDED(inode) ((inode)->i_flags & S_CASEFOLD)
2096 #define IS_VERITY(inode) ((inode)->i_flags & S_VERITY)
2098 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
2099 (inode)->i_rdev == WHITEOUT_DEV)
2101 static inline bool HAS_UNMAPPED_ID(struct inode *inode)
2103 return !uid_valid(inode->i_uid) || !gid_valid(inode->i_gid);
2106 static inline enum rw_hint file_write_hint(struct file *file)
2108 if (file->f_write_hint != WRITE_LIFE_NOT_SET)
2109 return file->f_write_hint;
2111 return file_inode(file)->i_write_hint;
2114 static inline int iocb_flags(struct file *file);
2116 static inline u16 ki_hint_validate(enum rw_hint hint)
2118 typeof(((struct kiocb *)0)->ki_hint) max_hint = -1;
2120 if (hint <= max_hint)
2125 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
2127 *kiocb = (struct kiocb) {
2129 .ki_flags = iocb_flags(filp),
2130 .ki_hint = ki_hint_validate(file_write_hint(filp)),
2131 .ki_ioprio = get_current_ioprio(),
2135 static inline void kiocb_clone(struct kiocb *kiocb, struct kiocb *kiocb_src,
2138 *kiocb = (struct kiocb) {
2140 .ki_flags = kiocb_src->ki_flags,
2141 .ki_hint = kiocb_src->ki_hint,
2142 .ki_ioprio = kiocb_src->ki_ioprio,
2143 .ki_pos = kiocb_src->ki_pos,
2148 * Inode state bits. Protected by inode->i_lock
2150 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
2151 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
2153 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
2154 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
2155 * various stages of removing an inode.
2157 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2159 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
2160 * fdatasync(). i_atime is the usual cause.
2161 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
2162 * these changes separately from I_DIRTY_SYNC so that we
2163 * don't have to write inode on fdatasync() when only
2164 * mtime has changed in it.
2165 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
2166 * I_NEW Serves as both a mutex and completion notification.
2167 * New inodes set I_NEW. If two processes both create
2168 * the same inode, one of them will release its inode and
2169 * wait for I_NEW to be released before returning.
2170 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2171 * also cause waiting on I_NEW, without I_NEW actually
2172 * being set. find_inode() uses this to prevent returning
2173 * nearly-dead inodes.
2174 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2175 * is zero. I_FREEING must be set when I_WILL_FREE is
2177 * I_FREEING Set when inode is about to be freed but still has dirty
2178 * pages or buffers attached or the inode itself is still
2180 * I_CLEAR Added by clear_inode(). In this state the inode is
2181 * clean and can be destroyed. Inode keeps I_FREEING.
2183 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2184 * prohibited for many purposes. iget() must wait for
2185 * the inode to be completely released, then create it
2186 * anew. Other functions will just ignore such inodes,
2187 * if appropriate. I_NEW is used for waiting.
2189 * I_SYNC Writeback of inode is running. The bit is set during
2190 * data writeback, and cleared with a wakeup on the bit
2191 * address once it is done. The bit is also used to pin
2192 * the inode in memory for flusher thread.
2194 * I_REFERENCED Marks the inode as recently references on the LRU list.
2196 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2198 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2199 * synchronize competing switching instances and to tell
2200 * wb stat updates to grab the i_pages lock. See
2201 * inode_switch_wbs_work_fn() for details.
2203 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2204 * and work dirs among overlayfs mounts.
2206 * I_CREATING New object's inode in the middle of setting up.
2208 * I_DONTCACHE Evict inode as soon as it is not used anymore.
2210 * I_SYNC_QUEUED Inode is queued in b_io or b_more_io writeback lists.
2211 * Used to detect that mark_inode_dirty() should not move
2212 * inode between dirty lists.
2214 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2216 #define I_DIRTY_SYNC (1 << 0)
2217 #define I_DIRTY_DATASYNC (1 << 1)
2218 #define I_DIRTY_PAGES (1 << 2)
2220 #define I_NEW (1 << __I_NEW)
2221 #define I_WILL_FREE (1 << 4)
2222 #define I_FREEING (1 << 5)
2223 #define I_CLEAR (1 << 6)
2225 #define I_SYNC (1 << __I_SYNC)
2226 #define I_REFERENCED (1 << 8)
2227 #define __I_DIO_WAKEUP 9
2228 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2229 #define I_LINKABLE (1 << 10)
2230 #define I_DIRTY_TIME (1 << 11)
2231 #define I_WB_SWITCH (1 << 13)
2232 #define I_OVL_INUSE (1 << 14)
2233 #define I_CREATING (1 << 15)
2234 #define I_DONTCACHE (1 << 16)
2235 #define I_SYNC_QUEUED (1 << 17)
2237 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2238 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2239 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2241 extern void __mark_inode_dirty(struct inode *, int);
2242 static inline void mark_inode_dirty(struct inode *inode)
2244 __mark_inode_dirty(inode, I_DIRTY);
2247 static inline void mark_inode_dirty_sync(struct inode *inode)
2249 __mark_inode_dirty(inode, I_DIRTY_SYNC);
2252 extern void inc_nlink(struct inode *inode);
2253 extern void drop_nlink(struct inode *inode);
2254 extern void clear_nlink(struct inode *inode);
2255 extern void set_nlink(struct inode *inode, unsigned int nlink);
2257 static inline void inode_inc_link_count(struct inode *inode)
2260 mark_inode_dirty(inode);
2263 static inline void inode_dec_link_count(struct inode *inode)
2266 mark_inode_dirty(inode);
2269 enum file_time_flags {
2276 extern bool atime_needs_update(const struct path *, struct inode *);
2277 extern void touch_atime(const struct path *);
2278 int inode_update_time(struct inode *inode, struct timespec64 *time, int flags);
2280 static inline void file_accessed(struct file *file)
2282 if (!(file->f_flags & O_NOATIME))
2283 touch_atime(&file->f_path);
2286 extern int file_modified(struct file *file);
2288 int sync_inode(struct inode *inode, struct writeback_control *wbc);
2289 int sync_inode_metadata(struct inode *inode, int wait);
2291 struct file_system_type {
2294 #define FS_REQUIRES_DEV 1
2295 #define FS_BINARY_MOUNTDATA 2
2296 #define FS_HAS_SUBTYPE 4
2297 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2298 #define FS_DISALLOW_NOTIFY_PERM 16 /* Disable fanotify permission events */
2299 #define FS_THP_SUPPORT 8192 /* Remove once all fs converted */
2300 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2301 int (*init_fs_context)(struct fs_context *);
2302 const struct fs_parameter_spec *parameters;
2303 struct dentry *(*mount) (struct file_system_type *, int,
2304 const char *, void *);
2305 void (*kill_sb) (struct super_block *);
2306 struct module *owner;
2307 struct file_system_type * next;
2308 struct hlist_head fs_supers;
2310 struct lock_class_key s_lock_key;
2311 struct lock_class_key s_umount_key;
2312 struct lock_class_key s_vfs_rename_key;
2313 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2315 struct lock_class_key i_lock_key;
2316 struct lock_class_key i_mutex_key;
2317 struct lock_class_key i_mutex_dir_key;
2320 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2322 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2323 int flags, const char *dev_name, void *data,
2324 int (*fill_super)(struct super_block *, void *, int));
2325 extern struct dentry *mount_single(struct file_system_type *fs_type,
2326 int flags, void *data,
2327 int (*fill_super)(struct super_block *, void *, int));
2328 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2329 int flags, void *data,
2330 int (*fill_super)(struct super_block *, void *, int));
2331 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2332 void generic_shutdown_super(struct super_block *sb);
2333 void kill_block_super(struct super_block *sb);
2334 void kill_anon_super(struct super_block *sb);
2335 void kill_litter_super(struct super_block *sb);
2336 void deactivate_super(struct super_block *sb);
2337 void deactivate_locked_super(struct super_block *sb);
2338 int set_anon_super(struct super_block *s, void *data);
2339 int set_anon_super_fc(struct super_block *s, struct fs_context *fc);
2340 int get_anon_bdev(dev_t *);
2341 void free_anon_bdev(dev_t);
2342 struct super_block *sget_fc(struct fs_context *fc,
2343 int (*test)(struct super_block *, struct fs_context *),
2344 int (*set)(struct super_block *, struct fs_context *));
2345 struct super_block *sget(struct file_system_type *type,
2346 int (*test)(struct super_block *,void *),
2347 int (*set)(struct super_block *,void *),
2348 int flags, void *data);
2350 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2351 #define fops_get(fops) \
2352 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2353 #define fops_put(fops) \
2354 do { if (fops) module_put((fops)->owner); } while(0)
2356 * This one is to be used *ONLY* from ->open() instances.
2357 * fops must be non-NULL, pinned down *and* module dependencies
2358 * should be sufficient to pin the caller down as well.
2360 #define replace_fops(f, fops) \
2362 struct file *__file = (f); \
2363 fops_put(__file->f_op); \
2364 BUG_ON(!(__file->f_op = (fops))); \
2367 extern int register_filesystem(struct file_system_type *);
2368 extern int unregister_filesystem(struct file_system_type *);
2369 extern struct vfsmount *kern_mount(struct file_system_type *);
2370 extern void kern_unmount(struct vfsmount *mnt);
2371 extern int may_umount_tree(struct vfsmount *);
2372 extern int may_umount(struct vfsmount *);
2373 extern long do_mount(const char *, const char __user *,
2374 const char *, unsigned long, void *);
2375 extern struct vfsmount *collect_mounts(const struct path *);
2376 extern void drop_collected_mounts(struct vfsmount *);
2377 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2379 extern int vfs_statfs(const struct path *, struct kstatfs *);
2380 extern int user_statfs(const char __user *, struct kstatfs *);
2381 extern int fd_statfs(int, struct kstatfs *);
2382 extern int freeze_super(struct super_block *super);
2383 extern int thaw_super(struct super_block *super);
2384 extern bool our_mnt(struct vfsmount *mnt);
2385 extern __printf(2, 3)
2386 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2387 extern int super_setup_bdi(struct super_block *sb);
2389 extern int current_umask(void);
2391 extern void ihold(struct inode * inode);
2392 extern void iput(struct inode *);
2393 extern int generic_update_time(struct inode *, struct timespec64 *, int);
2396 extern struct kobject *fs_kobj;
2398 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2400 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2401 extern int locks_mandatory_locked(struct file *);
2402 extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
2405 * Candidates for mandatory locking have the setgid bit set
2406 * but no group execute bit - an otherwise meaningless combination.
2409 static inline int __mandatory_lock(struct inode *ino)
2411 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2415 * ... and these candidates should be on SB_MANDLOCK mounted fs,
2416 * otherwise these will be advisory locks
2419 static inline int mandatory_lock(struct inode *ino)
2421 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2424 static inline int locks_verify_locked(struct file *file)
2426 if (mandatory_lock(locks_inode(file)))
2427 return locks_mandatory_locked(file);
2431 static inline int locks_verify_truncate(struct inode *inode,
2435 if (!inode->i_flctx || !mandatory_lock(inode))
2438 if (size < inode->i_size) {
2439 return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2442 return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2447 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2449 static inline int locks_mandatory_locked(struct file *file)
2454 static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2455 loff_t start, loff_t end, unsigned char type)
2460 static inline int __mandatory_lock(struct inode *inode)
2465 static inline int mandatory_lock(struct inode *inode)
2470 static inline int locks_verify_locked(struct file *file)
2475 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2481 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2484 #ifdef CONFIG_FILE_LOCKING
2485 static inline int break_lease(struct inode *inode, unsigned int mode)
2488 * Since this check is lockless, we must ensure that any refcounts
2489 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2490 * could end up racing with tasks trying to set a new lease on this
2494 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2495 return __break_lease(inode, mode, FL_LEASE);
2499 static inline int break_deleg(struct inode *inode, unsigned int mode)
2502 * Since this check is lockless, we must ensure that any refcounts
2503 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2504 * could end up racing with tasks trying to set a new lease on this
2508 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2509 return __break_lease(inode, mode, FL_DELEG);
2513 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2517 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2518 if (ret == -EWOULDBLOCK && delegated_inode) {
2519 *delegated_inode = inode;
2525 static inline int break_deleg_wait(struct inode **delegated_inode)
2529 ret = break_deleg(*delegated_inode, O_WRONLY);
2530 iput(*delegated_inode);
2531 *delegated_inode = NULL;
2535 static inline int break_layout(struct inode *inode, bool wait)
2538 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2539 return __break_lease(inode,
2540 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2545 #else /* !CONFIG_FILE_LOCKING */
2546 static inline int break_lease(struct inode *inode, unsigned int mode)
2551 static inline int break_deleg(struct inode *inode, unsigned int mode)
2556 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2561 static inline int break_deleg_wait(struct inode **delegated_inode)
2567 static inline int break_layout(struct inode *inode, bool wait)
2572 #endif /* CONFIG_FILE_LOCKING */
2577 const char *name; /* pointer to actual string */
2578 const __user char *uptr; /* original userland pointer */
2580 struct audit_names *aname;
2583 static_assert(offsetof(struct filename, iname) % sizeof(long) == 0);
2585 extern long vfs_truncate(const struct path *, loff_t);
2586 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2588 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2590 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2592 extern struct file *file_open_name(struct filename *, int, umode_t);
2593 extern struct file *filp_open(const char *, int, umode_t);
2594 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2595 const char *, int, umode_t);
2596 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2597 extern struct file * open_with_fake_path(const struct path *, int,
2598 struct inode*, const struct cred *);
2599 static inline struct file *file_clone_open(struct file *file)
2601 return dentry_open(&file->f_path, file->f_flags, file->f_cred);
2603 extern int filp_close(struct file *, fl_owner_t id);
2605 extern struct filename *getname_flags(const char __user *, int, int *);
2606 extern struct filename *getname(const char __user *);
2607 extern struct filename *getname_kernel(const char *);
2608 extern void putname(struct filename *name);
2610 extern int finish_open(struct file *file, struct dentry *dentry,
2611 int (*open)(struct inode *, struct file *));
2612 extern int finish_no_open(struct file *file, struct dentry *dentry);
2615 extern void __init vfs_caches_init_early(void);
2616 extern void __init vfs_caches_init(void);
2618 extern struct kmem_cache *names_cachep;
2620 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2621 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2623 extern struct super_block *blockdev_superblock;
2624 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2626 return IS_ENABLED(CONFIG_BLOCK) && sb == blockdev_superblock;
2629 void emergency_thaw_all(void);
2630 extern int sync_filesystem(struct super_block *);
2631 extern const struct file_operations def_blk_fops;
2632 extern const struct file_operations def_chr_fops;
2635 #define CHRDEV_MAJOR_MAX 512
2636 /* Marks the bottom of the first segment of free char majors */
2637 #define CHRDEV_MAJOR_DYN_END 234
2638 /* Marks the top and bottom of the second segment of free char majors */
2639 #define CHRDEV_MAJOR_DYN_EXT_START 511
2640 #define CHRDEV_MAJOR_DYN_EXT_END 384
2642 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2643 extern int register_chrdev_region(dev_t, unsigned, const char *);
2644 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2645 unsigned int count, const char *name,
2646 const struct file_operations *fops);
2647 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2648 unsigned int count, const char *name);
2649 extern void unregister_chrdev_region(dev_t, unsigned);
2650 extern void chrdev_show(struct seq_file *,off_t);
2652 static inline int register_chrdev(unsigned int major, const char *name,
2653 const struct file_operations *fops)
2655 return __register_chrdev(major, 0, 256, name, fops);
2658 static inline void unregister_chrdev(unsigned int major, const char *name)
2660 __unregister_chrdev(major, 0, 256, name);
2663 extern void init_special_inode(struct inode *, umode_t, dev_t);
2665 /* Invalid inode operations -- fs/bad_inode.c */
2666 extern void make_bad_inode(struct inode *);
2667 extern bool is_bad_inode(struct inode *);
2669 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2670 pgoff_t start, pgoff_t end);
2672 void invalidate_mapping_pagevec(struct address_space *mapping,
2673 pgoff_t start, pgoff_t end,
2674 unsigned long *nr_pagevec);
2676 static inline void invalidate_remote_inode(struct inode *inode)
2678 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2679 S_ISLNK(inode->i_mode))
2680 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2682 extern int invalidate_inode_pages2(struct address_space *mapping);
2683 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2684 pgoff_t start, pgoff_t end);
2685 extern int write_inode_now(struct inode *, int);
2686 extern int filemap_fdatawrite(struct address_space *);
2687 extern int filemap_flush(struct address_space *);
2688 extern int filemap_fdatawait_keep_errors(struct address_space *mapping);
2689 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2691 extern int filemap_fdatawait_range_keep_errors(struct address_space *mapping,
2692 loff_t start_byte, loff_t end_byte);
2694 static inline int filemap_fdatawait(struct address_space *mapping)
2696 return filemap_fdatawait_range(mapping, 0, LLONG_MAX);
2699 extern bool filemap_range_has_page(struct address_space *, loff_t lstart,
2701 extern int filemap_write_and_wait_range(struct address_space *mapping,
2702 loff_t lstart, loff_t lend);
2703 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2704 loff_t start, loff_t end, int sync_mode);
2705 extern int filemap_fdatawrite_range(struct address_space *mapping,
2706 loff_t start, loff_t end);
2707 extern int filemap_check_errors(struct address_space *mapping);
2708 extern void __filemap_set_wb_err(struct address_space *mapping, int err);
2710 static inline int filemap_write_and_wait(struct address_space *mapping)
2712 return filemap_write_and_wait_range(mapping, 0, LLONG_MAX);
2715 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2717 extern int __must_check file_check_and_advance_wb_err(struct file *file);
2718 extern int __must_check file_write_and_wait_range(struct file *file,
2719 loff_t start, loff_t end);
2721 static inline int file_write_and_wait(struct file *file)
2723 return file_write_and_wait_range(file, 0, LLONG_MAX);
2727 * filemap_set_wb_err - set a writeback error on an address_space
2728 * @mapping: mapping in which to set writeback error
2729 * @err: error to be set in mapping
2731 * When writeback fails in some way, we must record that error so that
2732 * userspace can be informed when fsync and the like are called. We endeavor
2733 * to report errors on any file that was open at the time of the error. Some
2734 * internal callers also need to know when writeback errors have occurred.
2736 * When a writeback error occurs, most filesystems will want to call
2737 * filemap_set_wb_err to record the error in the mapping so that it will be
2738 * automatically reported whenever fsync is called on the file.
2740 static inline void filemap_set_wb_err(struct address_space *mapping, int err)
2742 /* Fastpath for common case of no error */
2744 __filemap_set_wb_err(mapping, err);
2748 * filemap_check_wb_err - has an error occurred since the mark was sampled?
2749 * @mapping: mapping to check for writeback errors
2750 * @since: previously-sampled errseq_t
2752 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2753 * the given value was sampled.
2755 * If it has then report the latest error set, otherwise return 0.
2757 static inline int filemap_check_wb_err(struct address_space *mapping,
2760 return errseq_check(&mapping->wb_err, since);
2764 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2765 * @mapping: mapping to be sampled
2767 * Writeback errors are always reported relative to a particular sample point
2768 * in the past. This function provides those sample points.
2770 static inline errseq_t filemap_sample_wb_err(struct address_space *mapping)
2772 return errseq_sample(&mapping->wb_err);
2776 * file_sample_sb_err - sample the current errseq_t to test for later errors
2777 * @file: file pointer to be sampled
2779 * Grab the most current superblock-level errseq_t value for the given
2782 static inline errseq_t file_sample_sb_err(struct file *file)
2784 return errseq_sample(&file->f_path.dentry->d_sb->s_wb_err);
2787 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2789 extern int vfs_fsync(struct file *file, int datasync);
2791 extern int sync_file_range(struct file *file, loff_t offset, loff_t nbytes,
2792 unsigned int flags);
2795 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2796 * to already be updated for the write, and will return either the amount
2797 * of bytes passed in, or an error if syncing the file failed.
2799 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2801 if (iocb->ki_flags & IOCB_DSYNC) {
2802 int ret = vfs_fsync_range(iocb->ki_filp,
2803 iocb->ki_pos - count, iocb->ki_pos - 1,
2804 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2812 extern void emergency_sync(void);
2813 extern void emergency_remount(void);
2816 extern int bmap(struct inode *inode, sector_t *block);
2818 static inline int bmap(struct inode *inode, sector_t *block)
2824 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2825 extern int inode_permission(struct inode *, int);
2826 extern int generic_permission(struct inode *, int);
2827 extern int __check_sticky(struct inode *dir, struct inode *inode);
2829 static inline bool execute_ok(struct inode *inode)
2831 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2834 static inline bool inode_wrong_type(const struct inode *inode, umode_t mode)
2836 return (inode->i_mode ^ mode) & S_IFMT;
2839 static inline void file_start_write(struct file *file)
2841 if (!S_ISREG(file_inode(file)->i_mode))
2843 sb_start_write(file_inode(file)->i_sb);
2846 static inline bool file_start_write_trylock(struct file *file)
2848 if (!S_ISREG(file_inode(file)->i_mode))
2850 return sb_start_write_trylock(file_inode(file)->i_sb);
2853 static inline void file_end_write(struct file *file)
2855 if (!S_ISREG(file_inode(file)->i_mode))
2857 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2861 * get_write_access() gets write permission for a file.
2862 * put_write_access() releases this write permission.
2863 * This is used for regular files.
2864 * We cannot support write (and maybe mmap read-write shared) accesses and
2865 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2866 * can have the following values:
2867 * 0: no writers, no VM_DENYWRITE mappings
2868 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2869 * > 0: (i_writecount) users are writing to the file.
2871 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2872 * except for the cases where we don't hold i_writecount yet. Then we need to
2873 * use {get,deny}_write_access() - these functions check the sign and refuse
2874 * to do the change if sign is wrong.
2876 static inline int get_write_access(struct inode *inode)
2878 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2880 static inline int deny_write_access(struct file *file)
2882 struct inode *inode = file_inode(file);
2883 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2885 static inline void put_write_access(struct inode * inode)
2887 atomic_dec(&inode->i_writecount);
2889 static inline void allow_write_access(struct file *file)
2892 atomic_inc(&file_inode(file)->i_writecount);
2894 static inline bool inode_is_open_for_write(const struct inode *inode)
2896 return atomic_read(&inode->i_writecount) > 0;
2899 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
2900 static inline void i_readcount_dec(struct inode *inode)
2902 BUG_ON(!atomic_read(&inode->i_readcount));
2903 atomic_dec(&inode->i_readcount);
2905 static inline void i_readcount_inc(struct inode *inode)
2907 atomic_inc(&inode->i_readcount);
2910 static inline void i_readcount_dec(struct inode *inode)
2914 static inline void i_readcount_inc(struct inode *inode)
2919 extern int do_pipe_flags(int *, int);
2921 extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
2922 ssize_t __kernel_read(struct file *file, void *buf, size_t count, loff_t *pos);
2923 extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
2924 extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
2925 extern struct file * open_exec(const char *);
2927 /* fs/dcache.c -- generic fs support functions */
2928 extern bool is_subdir(struct dentry *, struct dentry *);
2929 extern bool path_is_under(const struct path *, const struct path *);
2931 extern char *file_path(struct file *, char *, int);
2933 #include <linux/err.h>
2935 /* needed for stackable file system support */
2936 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2938 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2940 extern int inode_init_always(struct super_block *, struct inode *);
2941 extern void inode_init_once(struct inode *);
2942 extern void address_space_init_once(struct address_space *mapping);
2943 extern struct inode * igrab(struct inode *);
2944 extern ino_t iunique(struct super_block *, ino_t);
2945 extern int inode_needs_sync(struct inode *inode);
2946 extern int generic_delete_inode(struct inode *inode);
2947 static inline int generic_drop_inode(struct inode *inode)
2949 return !inode->i_nlink || inode_unhashed(inode);
2951 extern void d_mark_dontcache(struct inode *inode);
2953 extern struct inode *ilookup5_nowait(struct super_block *sb,
2954 unsigned long hashval, int (*test)(struct inode *, void *),
2956 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2957 int (*test)(struct inode *, void *), void *data);
2958 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2960 extern struct inode *inode_insert5(struct inode *inode, unsigned long hashval,
2961 int (*test)(struct inode *, void *),
2962 int (*set)(struct inode *, void *),
2964 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2965 extern struct inode * iget_locked(struct super_block *, unsigned long);
2966 extern struct inode *find_inode_nowait(struct super_block *,
2968 int (*match)(struct inode *,
2969 unsigned long, void *),
2971 extern struct inode *find_inode_rcu(struct super_block *, unsigned long,
2972 int (*)(struct inode *, void *), void *);
2973 extern struct inode *find_inode_by_ino_rcu(struct super_block *, unsigned long);
2974 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2975 extern int insert_inode_locked(struct inode *);
2976 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2977 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2979 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2981 extern void unlock_new_inode(struct inode *);
2982 extern void discard_new_inode(struct inode *);
2983 extern unsigned int get_next_ino(void);
2984 extern void evict_inodes(struct super_block *sb);
2987 * Userspace may rely on the the inode number being non-zero. For example, glibc
2988 * simply ignores files with zero i_ino in unlink() and other places.
2990 * As an additional complication, if userspace was compiled with
2991 * _FILE_OFFSET_BITS=32 on a 64-bit kernel we'll only end up reading out the
2992 * lower 32 bits, so we need to check that those aren't zero explicitly. With
2993 * _FILE_OFFSET_BITS=64, this may cause some harmless false-negatives, but
2994 * better safe than sorry.
2996 static inline bool is_zero_ino(ino_t ino)
2998 return (u32)ino == 0;
3001 extern void __iget(struct inode * inode);
3002 extern void iget_failed(struct inode *);
3003 extern void clear_inode(struct inode *);
3004 extern void __destroy_inode(struct inode *);
3005 extern struct inode *new_inode_pseudo(struct super_block *sb);
3006 extern struct inode *new_inode(struct super_block *sb);
3007 extern void free_inode_nonrcu(struct inode *inode);
3008 extern int setattr_should_drop_suidgid(struct inode *);
3009 extern int file_remove_privs(struct file *);
3011 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
3012 static inline void insert_inode_hash(struct inode *inode)
3014 __insert_inode_hash(inode, inode->i_ino);
3017 extern void __remove_inode_hash(struct inode *);
3018 static inline void remove_inode_hash(struct inode *inode)
3020 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
3021 __remove_inode_hash(inode);
3024 extern void inode_sb_list_add(struct inode *inode);
3026 extern int sb_set_blocksize(struct super_block *, int);
3027 extern int sb_min_blocksize(struct super_block *, int);
3029 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
3030 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
3031 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
3032 extern int generic_write_check_limits(struct file *file, loff_t pos,
3034 extern int generic_file_rw_checks(struct file *file_in, struct file *file_out);
3035 extern ssize_t generic_file_buffered_read(struct kiocb *iocb,
3036 struct iov_iter *to, ssize_t already_read);
3037 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
3038 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
3039 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
3040 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
3041 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
3043 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
3045 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
3047 ssize_t vfs_iocb_iter_read(struct file *file, struct kiocb *iocb,
3048 struct iov_iter *iter);
3049 ssize_t vfs_iocb_iter_write(struct file *file, struct kiocb *iocb,
3050 struct iov_iter *iter);
3052 /* fs/block_dev.c */
3053 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
3054 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
3055 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
3057 extern void block_sync_page(struct page *page);
3060 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
3061 struct pipe_inode_info *, size_t, unsigned int);
3062 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
3063 struct file *, loff_t *, size_t, unsigned int);
3064 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
3065 struct file *out, loff_t *, size_t len, unsigned int flags);
3066 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
3067 loff_t *opos, size_t len, unsigned int flags);
3071 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
3072 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
3073 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
3074 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
3075 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
3076 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
3077 int whence, loff_t maxsize, loff_t eof);
3078 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
3079 int whence, loff_t size);
3080 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
3081 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
3082 extern int generic_file_open(struct inode * inode, struct file * filp);
3083 extern int nonseekable_open(struct inode * inode, struct file * filp);
3084 extern int stream_open(struct inode * inode, struct file * filp);
3087 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
3088 loff_t file_offset);
3091 /* need locking between buffered and direct access */
3094 /* filesystem does not support filling holes */
3095 DIO_SKIP_HOLES = 0x02,
3098 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
3099 struct block_device *bdev, struct iov_iter *iter,
3100 get_block_t get_block,
3101 dio_iodone_t end_io, dio_submit_t submit_io,
3104 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
3105 struct inode *inode,
3106 struct iov_iter *iter,
3107 get_block_t get_block)
3109 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
3110 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
3114 void inode_dio_wait(struct inode *inode);
3117 * inode_dio_begin - signal start of a direct I/O requests
3118 * @inode: inode the direct I/O happens on
3120 * This is called once we've finished processing a direct I/O request,
3121 * and is used to wake up callers waiting for direct I/O to be quiesced.
3123 static inline void inode_dio_begin(struct inode *inode)
3125 atomic_inc(&inode->i_dio_count);
3129 * inode_dio_end - signal finish of a direct I/O requests
3130 * @inode: inode the direct I/O happens on
3132 * This is called once we've finished processing a direct I/O request,
3133 * and is used to wake up callers waiting for direct I/O to be quiesced.
3135 static inline void inode_dio_end(struct inode *inode)
3137 if (atomic_dec_and_test(&inode->i_dio_count))
3138 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
3142 * Warn about a page cache invalidation failure diring a direct I/O write.
3144 void dio_warn_stale_pagecache(struct file *filp);
3146 extern void inode_set_flags(struct inode *inode, unsigned int flags,
3149 extern const struct file_operations generic_ro_fops;
3151 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3153 extern int readlink_copy(char __user *, int, const char *);
3154 extern int page_readlink(struct dentry *, char __user *, int);
3155 extern const char *page_get_link(struct dentry *, struct inode *,
3156 struct delayed_call *);
3157 extern void page_put_link(void *);
3158 extern int __page_symlink(struct inode *inode, const char *symname, int len,
3160 extern int page_symlink(struct inode *inode, const char *symname, int len);
3161 extern const struct inode_operations page_symlink_inode_operations;
3162 extern void kfree_link(void *);
3163 extern void generic_fillattr(struct inode *, struct kstat *);
3164 extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
3165 extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
3166 void __inode_add_bytes(struct inode *inode, loff_t bytes);
3167 void inode_add_bytes(struct inode *inode, loff_t bytes);
3168 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
3169 void inode_sub_bytes(struct inode *inode, loff_t bytes);
3170 static inline loff_t __inode_get_bytes(struct inode *inode)
3172 return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes;
3174 loff_t inode_get_bytes(struct inode *inode);
3175 void inode_set_bytes(struct inode *inode, loff_t bytes);
3176 const char *simple_get_link(struct dentry *, struct inode *,
3177 struct delayed_call *);
3178 extern const struct inode_operations simple_symlink_inode_operations;
3180 extern int iterate_dir(struct file *, struct dir_context *);
3182 int vfs_fstatat(int dfd, const char __user *filename, struct kstat *stat,
3184 int vfs_fstat(int fd, struct kstat *stat);
3186 static inline int vfs_stat(const char __user *filename, struct kstat *stat)
3188 return vfs_fstatat(AT_FDCWD, filename, stat, 0);
3190 static inline int vfs_lstat(const char __user *name, struct kstat *stat)
3192 return vfs_fstatat(AT_FDCWD, name, stat, AT_SYMLINK_NOFOLLOW);
3195 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
3196 extern int vfs_readlink(struct dentry *, char __user *, int);
3198 extern struct file_system_type *get_filesystem(struct file_system_type *fs);
3199 extern void put_filesystem(struct file_system_type *fs);
3200 extern struct file_system_type *get_fs_type(const char *name);
3201 extern struct super_block *get_super(struct block_device *);
3202 extern struct super_block *get_super_thawed(struct block_device *);
3203 extern struct super_block *get_super_exclusive_thawed(struct block_device *bdev);
3204 extern struct super_block *get_active_super(struct block_device *bdev);
3205 extern void drop_super(struct super_block *sb);
3206 extern void drop_super_exclusive(struct super_block *sb);
3207 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
3208 extern void iterate_supers_type(struct file_system_type *,
3209 void (*)(struct super_block *, void *), void *);
3211 extern int dcache_dir_open(struct inode *, struct file *);
3212 extern int dcache_dir_close(struct inode *, struct file *);
3213 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
3214 extern int dcache_readdir(struct file *, struct dir_context *);
3215 extern int simple_setattr(struct dentry *, struct iattr *);
3216 extern int simple_getattr(const struct path *, struct kstat *, u32, unsigned int);
3217 extern int simple_statfs(struct dentry *, struct kstatfs *);
3218 extern int simple_open(struct inode *inode, struct file *file);
3219 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3220 extern int simple_unlink(struct inode *, struct dentry *);
3221 extern int simple_rmdir(struct inode *, struct dentry *);
3222 extern int simple_rename(struct inode *, struct dentry *,
3223 struct inode *, struct dentry *, unsigned int);
3224 extern void simple_recursive_removal(struct dentry *,
3225 void (*callback)(struct dentry *));
3226 extern int noop_fsync(struct file *, loff_t, loff_t, int);
3227 extern int noop_set_page_dirty(struct page *page);
3228 extern void noop_invalidatepage(struct page *page, unsigned int offset,
3229 unsigned int length);
3230 extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter);
3231 extern int simple_empty(struct dentry *);
3232 extern int simple_readpage(struct file *file, struct page *page);
3233 extern int simple_write_begin(struct file *file, struct address_space *mapping,
3234 loff_t pos, unsigned len, unsigned flags,
3235 struct page **pagep, void **fsdata);
3236 extern int simple_write_end(struct file *file, struct address_space *mapping,
3237 loff_t pos, unsigned len, unsigned copied,
3238 struct page *page, void *fsdata);
3239 extern int always_delete_dentry(const struct dentry *);
3240 extern struct inode *alloc_anon_inode(struct super_block *);
3241 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
3242 extern const struct dentry_operations simple_dentry_operations;
3244 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3245 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3246 extern const struct file_operations simple_dir_operations;
3247 extern const struct inode_operations simple_dir_inode_operations;
3248 extern void make_empty_dir_inode(struct inode *inode);
3249 extern bool is_empty_dir_inode(struct inode *inode);
3250 struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
3251 struct dentry *d_alloc_name(struct dentry *, const char *);
3252 extern int simple_fill_super(struct super_block *, unsigned long,
3253 const struct tree_descr *);
3254 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3255 extern void simple_release_fs(struct vfsmount **mount, int *count);
3257 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3258 loff_t *ppos, const void *from, size_t available);
3259 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3260 const void __user *from, size_t count);
3262 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3263 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3265 extern int generic_check_addressable(unsigned, u64);
3267 #ifdef CONFIG_UNICODE
3268 extern int generic_ci_d_hash(const struct dentry *dentry, struct qstr *str);
3269 extern int generic_ci_d_compare(const struct dentry *dentry, unsigned int len,
3270 const char *str, const struct qstr *name);
3273 #ifdef CONFIG_MIGRATION
3274 extern int buffer_migrate_page(struct address_space *,
3275 struct page *, struct page *,
3277 extern int buffer_migrate_page_norefs(struct address_space *,
3278 struct page *, struct page *,
3281 #define buffer_migrate_page NULL
3282 #define buffer_migrate_page_norefs NULL
3285 extern int setattr_prepare(struct dentry *, struct iattr *);
3286 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3287 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
3289 extern int file_update_time(struct file *file);
3291 static inline bool vma_is_dax(const struct vm_area_struct *vma)
3293 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3296 static inline bool vma_is_fsdax(struct vm_area_struct *vma)
3298 struct inode *inode;
3302 if (!vma_is_dax(vma))
3304 inode = file_inode(vma->vm_file);
3305 if (S_ISCHR(inode->i_mode))
3306 return false; /* device-dax */
3310 static inline int iocb_flags(struct file *file)
3313 if (file->f_flags & O_APPEND)
3315 if (file->f_flags & O_DIRECT)
3317 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3319 if (file->f_flags & __O_SYNC)
3324 static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
3326 int kiocb_flags = 0;
3328 /* make sure there's no overlap between RWF and private IOCB flags */
3329 BUILD_BUG_ON((__force int) RWF_SUPPORTED & IOCB_EVENTFD);
3333 if (unlikely(flags & ~RWF_SUPPORTED))
3336 if (flags & RWF_NOWAIT) {
3337 if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
3339 kiocb_flags |= IOCB_NOIO;
3341 kiocb_flags |= (__force int) (flags & RWF_SUPPORTED);
3342 if (flags & RWF_SYNC)
3343 kiocb_flags |= IOCB_DSYNC;
3345 ki->ki_flags |= kiocb_flags;
3349 static inline ino_t parent_ino(struct dentry *dentry)
3354 * Don't strictly need d_lock here? If the parent ino could change
3355 * then surely we'd have a deeper race in the caller?
3357 spin_lock(&dentry->d_lock);
3358 res = dentry->d_parent->d_inode->i_ino;
3359 spin_unlock(&dentry->d_lock);
3363 /* Transaction based IO helpers */
3366 * An argresp is stored in an allocated page and holds the
3367 * size of the argument or response, along with its content
3369 struct simple_transaction_argresp {
3374 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3376 char *simple_transaction_get(struct file *file, const char __user *buf,
3378 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3379 size_t size, loff_t *pos);
3380 int simple_transaction_release(struct inode *inode, struct file *file);
3382 void simple_transaction_set(struct file *file, size_t n);
3385 * simple attribute files
3387 * These attributes behave similar to those in sysfs:
3389 * Writing to an attribute immediately sets a value, an open file can be
3390 * written to multiple times.
3392 * Reading from an attribute creates a buffer from the value that might get
3393 * read with multiple read calls. When the attribute has been read
3394 * completely, no further read calls are possible until the file is opened
3397 * All attributes contain a text representation of a numeric value
3398 * that are accessed with the get() and set() functions.
3400 #define DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, __is_signed) \
3401 static int __fops ## _open(struct inode *inode, struct file *file) \
3403 __simple_attr_check_format(__fmt, 0ull); \
3404 return simple_attr_open(inode, file, __get, __set, __fmt); \
3406 static const struct file_operations __fops = { \
3407 .owner = THIS_MODULE, \
3408 .open = __fops ## _open, \
3409 .release = simple_attr_release, \
3410 .read = simple_attr_read, \
3411 .write = (__is_signed) ? simple_attr_write_signed : simple_attr_write, \
3412 .llseek = generic_file_llseek, \
3415 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3416 DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, false)
3418 #define DEFINE_SIMPLE_ATTRIBUTE_SIGNED(__fops, __get, __set, __fmt) \
3419 DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, true)
3421 static inline __printf(1, 2)
3422 void __simple_attr_check_format(const char *fmt, ...)
3424 /* don't do anything, just let the compiler check the arguments; */
3427 int simple_attr_open(struct inode *inode, struct file *file,
3428 int (*get)(void *, u64 *), int (*set)(void *, u64),
3430 int simple_attr_release(struct inode *inode, struct file *file);
3431 ssize_t simple_attr_read(struct file *file, char __user *buf,
3432 size_t len, loff_t *ppos);
3433 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3434 size_t len, loff_t *ppos);
3435 ssize_t simple_attr_write_signed(struct file *file, const char __user *buf,
3436 size_t len, loff_t *ppos);
3439 int proc_nr_files(struct ctl_table *table, int write,
3440 void *buffer, size_t *lenp, loff_t *ppos);
3441 int proc_nr_dentry(struct ctl_table *table, int write,
3442 void *buffer, size_t *lenp, loff_t *ppos);
3443 int proc_nr_inodes(struct ctl_table *table, int write,
3444 void *buffer, size_t *lenp, loff_t *ppos);
3445 int __init get_filesystem_list(char *buf);
3447 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3448 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3450 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3451 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3452 (flag & __FMODE_NONOTIFY)))
3454 static inline bool is_sxid(umode_t mode)
3456 return mode & (S_ISUID | S_ISGID);
3459 static inline int check_sticky(struct inode *dir, struct inode *inode)
3461 if (!(dir->i_mode & S_ISVTX))
3464 return __check_sticky(dir, inode);
3467 static inline void inode_has_no_xattr(struct inode *inode)
3469 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
3470 inode->i_flags |= S_NOSEC;
3473 static inline bool is_root_inode(struct inode *inode)
3475 return inode == inode->i_sb->s_root->d_inode;
3478 static inline bool dir_emit(struct dir_context *ctx,
3479 const char *name, int namelen,
3480 u64 ino, unsigned type)
3482 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3484 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3486 return ctx->actor(ctx, ".", 1, ctx->pos,
3487 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3489 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3491 return ctx->actor(ctx, "..", 2, ctx->pos,
3492 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3494 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3496 if (ctx->pos == 0) {
3497 if (!dir_emit_dot(file, ctx))
3501 if (ctx->pos == 1) {
3502 if (!dir_emit_dotdot(file, ctx))
3508 static inline bool dir_relax(struct inode *inode)
3510 inode_unlock(inode);
3512 return !IS_DEADDIR(inode);
3515 static inline bool dir_relax_shared(struct inode *inode)
3517 inode_unlock_shared(inode);
3518 inode_lock_shared(inode);
3519 return !IS_DEADDIR(inode);
3522 extern bool path_noexec(const struct path *path);
3523 extern void inode_nohighmem(struct inode *inode);
3526 extern int vfs_fadvise(struct file *file, loff_t offset, loff_t len,
3528 extern int generic_fadvise(struct file *file, loff_t offset, loff_t len,
3531 int vfs_ioc_setflags_prepare(struct inode *inode, unsigned int oldflags,
3532 unsigned int flags);
3534 int vfs_ioc_fssetxattr_check(struct inode *inode, const struct fsxattr *old_fa,
3535 struct fsxattr *fa);
3537 static inline void simple_fill_fsxattr(struct fsxattr *fa, __u32 xflags)
3539 memset(fa, 0, sizeof(*fa));
3540 fa->fsx_xflags = xflags;
3544 * Flush file data before changing attributes. Caller must hold any locks
3545 * required to prevent further writes to this file until we're done setting
3548 static inline int inode_drain_writes(struct inode *inode)
3550 inode_dio_wait(inode);
3551 return filemap_write_and_wait(inode->i_mapping);
3554 #endif /* _LINUX_FS_H */