1 /* SPDX-License-Identifier: GPL-2.0 */
5 #include <linux/linkage.h>
6 #include <linux/wait_bit.h>
7 #include <linux/kdev_t.h>
8 #include <linux/dcache.h>
9 #include <linux/path.h>
10 #include <linux/stat.h>
11 #include <linux/cache.h>
12 #include <linux/list.h>
13 #include <linux/list_lru.h>
14 #include <linux/llist.h>
15 #include <linux/radix-tree.h>
16 #include <linux/xarray.h>
17 #include <linux/rbtree.h>
18 #include <linux/init.h>
19 #include <linux/pid.h>
20 #include <linux/bug.h>
21 #include <linux/mutex.h>
22 #include <linux/rwsem.h>
23 #include <linux/mm_types.h>
24 #include <linux/capability.h>
25 #include <linux/semaphore.h>
26 #include <linux/fcntl.h>
27 #include <linux/rculist_bl.h>
28 #include <linux/atomic.h>
29 #include <linux/shrinker.h>
30 #include <linux/migrate_mode.h>
31 #include <linux/uidgid.h>
32 #include <linux/lockdep.h>
33 #include <linux/percpu-rwsem.h>
34 #include <linux/workqueue.h>
35 #include <linux/delayed_call.h>
36 #include <linux/uuid.h>
37 #include <linux/errseq.h>
38 #include <linux/ioprio.h>
39 #include <linux/fs_types.h>
40 #include <linux/build_bug.h>
41 #include <linux/stddef.h>
42 #include <linux/mount.h>
43 #include <linux/cred.h>
44 #include <linux/mnt_idmapping.h>
45 #include <linux/slab.h>
47 #include <asm/byteorder.h>
48 #include <uapi/linux/fs.h>
50 struct backing_dev_info;
54 struct export_operations;
55 struct fiemap_extent_info;
60 struct pipe_inode_info;
61 struct poll_table_struct;
63 struct vm_area_struct;
66 struct swap_info_struct;
68 struct workqueue_struct;
71 struct fscrypt_operations;
73 struct fsverity_operations;
75 struct fs_parameter_spec;
78 extern void __init inode_init(void);
79 extern void __init inode_init_early(void);
80 extern void __init files_init(void);
81 extern void __init files_maxfiles_init(void);
83 extern unsigned long get_max_files(void);
84 extern unsigned int sysctl_nr_open;
86 typedef __kernel_rwf_t rwf_t;
89 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
90 struct buffer_head *bh_result, int create);
91 typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
92 ssize_t bytes, void *private);
94 #define MAY_EXEC 0x00000001
95 #define MAY_WRITE 0x00000002
96 #define MAY_READ 0x00000004
97 #define MAY_APPEND 0x00000008
98 #define MAY_ACCESS 0x00000010
99 #define MAY_OPEN 0x00000020
100 #define MAY_CHDIR 0x00000040
101 /* called from RCU mode, don't block */
102 #define MAY_NOT_BLOCK 0x00000080
105 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
106 * to O_WRONLY and O_RDWR via the strange trick in do_dentry_open()
109 /* file is open for reading */
110 #define FMODE_READ ((__force fmode_t)0x1)
111 /* file is open for writing */
112 #define FMODE_WRITE ((__force fmode_t)0x2)
113 /* file is seekable */
114 #define FMODE_LSEEK ((__force fmode_t)0x4)
115 /* file can be accessed using pread */
116 #define FMODE_PREAD ((__force fmode_t)0x8)
117 /* file can be accessed using pwrite */
118 #define FMODE_PWRITE ((__force fmode_t)0x10)
119 /* File is opened for execution with sys_execve / sys_uselib */
120 #define FMODE_EXEC ((__force fmode_t)0x20)
121 /* File is opened with O_NDELAY (only set for block devices) */
122 #define FMODE_NDELAY ((__force fmode_t)0x40)
123 /* File is opened with O_EXCL (only set for block devices) */
124 #define FMODE_EXCL ((__force fmode_t)0x80)
125 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
126 (specialy hack for floppy.c) */
127 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
128 /* 32bit hashes as llseek() offset (for directories) */
129 #define FMODE_32BITHASH ((__force fmode_t)0x200)
130 /* 64bit hashes as llseek() offset (for directories) */
131 #define FMODE_64BITHASH ((__force fmode_t)0x400)
134 * Don't update ctime and mtime.
136 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
137 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
139 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
141 /* Expect random access pattern */
142 #define FMODE_RANDOM ((__force fmode_t)0x1000)
144 /* File is huge (eg. /dev/mem): treat loff_t as unsigned */
145 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
147 /* File is opened with O_PATH; almost nothing can be done with it */
148 #define FMODE_PATH ((__force fmode_t)0x4000)
150 /* File needs atomic accesses to f_pos */
151 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
152 /* Write access to underlying fs */
153 #define FMODE_WRITER ((__force fmode_t)0x10000)
154 /* Has read method(s) */
155 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
156 /* Has write method(s) */
157 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
159 #define FMODE_OPENED ((__force fmode_t)0x80000)
160 #define FMODE_CREATED ((__force fmode_t)0x100000)
162 /* File is stream-like */
163 #define FMODE_STREAM ((__force fmode_t)0x200000)
165 /* File was opened by fanotify and shouldn't generate fanotify events */
166 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
168 /* File is capable of returning -EAGAIN if I/O will block */
169 #define FMODE_NOWAIT ((__force fmode_t)0x8000000)
171 /* File represents mount that needs unmounting */
172 #define FMODE_NEED_UNMOUNT ((__force fmode_t)0x10000000)
174 /* File does not contribute to nr_files count */
175 #define FMODE_NOACCOUNT ((__force fmode_t)0x20000000)
177 /* File supports async buffered reads */
178 #define FMODE_BUF_RASYNC ((__force fmode_t)0x40000000)
181 * Attribute flags. These should be or-ed together to figure out what
184 #define ATTR_MODE (1 << 0)
185 #define ATTR_UID (1 << 1)
186 #define ATTR_GID (1 << 2)
187 #define ATTR_SIZE (1 << 3)
188 #define ATTR_ATIME (1 << 4)
189 #define ATTR_MTIME (1 << 5)
190 #define ATTR_CTIME (1 << 6)
191 #define ATTR_ATIME_SET (1 << 7)
192 #define ATTR_MTIME_SET (1 << 8)
193 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
194 #define ATTR_KILL_SUID (1 << 11)
195 #define ATTR_KILL_SGID (1 << 12)
196 #define ATTR_FILE (1 << 13)
197 #define ATTR_KILL_PRIV (1 << 14)
198 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
199 #define ATTR_TIMES_SET (1 << 16)
200 #define ATTR_TOUCH (1 << 17)
203 * Whiteout is represented by a char device. The following constants define the
204 * mode and device number to use.
206 #define WHITEOUT_MODE 0
207 #define WHITEOUT_DEV 0
210 * This is the Inode Attributes structure, used for notify_change(). It
211 * uses the above definitions as flags, to know which values have changed.
212 * Also, in this manner, a Filesystem can look at only the values it cares
213 * about. Basically, these are the attributes that the VFS layer can
214 * request to change from the FS layer.
216 * Derek Atkins <warlord@MIT.EDU> 94-10-20
219 unsigned int ia_valid;
224 struct timespec64 ia_atime;
225 struct timespec64 ia_mtime;
226 struct timespec64 ia_ctime;
229 * Not an attribute, but an auxiliary info for filesystems wanting to
230 * implement an ftruncate() like method. NOTE: filesystem should
231 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
233 struct file *ia_file;
237 * Includes for diskquotas.
239 #include <linux/quota.h>
242 * Maximum number of layers of fs stack. Needs to be limited to
243 * prevent kernel stack overflow
245 #define FILESYSTEM_MAX_STACK_DEPTH 2
248 * enum positive_aop_returns - aop return codes with specific semantics
250 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
251 * completed, that the page is still locked, and
252 * should be considered active. The VM uses this hint
253 * to return the page to the active list -- it won't
254 * be a candidate for writeback again in the near
255 * future. Other callers must be careful to unlock
256 * the page if they get this return. Returned by
259 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
260 * unlocked it and the page might have been truncated.
261 * The caller should back up to acquiring a new page and
262 * trying again. The aop will be taking reasonable
263 * precautions not to livelock. If the caller held a page
264 * reference, it should drop it before retrying. Returned
267 * address_space_operation functions return these large constants to indicate
268 * special semantics to the caller. These are much larger than the bytes in a
269 * page to allow for functions that return the number of bytes operated on in a
273 enum positive_aop_returns {
274 AOP_WRITEPAGE_ACTIVATE = 0x80000,
275 AOP_TRUNCATED_PAGE = 0x80001,
278 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
279 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
280 * helper code (eg buffer layer)
281 * to clear GFP_FS from alloc */
284 * oh the beauties of C type declarations.
287 struct address_space;
288 struct writeback_control;
289 struct readahead_control;
292 * Write life time hint values.
293 * Stored in struct inode as u8.
296 WRITE_LIFE_NOT_SET = 0,
297 WRITE_LIFE_NONE = RWH_WRITE_LIFE_NONE,
298 WRITE_LIFE_SHORT = RWH_WRITE_LIFE_SHORT,
299 WRITE_LIFE_MEDIUM = RWH_WRITE_LIFE_MEDIUM,
300 WRITE_LIFE_LONG = RWH_WRITE_LIFE_LONG,
301 WRITE_LIFE_EXTREME = RWH_WRITE_LIFE_EXTREME,
304 /* Match RWF_* bits to IOCB bits */
305 #define IOCB_HIPRI (__force int) RWF_HIPRI
306 #define IOCB_DSYNC (__force int) RWF_DSYNC
307 #define IOCB_SYNC (__force int) RWF_SYNC
308 #define IOCB_NOWAIT (__force int) RWF_NOWAIT
309 #define IOCB_APPEND (__force int) RWF_APPEND
311 /* non-RWF related bits - start at 16 */
312 #define IOCB_EVENTFD (1 << 16)
313 #define IOCB_DIRECT (1 << 17)
314 #define IOCB_WRITE (1 << 18)
315 /* iocb->ki_waitq is valid */
316 #define IOCB_WAITQ (1 << 19)
317 #define IOCB_NOIO (1 << 20)
318 /* can use bio alloc cache */
319 #define IOCB_ALLOC_CACHE (1 << 21)
322 struct file *ki_filp;
324 /* The 'ki_filp' pointer is shared in a union for aio */
325 randomized_struct_fields_start
328 void (*ki_complete)(struct kiocb *iocb, long ret);
331 u16 ki_ioprio; /* See linux/ioprio.h */
332 struct wait_page_queue *ki_waitq; /* for async buffered IO */
333 randomized_struct_fields_end
336 static inline bool is_sync_kiocb(struct kiocb *kiocb)
338 return kiocb->ki_complete == NULL;
342 * "descriptor" for what we're up to with a read.
343 * This allows us to use the same read code yet
344 * have multiple different users of the data that
345 * we read from a file.
347 * The simplest case just copies the data to user
360 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
361 unsigned long, unsigned long);
363 struct address_space_operations {
364 int (*writepage)(struct page *page, struct writeback_control *wbc);
365 int (*readpage)(struct file *, struct page *);
367 /* Write back some dirty pages from this mapping. */
368 int (*writepages)(struct address_space *, struct writeback_control *);
370 /* Mark a folio dirty. Return true if this dirtied it */
371 bool (*dirty_folio)(struct address_space *, struct folio *);
374 * Reads in the requested pages. Unlike ->readpage(), this is
375 * PURELY used for read-ahead!.
377 int (*readpages)(struct file *filp, struct address_space *mapping,
378 struct list_head *pages, unsigned nr_pages);
379 void (*readahead)(struct readahead_control *);
381 int (*write_begin)(struct file *, struct address_space *mapping,
382 loff_t pos, unsigned len, unsigned flags,
383 struct page **pagep, void **fsdata);
384 int (*write_end)(struct file *, struct address_space *mapping,
385 loff_t pos, unsigned len, unsigned copied,
386 struct page *page, void *fsdata);
388 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
389 sector_t (*bmap)(struct address_space *, sector_t);
390 void (*invalidate_folio) (struct folio *, size_t offset, size_t len);
391 int (*releasepage) (struct page *, gfp_t);
392 void (*freepage)(struct page *);
393 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
395 * migrate the contents of a page to the specified target. If
396 * migrate_mode is MIGRATE_ASYNC, it must not block.
398 int (*migratepage) (struct address_space *,
399 struct page *, struct page *, enum migrate_mode);
400 bool (*isolate_page)(struct page *, isolate_mode_t);
401 void (*putback_page)(struct page *);
402 int (*launder_folio)(struct folio *);
403 bool (*is_partially_uptodate) (struct folio *, size_t from,
405 void (*is_dirty_writeback) (struct page *, bool *, bool *);
406 int (*error_remove_page)(struct address_space *, struct page *);
408 /* swapfile support */
409 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
411 void (*swap_deactivate)(struct file *file);
414 extern const struct address_space_operations empty_aops;
417 * pagecache_write_begin/pagecache_write_end must be used by general code
418 * to write into the pagecache.
420 int pagecache_write_begin(struct file *, struct address_space *mapping,
421 loff_t pos, unsigned len, unsigned flags,
422 struct page **pagep, void **fsdata);
424 int pagecache_write_end(struct file *, struct address_space *mapping,
425 loff_t pos, unsigned len, unsigned copied,
426 struct page *page, void *fsdata);
429 * struct address_space - Contents of a cacheable, mappable object.
430 * @host: Owner, either the inode or the block_device.
431 * @i_pages: Cached pages.
432 * @invalidate_lock: Guards coherency between page cache contents and
433 * file offset->disk block mappings in the filesystem during invalidates.
434 * It is also used to block modification of page cache contents through
436 * @gfp_mask: Memory allocation flags to use for allocating pages.
437 * @i_mmap_writable: Number of VM_SHARED mappings.
438 * @nr_thps: Number of THPs in the pagecache (non-shmem only).
439 * @i_mmap: Tree of private and shared mappings.
440 * @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable.
441 * @nrpages: Number of page entries, protected by the i_pages lock.
442 * @writeback_index: Writeback starts here.
444 * @flags: Error bits and flags (AS_*).
445 * @wb_err: The most recent error which has occurred.
446 * @private_lock: For use by the owner of the address_space.
447 * @private_list: For use by the owner of the address_space.
448 * @private_data: For use by the owner of the address_space.
450 struct address_space {
452 struct xarray i_pages;
453 struct rw_semaphore invalidate_lock;
455 atomic_t i_mmap_writable;
456 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
457 /* number of thp, only for non-shmem files */
460 struct rb_root_cached i_mmap;
461 struct rw_semaphore i_mmap_rwsem;
462 unsigned long nrpages;
463 pgoff_t writeback_index;
464 const struct address_space_operations *a_ops;
467 spinlock_t private_lock;
468 struct list_head private_list;
470 } __attribute__((aligned(sizeof(long)))) __randomize_layout;
472 * On most architectures that alignment is already the case; but
473 * must be enforced here for CRIS, to let the least significant bit
474 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
477 /* XArray tags, for tagging dirty and writeback pages in the pagecache. */
478 #define PAGECACHE_TAG_DIRTY XA_MARK_0
479 #define PAGECACHE_TAG_WRITEBACK XA_MARK_1
480 #define PAGECACHE_TAG_TOWRITE XA_MARK_2
483 * Returns true if any of the pages in the mapping are marked with the tag.
485 static inline bool mapping_tagged(struct address_space *mapping, xa_mark_t tag)
487 return xa_marked(&mapping->i_pages, tag);
490 static inline void i_mmap_lock_write(struct address_space *mapping)
492 down_write(&mapping->i_mmap_rwsem);
495 static inline int i_mmap_trylock_write(struct address_space *mapping)
497 return down_write_trylock(&mapping->i_mmap_rwsem);
500 static inline void i_mmap_unlock_write(struct address_space *mapping)
502 up_write(&mapping->i_mmap_rwsem);
505 static inline void i_mmap_lock_read(struct address_space *mapping)
507 down_read(&mapping->i_mmap_rwsem);
510 static inline void i_mmap_unlock_read(struct address_space *mapping)
512 up_read(&mapping->i_mmap_rwsem);
515 static inline void i_mmap_assert_locked(struct address_space *mapping)
517 lockdep_assert_held(&mapping->i_mmap_rwsem);
520 static inline void i_mmap_assert_write_locked(struct address_space *mapping)
522 lockdep_assert_held_write(&mapping->i_mmap_rwsem);
526 * Might pages of this file be mapped into userspace?
528 static inline int mapping_mapped(struct address_space *mapping)
530 return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root);
534 * Might pages of this file have been modified in userspace?
535 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap
536 * marks vma as VM_SHARED if it is shared, and the file was opened for
537 * writing i.e. vma may be mprotected writable even if now readonly.
539 * If i_mmap_writable is negative, no new writable mappings are allowed. You
540 * can only deny writable mappings, if none exists right now.
542 static inline int mapping_writably_mapped(struct address_space *mapping)
544 return atomic_read(&mapping->i_mmap_writable) > 0;
547 static inline int mapping_map_writable(struct address_space *mapping)
549 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
553 static inline void mapping_unmap_writable(struct address_space *mapping)
555 atomic_dec(&mapping->i_mmap_writable);
558 static inline int mapping_deny_writable(struct address_space *mapping)
560 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
564 static inline void mapping_allow_writable(struct address_space *mapping)
566 atomic_inc(&mapping->i_mmap_writable);
570 * Use sequence counter to get consistent i_size on 32-bit processors.
572 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
573 #include <linux/seqlock.h>
574 #define __NEED_I_SIZE_ORDERED
575 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
577 #define i_size_ordered_init(inode) do { } while (0)
581 #define ACL_NOT_CACHED ((void *)(-1))
583 * ACL_DONT_CACHE is for stacked filesystems, that rely on underlying fs to
584 * cache the ACL. This also means that ->get_acl() can be called in RCU mode
585 * with the LOOKUP_RCU flag.
587 #define ACL_DONT_CACHE ((void *)(-3))
589 static inline struct posix_acl *
590 uncached_acl_sentinel(struct task_struct *task)
592 return (void *)task + 1;
596 is_uncached_acl(struct posix_acl *acl)
598 return (long)acl & 1;
601 #define IOP_FASTPERM 0x0001
602 #define IOP_LOOKUP 0x0002
603 #define IOP_NOFOLLOW 0x0004
604 #define IOP_XATTR 0x0008
605 #define IOP_DEFAULT_READLINK 0x0010
607 struct fsnotify_mark_connector;
610 * Keep mostly read-only and often accessed (especially for
611 * the RCU path lookup and 'stat' data) fields at the beginning
612 * of the 'struct inode'
616 unsigned short i_opflags;
619 unsigned int i_flags;
621 #ifdef CONFIG_FS_POSIX_ACL
622 struct posix_acl *i_acl;
623 struct posix_acl *i_default_acl;
626 const struct inode_operations *i_op;
627 struct super_block *i_sb;
628 struct address_space *i_mapping;
630 #ifdef CONFIG_SECURITY
634 /* Stat data, not accessed from path walking */
637 * Filesystems may only read i_nlink directly. They shall use the
638 * following functions for modification:
640 * (set|clear|inc|drop)_nlink
641 * inode_(inc|dec)_link_count
644 const unsigned int i_nlink;
645 unsigned int __i_nlink;
649 struct timespec64 i_atime;
650 struct timespec64 i_mtime;
651 struct timespec64 i_ctime;
652 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
653 unsigned short i_bytes;
658 #ifdef __NEED_I_SIZE_ORDERED
659 seqcount_t i_size_seqcount;
663 unsigned long i_state;
664 struct rw_semaphore i_rwsem;
666 unsigned long dirtied_when; /* jiffies of first dirtying */
667 unsigned long dirtied_time_when;
669 struct hlist_node i_hash;
670 struct list_head i_io_list; /* backing dev IO list */
671 #ifdef CONFIG_CGROUP_WRITEBACK
672 struct bdi_writeback *i_wb; /* the associated cgroup wb */
674 /* foreign inode detection, see wbc_detach_inode() */
676 u16 i_wb_frn_avg_time;
677 u16 i_wb_frn_history;
679 struct list_head i_lru; /* inode LRU list */
680 struct list_head i_sb_list;
681 struct list_head i_wb_list; /* backing dev writeback list */
683 struct hlist_head i_dentry;
684 struct rcu_head i_rcu;
686 atomic64_t i_version;
687 atomic64_t i_sequence; /* see futex */
689 atomic_t i_dio_count;
690 atomic_t i_writecount;
691 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
692 atomic_t i_readcount; /* struct files open RO */
695 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
696 void (*free_inode)(struct inode *);
698 struct file_lock_context *i_flctx;
699 struct address_space i_data;
700 struct list_head i_devices;
702 struct pipe_inode_info *i_pipe;
710 #ifdef CONFIG_FSNOTIFY
711 __u32 i_fsnotify_mask; /* all events this inode cares about */
712 struct fsnotify_mark_connector __rcu *i_fsnotify_marks;
715 #ifdef CONFIG_FS_ENCRYPTION
716 struct fscrypt_info *i_crypt_info;
719 #ifdef CONFIG_FS_VERITY
720 struct fsverity_info *i_verity_info;
723 void *i_private; /* fs or device private pointer */
724 } __randomize_layout;
726 struct timespec64 timestamp_truncate(struct timespec64 t, struct inode *inode);
728 static inline unsigned int i_blocksize(const struct inode *node)
730 return (1 << node->i_blkbits);
733 static inline int inode_unhashed(struct inode *inode)
735 return hlist_unhashed(&inode->i_hash);
739 * __mark_inode_dirty expects inodes to be hashed. Since we don't
740 * want special inodes in the fileset inode space, we make them
741 * appear hashed, but do not put on any lists. hlist_del()
742 * will work fine and require no locking.
744 static inline void inode_fake_hash(struct inode *inode)
746 hlist_add_fake(&inode->i_hash);
750 * inode->i_mutex nesting subclasses for the lock validator:
752 * 0: the object of the current VFS operation
756 * 4: second non-directory
757 * 5: second parent (when locking independent directories in rename)
759 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
760 * non-directories at once.
762 * The locking order between these classes is
763 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
765 enum inode_i_mutex_lock_class
775 static inline void inode_lock(struct inode *inode)
777 down_write(&inode->i_rwsem);
780 static inline void inode_unlock(struct inode *inode)
782 up_write(&inode->i_rwsem);
785 static inline void inode_lock_shared(struct inode *inode)
787 down_read(&inode->i_rwsem);
790 static inline void inode_unlock_shared(struct inode *inode)
792 up_read(&inode->i_rwsem);
795 static inline int inode_trylock(struct inode *inode)
797 return down_write_trylock(&inode->i_rwsem);
800 static inline int inode_trylock_shared(struct inode *inode)
802 return down_read_trylock(&inode->i_rwsem);
805 static inline int inode_is_locked(struct inode *inode)
807 return rwsem_is_locked(&inode->i_rwsem);
810 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
812 down_write_nested(&inode->i_rwsem, subclass);
815 static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass)
817 down_read_nested(&inode->i_rwsem, subclass);
820 static inline void filemap_invalidate_lock(struct address_space *mapping)
822 down_write(&mapping->invalidate_lock);
825 static inline void filemap_invalidate_unlock(struct address_space *mapping)
827 up_write(&mapping->invalidate_lock);
830 static inline void filemap_invalidate_lock_shared(struct address_space *mapping)
832 down_read(&mapping->invalidate_lock);
835 static inline int filemap_invalidate_trylock_shared(
836 struct address_space *mapping)
838 return down_read_trylock(&mapping->invalidate_lock);
841 static inline void filemap_invalidate_unlock_shared(
842 struct address_space *mapping)
844 up_read(&mapping->invalidate_lock);
847 void lock_two_nondirectories(struct inode *, struct inode*);
848 void unlock_two_nondirectories(struct inode *, struct inode*);
850 void filemap_invalidate_lock_two(struct address_space *mapping1,
851 struct address_space *mapping2);
852 void filemap_invalidate_unlock_two(struct address_space *mapping1,
853 struct address_space *mapping2);
857 * NOTE: in a 32bit arch with a preemptable kernel and
858 * an UP compile the i_size_read/write must be atomic
859 * with respect to the local cpu (unlike with preempt disabled),
860 * but they don't need to be atomic with respect to other cpus like in
861 * true SMP (so they need either to either locally disable irq around
862 * the read or for example on x86 they can be still implemented as a
863 * cmpxchg8b without the need of the lock prefix). For SMP compiles
864 * and 64bit archs it makes no difference if preempt is enabled or not.
866 static inline loff_t i_size_read(const struct inode *inode)
868 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
873 seq = read_seqcount_begin(&inode->i_size_seqcount);
874 i_size = inode->i_size;
875 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
877 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
881 i_size = inode->i_size;
885 return inode->i_size;
890 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
891 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
892 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
894 static inline void i_size_write(struct inode *inode, loff_t i_size)
896 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
898 write_seqcount_begin(&inode->i_size_seqcount);
899 inode->i_size = i_size;
900 write_seqcount_end(&inode->i_size_seqcount);
902 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
904 inode->i_size = i_size;
907 inode->i_size = i_size;
911 static inline unsigned iminor(const struct inode *inode)
913 return MINOR(inode->i_rdev);
916 static inline unsigned imajor(const struct inode *inode)
918 return MAJOR(inode->i_rdev);
922 rwlock_t lock; /* protects pid, uid, euid fields */
923 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
924 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
925 kuid_t uid, euid; /* uid/euid of process setting the owner */
926 int signum; /* posix.1b rt signal to be delivered on IO */
930 * struct file_ra_state - Track a file's readahead state.
931 * @start: Where the most recent readahead started.
932 * @size: Number of pages read in the most recent readahead.
933 * @async_size: Numer of pages that were/are not needed immediately
934 * and so were/are genuinely "ahead". Start next readahead when
935 * the first of these pages is accessed.
936 * @ra_pages: Maximum size of a readahead request, copied from the bdi.
937 * @mmap_miss: How many mmap accesses missed in the page cache.
938 * @prev_pos: The last byte in the most recent read request.
940 * When this structure is passed to ->readahead(), the "most recent"
941 * readahead means the current readahead.
943 struct file_ra_state {
946 unsigned int async_size;
947 unsigned int ra_pages;
948 unsigned int mmap_miss;
953 * Check if @index falls in the readahead windows.
955 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
957 return (index >= ra->start &&
958 index < ra->start + ra->size);
963 struct llist_node fu_llist;
964 struct rcu_head fu_rcuhead;
967 struct inode *f_inode; /* cached value */
968 const struct file_operations *f_op;
971 * Protects f_ep, f_flags.
972 * Must not be taken from IRQ context.
975 atomic_long_t f_count;
976 unsigned int f_flags;
978 struct mutex f_pos_lock;
980 struct fown_struct f_owner;
981 const struct cred *f_cred;
982 struct file_ra_state f_ra;
985 #ifdef CONFIG_SECURITY
988 /* needed for tty driver, and maybe others */
992 /* Used by fs/eventpoll.c to link all the hooks to this file */
993 struct hlist_head *f_ep;
994 #endif /* #ifdef CONFIG_EPOLL */
995 struct address_space *f_mapping;
997 errseq_t f_sb_err; /* for syncfs */
999 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
1001 struct file_handle {
1004 /* file identifier */
1005 unsigned char f_handle[];
1008 static inline struct file *get_file(struct file *f)
1010 atomic_long_inc(&f->f_count);
1013 #define get_file_rcu_many(x, cnt) \
1014 atomic_long_add_unless(&(x)->f_count, (cnt), 0)
1015 #define get_file_rcu(x) get_file_rcu_many((x), 1)
1016 #define file_count(x) atomic_long_read(&(x)->f_count)
1018 #define MAX_NON_LFS ((1UL<<31) - 1)
1020 /* Page cache limit. The filesystems should put that into their s_maxbytes
1021 limits, otherwise bad things can happen in VM. */
1022 #if BITS_PER_LONG==32
1023 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
1024 #elif BITS_PER_LONG==64
1025 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
1030 #define FL_DELEG 4 /* NFSv4 delegation */
1031 #define FL_ACCESS 8 /* not trying to lock, just looking */
1032 #define FL_EXISTS 16 /* when unlocking, test for existence */
1033 #define FL_LEASE 32 /* lease held on this file */
1034 #define FL_CLOSE 64 /* unlock on close */
1035 #define FL_SLEEP 128 /* A blocking lock */
1036 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
1037 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
1038 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
1039 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
1040 #define FL_RECLAIM 4096 /* reclaiming from a reboot server */
1042 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
1045 * Special return value from posix_lock_file() and vfs_lock_file() for
1046 * asynchronous locking.
1048 #define FILE_LOCK_DEFERRED 1
1050 /* legacy typedef, should eventually be removed */
1051 typedef void *fl_owner_t;
1055 struct file_lock_operations {
1056 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1057 void (*fl_release_private)(struct file_lock *);
1060 struct lock_manager_operations {
1061 fl_owner_t (*lm_get_owner)(fl_owner_t);
1062 void (*lm_put_owner)(fl_owner_t);
1063 void (*lm_notify)(struct file_lock *); /* unblock callback */
1064 int (*lm_grant)(struct file_lock *, int);
1065 bool (*lm_break)(struct file_lock *);
1066 int (*lm_change)(struct file_lock *, int, struct list_head *);
1067 void (*lm_setup)(struct file_lock *, void **);
1068 bool (*lm_breaker_owns_lease)(struct file_lock *);
1071 struct lock_manager {
1072 struct list_head list;
1074 * NFSv4 and up also want opens blocked during the grace period;
1081 void locks_start_grace(struct net *, struct lock_manager *);
1082 void locks_end_grace(struct lock_manager *);
1083 bool locks_in_grace(struct net *);
1084 bool opens_in_grace(struct net *);
1086 /* that will die - we need it for nfs_lock_info */
1087 #include <linux/nfs_fs_i.h>
1090 * struct file_lock represents a generic "file lock". It's used to represent
1091 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1092 * note that the same struct is used to represent both a request for a lock and
1093 * the lock itself, but the same object is never used for both.
1095 * FIXME: should we create a separate "struct lock_request" to help distinguish
1098 * The varous i_flctx lists are ordered by:
1101 * 2) lock range start
1104 * Obviously, the last two criteria only matter for POSIX locks.
1107 struct file_lock *fl_blocker; /* The lock, that is blocking us */
1108 struct list_head fl_list; /* link into file_lock_context */
1109 struct hlist_node fl_link; /* node in global lists */
1110 struct list_head fl_blocked_requests; /* list of requests with
1111 * ->fl_blocker pointing here
1113 struct list_head fl_blocked_member; /* node in
1114 * ->fl_blocker->fl_blocked_requests
1116 fl_owner_t fl_owner;
1117 unsigned int fl_flags;
1118 unsigned char fl_type;
1119 unsigned int fl_pid;
1120 int fl_link_cpu; /* what cpu's list is this on? */
1121 wait_queue_head_t fl_wait;
1122 struct file *fl_file;
1126 struct fasync_struct * fl_fasync; /* for lease break notifications */
1127 /* for lease breaks: */
1128 unsigned long fl_break_time;
1129 unsigned long fl_downgrade_time;
1131 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1132 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1134 struct nfs_lock_info nfs_fl;
1135 struct nfs4_lock_info nfs4_fl;
1137 struct list_head link; /* link in AFS vnode's pending_locks list */
1138 int state; /* state of grant or error if -ve */
1139 unsigned int debug_id;
1142 } __randomize_layout;
1144 struct file_lock_context {
1145 spinlock_t flc_lock;
1146 struct list_head flc_flock;
1147 struct list_head flc_posix;
1148 struct list_head flc_lease;
1151 /* The following constant reflects the upper bound of the file/locking space */
1153 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1154 #define OFFSET_MAX INT_LIMIT(loff_t)
1155 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1158 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1160 #define locks_inode(f) file_inode(f)
1162 #ifdef CONFIG_FILE_LOCKING
1163 extern int fcntl_getlk(struct file *, unsigned int, struct flock *);
1164 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1167 #if BITS_PER_LONG == 32
1168 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *);
1169 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1173 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1174 extern int fcntl_getlease(struct file *filp);
1177 void locks_free_lock_context(struct inode *inode);
1178 void locks_free_lock(struct file_lock *fl);
1179 extern void locks_init_lock(struct file_lock *);
1180 extern struct file_lock * locks_alloc_lock(void);
1181 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1182 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1183 extern void locks_remove_posix(struct file *, fl_owner_t);
1184 extern void locks_remove_file(struct file *);
1185 extern void locks_release_private(struct file_lock *);
1186 extern void posix_test_lock(struct file *, struct file_lock *);
1187 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1188 extern int locks_delete_block(struct file_lock *);
1189 extern int vfs_test_lock(struct file *, struct file_lock *);
1190 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1191 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1192 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1193 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1194 extern void lease_get_mtime(struct inode *, struct timespec64 *time);
1195 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1196 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1197 extern int lease_modify(struct file_lock *, int, struct list_head *);
1199 struct notifier_block;
1200 extern int lease_register_notifier(struct notifier_block *);
1201 extern void lease_unregister_notifier(struct notifier_block *);
1203 struct files_struct;
1204 extern void show_fd_locks(struct seq_file *f,
1205 struct file *filp, struct files_struct *files);
1206 #else /* !CONFIG_FILE_LOCKING */
1207 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1208 struct flock __user *user)
1213 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1214 unsigned int cmd, struct flock __user *user)
1219 #if BITS_PER_LONG == 32
1220 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1221 struct flock64 *user)
1226 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1227 unsigned int cmd, struct flock64 *user)
1232 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1237 static inline int fcntl_getlease(struct file *filp)
1243 locks_free_lock_context(struct inode *inode)
1247 static inline void locks_init_lock(struct file_lock *fl)
1252 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1257 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1262 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1267 static inline void locks_remove_file(struct file *filp)
1272 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1277 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1278 struct file_lock *conflock)
1283 static inline int locks_delete_block(struct file_lock *waiter)
1288 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1293 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1294 struct file_lock *fl, struct file_lock *conf)
1299 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1304 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1309 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1314 static inline void lease_get_mtime(struct inode *inode,
1315 struct timespec64 *time)
1320 static inline int generic_setlease(struct file *filp, long arg,
1321 struct file_lock **flp, void **priv)
1326 static inline int vfs_setlease(struct file *filp, long arg,
1327 struct file_lock **lease, void **priv)
1332 static inline int lease_modify(struct file_lock *fl, int arg,
1333 struct list_head *dispose)
1338 struct files_struct;
1339 static inline void show_fd_locks(struct seq_file *f,
1340 struct file *filp, struct files_struct *files) {}
1341 #endif /* !CONFIG_FILE_LOCKING */
1343 static inline struct inode *file_inode(const struct file *f)
1348 static inline struct dentry *file_dentry(const struct file *file)
1350 return d_real(file->f_path.dentry, file_inode(file));
1353 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1355 return locks_lock_inode_wait(locks_inode(filp), fl);
1358 struct fasync_struct {
1362 struct fasync_struct *fa_next; /* singly linked list */
1363 struct file *fa_file;
1364 struct rcu_head fa_rcu;
1367 #define FASYNC_MAGIC 0x4601
1369 /* SMP safe fasync helpers: */
1370 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1371 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1372 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1373 extern struct fasync_struct *fasync_alloc(void);
1374 extern void fasync_free(struct fasync_struct *);
1376 /* can be called from interrupts */
1377 extern void kill_fasync(struct fasync_struct **, int, int);
1379 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1380 extern int f_setown(struct file *filp, unsigned long arg, int force);
1381 extern void f_delown(struct file *filp);
1382 extern pid_t f_getown(struct file *filp);
1383 extern int send_sigurg(struct fown_struct *fown);
1386 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1387 * represented in both.
1389 #define SB_RDONLY 1 /* Mount read-only */
1390 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1391 #define SB_NODEV 4 /* Disallow access to device special files */
1392 #define SB_NOEXEC 8 /* Disallow program execution */
1393 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1394 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1395 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1396 #define SB_NOATIME 1024 /* Do not update access times. */
1397 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1398 #define SB_SILENT 32768
1399 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1400 #define SB_INLINECRYPT (1<<17) /* Use blk-crypto for encrypted files */
1401 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1402 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1403 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1405 /* These sb flags are internal to the kernel */
1406 #define SB_SUBMOUNT (1<<26)
1407 #define SB_FORCE (1<<27)
1408 #define SB_NOSEC (1<<28)
1409 #define SB_BORN (1<<29)
1410 #define SB_ACTIVE (1<<30)
1411 #define SB_NOUSER (1<<31)
1413 /* These flags relate to encoding and casefolding */
1414 #define SB_ENC_STRICT_MODE_FL (1 << 0)
1416 #define sb_has_strict_encoding(sb) \
1417 (sb->s_encoding_flags & SB_ENC_STRICT_MODE_FL)
1423 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1424 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1425 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1426 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1427 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1430 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1431 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1432 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1433 #define SB_I_STABLE_WRITES 0x00000008 /* don't modify blks until WB is done */
1435 /* sb->s_iflags to limit user namespace mounts */
1436 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1437 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1438 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1440 #define SB_I_SKIP_SYNC 0x00000100 /* Skip superblock at global sync */
1441 #define SB_I_PERSB_BDI 0x00000200 /* has a per-sb bdi */
1442 #define SB_I_TS_EXPIRY_WARNED 0x00000400 /* warned about timestamp range expiry */
1444 /* Possible states of 'frozen' field */
1446 SB_UNFROZEN = 0, /* FS is unfrozen */
1447 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1448 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1449 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1450 * internal threads if needed) */
1451 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1454 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1457 int frozen; /* Is sb frozen? */
1458 wait_queue_head_t wait_unfrozen; /* wait for thaw */
1459 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1462 struct super_block {
1463 struct list_head s_list; /* Keep this first */
1464 dev_t s_dev; /* search index; _not_ kdev_t */
1465 unsigned char s_blocksize_bits;
1466 unsigned long s_blocksize;
1467 loff_t s_maxbytes; /* Max file size */
1468 struct file_system_type *s_type;
1469 const struct super_operations *s_op;
1470 const struct dquot_operations *dq_op;
1471 const struct quotactl_ops *s_qcop;
1472 const struct export_operations *s_export_op;
1473 unsigned long s_flags;
1474 unsigned long s_iflags; /* internal SB_I_* flags */
1475 unsigned long s_magic;
1476 struct dentry *s_root;
1477 struct rw_semaphore s_umount;
1480 #ifdef CONFIG_SECURITY
1483 const struct xattr_handler **s_xattr;
1484 #ifdef CONFIG_FS_ENCRYPTION
1485 const struct fscrypt_operations *s_cop;
1486 struct key *s_master_keys; /* master crypto keys in use */
1488 #ifdef CONFIG_FS_VERITY
1489 const struct fsverity_operations *s_vop;
1491 #if IS_ENABLED(CONFIG_UNICODE)
1492 struct unicode_map *s_encoding;
1493 __u16 s_encoding_flags;
1495 struct hlist_bl_head s_roots; /* alternate root dentries for NFS */
1496 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1497 struct block_device *s_bdev;
1498 struct backing_dev_info *s_bdi;
1499 struct mtd_info *s_mtd;
1500 struct hlist_node s_instances;
1501 unsigned int s_quota_types; /* Bitmask of supported quota types */
1502 struct quota_info s_dquot; /* Diskquota specific options */
1504 struct sb_writers s_writers;
1507 * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and
1508 * s_fsnotify_marks together for cache efficiency. They are frequently
1509 * accessed and rarely modified.
1511 void *s_fs_info; /* Filesystem private info */
1513 /* Granularity of c/m/atime in ns (cannot be worse than a second) */
1515 /* Time limits for c/m/atime in seconds */
1516 time64_t s_time_min;
1517 time64_t s_time_max;
1518 #ifdef CONFIG_FSNOTIFY
1519 __u32 s_fsnotify_mask;
1520 struct fsnotify_mark_connector __rcu *s_fsnotify_marks;
1523 char s_id[32]; /* Informational name */
1524 uuid_t s_uuid; /* UUID */
1526 unsigned int s_max_links;
1530 * The next field is for VFS *only*. No filesystems have any business
1531 * even looking at it. You had been warned.
1533 struct mutex s_vfs_rename_mutex; /* Kludge */
1536 * Filesystem subtype. If non-empty the filesystem type field
1537 * in /proc/mounts will be "type.subtype"
1539 const char *s_subtype;
1541 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1543 struct shrinker s_shrink; /* per-sb shrinker handle */
1545 /* Number of inodes with nlink == 0 but still referenced */
1546 atomic_long_t s_remove_count;
1549 * Number of inode/mount/sb objects that are being watched, note that
1550 * inodes objects are currently double-accounted.
1552 atomic_long_t s_fsnotify_connectors;
1554 /* Being remounted read-only */
1555 int s_readonly_remount;
1557 /* per-sb errseq_t for reporting writeback errors via syncfs */
1560 /* AIO completions deferred from interrupt context */
1561 struct workqueue_struct *s_dio_done_wq;
1562 struct hlist_head s_pins;
1565 * Owning user namespace and default context in which to
1566 * interpret filesystem uids, gids, quotas, device nodes,
1567 * xattrs and security labels.
1569 struct user_namespace *s_user_ns;
1572 * The list_lru structure is essentially just a pointer to a table
1573 * of per-node lru lists, each of which has its own spinlock.
1574 * There is no need to put them into separate cachelines.
1576 struct list_lru s_dentry_lru;
1577 struct list_lru s_inode_lru;
1578 struct rcu_head rcu;
1579 struct work_struct destroy_work;
1581 struct mutex s_sync_lock; /* sync serialisation lock */
1584 * Indicates how deep in a filesystem stack this SB is
1588 /* s_inode_list_lock protects s_inodes */
1589 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1590 struct list_head s_inodes; /* all inodes */
1592 spinlock_t s_inode_wblist_lock;
1593 struct list_head s_inodes_wb; /* writeback inodes */
1594 } __randomize_layout;
1596 static inline struct user_namespace *i_user_ns(const struct inode *inode)
1598 return inode->i_sb->s_user_ns;
1601 /* Helper functions so that in most cases filesystems will
1602 * not need to deal directly with kuid_t and kgid_t and can
1603 * instead deal with the raw numeric values that are stored
1604 * in the filesystem.
1606 static inline uid_t i_uid_read(const struct inode *inode)
1608 return from_kuid(i_user_ns(inode), inode->i_uid);
1611 static inline gid_t i_gid_read(const struct inode *inode)
1613 return from_kgid(i_user_ns(inode), inode->i_gid);
1616 static inline void i_uid_write(struct inode *inode, uid_t uid)
1618 inode->i_uid = make_kuid(i_user_ns(inode), uid);
1621 static inline void i_gid_write(struct inode *inode, gid_t gid)
1623 inode->i_gid = make_kgid(i_user_ns(inode), gid);
1627 * i_uid_into_mnt - map an inode's i_uid down into a mnt_userns
1628 * @mnt_userns: user namespace of the mount the inode was found from
1629 * @inode: inode to map
1631 * Return: the inode's i_uid mapped down according to @mnt_userns.
1632 * If the inode's i_uid has no mapping INVALID_UID is returned.
1634 static inline kuid_t i_uid_into_mnt(struct user_namespace *mnt_userns,
1635 const struct inode *inode)
1637 return mapped_kuid_fs(mnt_userns, i_user_ns(inode), inode->i_uid);
1641 * i_gid_into_mnt - map an inode's i_gid down into a mnt_userns
1642 * @mnt_userns: user namespace of the mount the inode was found from
1643 * @inode: inode to map
1645 * Return: the inode's i_gid mapped down according to @mnt_userns.
1646 * If the inode's i_gid has no mapping INVALID_GID is returned.
1648 static inline kgid_t i_gid_into_mnt(struct user_namespace *mnt_userns,
1649 const struct inode *inode)
1651 return mapped_kgid_fs(mnt_userns, i_user_ns(inode), inode->i_gid);
1655 * inode_fsuid_set - initialize inode's i_uid field with callers fsuid
1656 * @inode: inode to initialize
1657 * @mnt_userns: user namespace of the mount the inode was found from
1659 * Initialize the i_uid field of @inode. If the inode was found/created via
1660 * an idmapped mount map the caller's fsuid according to @mnt_users.
1662 static inline void inode_fsuid_set(struct inode *inode,
1663 struct user_namespace *mnt_userns)
1665 inode->i_uid = mapped_fsuid(mnt_userns, i_user_ns(inode));
1669 * inode_fsgid_set - initialize inode's i_gid field with callers fsgid
1670 * @inode: inode to initialize
1671 * @mnt_userns: user namespace of the mount the inode was found from
1673 * Initialize the i_gid field of @inode. If the inode was found/created via
1674 * an idmapped mount map the caller's fsgid according to @mnt_users.
1676 static inline void inode_fsgid_set(struct inode *inode,
1677 struct user_namespace *mnt_userns)
1679 inode->i_gid = mapped_fsgid(mnt_userns, i_user_ns(inode));
1683 * fsuidgid_has_mapping() - check whether caller's fsuid/fsgid is mapped
1684 * @sb: the superblock we want a mapping in
1685 * @mnt_userns: user namespace of the relevant mount
1687 * Check whether the caller's fsuid and fsgid have a valid mapping in the
1688 * s_user_ns of the superblock @sb. If the caller is on an idmapped mount map
1689 * the caller's fsuid and fsgid according to the @mnt_userns first.
1691 * Return: true if fsuid and fsgid is mapped, false if not.
1693 static inline bool fsuidgid_has_mapping(struct super_block *sb,
1694 struct user_namespace *mnt_userns)
1696 struct user_namespace *fs_userns = sb->s_user_ns;
1700 kuid = mapped_fsuid(mnt_userns, fs_userns);
1701 if (!uid_valid(kuid))
1703 kgid = mapped_fsgid(mnt_userns, fs_userns);
1704 if (!gid_valid(kgid))
1706 return kuid_has_mapping(fs_userns, kuid) &&
1707 kgid_has_mapping(fs_userns, kgid);
1710 extern struct timespec64 current_time(struct inode *inode);
1713 * Snapshotting support.
1717 * These are internal functions, please use sb_start_{write,pagefault,intwrite}
1720 static inline void __sb_end_write(struct super_block *sb, int level)
1722 percpu_up_read(sb->s_writers.rw_sem + level-1);
1725 static inline void __sb_start_write(struct super_block *sb, int level)
1727 percpu_down_read(sb->s_writers.rw_sem + level - 1);
1730 static inline bool __sb_start_write_trylock(struct super_block *sb, int level)
1732 return percpu_down_read_trylock(sb->s_writers.rw_sem + level - 1);
1735 #define __sb_writers_acquired(sb, lev) \
1736 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1737 #define __sb_writers_release(sb, lev) \
1738 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1741 * sb_end_write - drop write access to a superblock
1742 * @sb: the super we wrote to
1744 * Decrement number of writers to the filesystem. Wake up possible waiters
1745 * wanting to freeze the filesystem.
1747 static inline void sb_end_write(struct super_block *sb)
1749 __sb_end_write(sb, SB_FREEZE_WRITE);
1753 * sb_end_pagefault - drop write access to a superblock from a page fault
1754 * @sb: the super we wrote to
1756 * Decrement number of processes handling write page fault to the filesystem.
1757 * Wake up possible waiters wanting to freeze the filesystem.
1759 static inline void sb_end_pagefault(struct super_block *sb)
1761 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1765 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1766 * @sb: the super we wrote to
1768 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1769 * waiters wanting to freeze the filesystem.
1771 static inline void sb_end_intwrite(struct super_block *sb)
1773 __sb_end_write(sb, SB_FREEZE_FS);
1777 * sb_start_write - get write access to a superblock
1778 * @sb: the super we write to
1780 * When a process wants to write data or metadata to a file system (i.e. dirty
1781 * a page or an inode), it should embed the operation in a sb_start_write() -
1782 * sb_end_write() pair to get exclusion against file system freezing. This
1783 * function increments number of writers preventing freezing. If the file
1784 * system is already frozen, the function waits until the file system is
1787 * Since freeze protection behaves as a lock, users have to preserve
1788 * ordering of freeze protection and other filesystem locks. Generally,
1789 * freeze protection should be the outermost lock. In particular, we have:
1792 * -> i_mutex (write path, truncate, directory ops, ...)
1793 * -> s_umount (freeze_super, thaw_super)
1795 static inline void sb_start_write(struct super_block *sb)
1797 __sb_start_write(sb, SB_FREEZE_WRITE);
1800 static inline bool sb_start_write_trylock(struct super_block *sb)
1802 return __sb_start_write_trylock(sb, SB_FREEZE_WRITE);
1806 * sb_start_pagefault - get write access to a superblock from a page fault
1807 * @sb: the super we write to
1809 * When a process starts handling write page fault, it should embed the
1810 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1811 * exclusion against file system freezing. This is needed since the page fault
1812 * is going to dirty a page. This function increments number of running page
1813 * faults preventing freezing. If the file system is already frozen, the
1814 * function waits until the file system is thawed.
1816 * Since page fault freeze protection behaves as a lock, users have to preserve
1817 * ordering of freeze protection and other filesystem locks. It is advised to
1818 * put sb_start_pagefault() close to mmap_lock in lock ordering. Page fault
1819 * handling code implies lock dependency:
1822 * -> sb_start_pagefault
1824 static inline void sb_start_pagefault(struct super_block *sb)
1826 __sb_start_write(sb, SB_FREEZE_PAGEFAULT);
1830 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1831 * @sb: the super we write to
1833 * This is the third level of protection against filesystem freezing. It is
1834 * free for use by a filesystem. The only requirement is that it must rank
1835 * below sb_start_pagefault.
1837 * For example filesystem can call sb_start_intwrite() when starting a
1838 * transaction which somewhat eases handling of freezing for internal sources
1839 * of filesystem changes (internal fs threads, discarding preallocation on file
1842 static inline void sb_start_intwrite(struct super_block *sb)
1844 __sb_start_write(sb, SB_FREEZE_FS);
1847 static inline bool sb_start_intwrite_trylock(struct super_block *sb)
1849 return __sb_start_write_trylock(sb, SB_FREEZE_FS);
1852 bool inode_owner_or_capable(struct user_namespace *mnt_userns,
1853 const struct inode *inode);
1856 * VFS helper functions..
1858 int vfs_create(struct user_namespace *, struct inode *,
1859 struct dentry *, umode_t, bool);
1860 int vfs_mkdir(struct user_namespace *, struct inode *,
1861 struct dentry *, umode_t);
1862 int vfs_mknod(struct user_namespace *, struct inode *, struct dentry *,
1864 int vfs_symlink(struct user_namespace *, struct inode *,
1865 struct dentry *, const char *);
1866 int vfs_link(struct dentry *, struct user_namespace *, struct inode *,
1867 struct dentry *, struct inode **);
1868 int vfs_rmdir(struct user_namespace *, struct inode *, struct dentry *);
1869 int vfs_unlink(struct user_namespace *, struct inode *, struct dentry *,
1873 * struct renamedata - contains all information required for renaming
1874 * @old_mnt_userns: old user namespace of the mount the inode was found from
1875 * @old_dir: parent of source
1876 * @old_dentry: source
1877 * @new_mnt_userns: new user namespace of the mount the inode was found from
1878 * @new_dir: parent of destination
1879 * @new_dentry: destination
1880 * @delegated_inode: returns an inode needing a delegation break
1881 * @flags: rename flags
1884 struct user_namespace *old_mnt_userns;
1885 struct inode *old_dir;
1886 struct dentry *old_dentry;
1887 struct user_namespace *new_mnt_userns;
1888 struct inode *new_dir;
1889 struct dentry *new_dentry;
1890 struct inode **delegated_inode;
1892 } __randomize_layout;
1894 int vfs_rename(struct renamedata *);
1896 static inline int vfs_whiteout(struct user_namespace *mnt_userns,
1897 struct inode *dir, struct dentry *dentry)
1899 return vfs_mknod(mnt_userns, dir, dentry, S_IFCHR | WHITEOUT_MODE,
1903 struct dentry *vfs_tmpfile(struct user_namespace *mnt_userns,
1904 struct dentry *dentry, umode_t mode, int open_flag);
1906 int vfs_mkobj(struct dentry *, umode_t,
1907 int (*f)(struct dentry *, umode_t, void *),
1910 int vfs_fchown(struct file *file, uid_t user, gid_t group);
1911 int vfs_fchmod(struct file *file, umode_t mode);
1912 int vfs_utimes(const struct path *path, struct timespec64 *times);
1914 extern long vfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
1916 #ifdef CONFIG_COMPAT
1917 extern long compat_ptr_ioctl(struct file *file, unsigned int cmd,
1920 #define compat_ptr_ioctl NULL
1924 * VFS file helper functions.
1926 void inode_init_owner(struct user_namespace *mnt_userns, struct inode *inode,
1927 const struct inode *dir, umode_t mode);
1928 extern bool may_open_dev(const struct path *path);
1931 * This is the "filldir" function type, used by readdir() to let
1932 * the kernel specify what kind of dirent layout it wants to have.
1933 * This allows the kernel to read directories into kernel space or
1934 * to have different dirent layouts depending on the binary type.
1937 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1940 struct dir_context {
1946 * These flags let !MMU mmap() govern direct device mapping vs immediate
1947 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1949 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1950 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1951 * NOMMU_MAP_READ: Can be mapped for reading
1952 * NOMMU_MAP_WRITE: Can be mapped for writing
1953 * NOMMU_MAP_EXEC: Can be mapped for execution
1955 #define NOMMU_MAP_COPY 0x00000001
1956 #define NOMMU_MAP_DIRECT 0x00000008
1957 #define NOMMU_MAP_READ VM_MAYREAD
1958 #define NOMMU_MAP_WRITE VM_MAYWRITE
1959 #define NOMMU_MAP_EXEC VM_MAYEXEC
1961 #define NOMMU_VMFLAGS \
1962 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1965 * These flags control the behavior of the remap_file_range function pointer.
1966 * If it is called with len == 0 that means "remap to end of source file".
1967 * See Documentation/filesystems/vfs.rst for more details about this call.
1969 * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate)
1970 * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request
1972 #define REMAP_FILE_DEDUP (1 << 0)
1973 #define REMAP_FILE_CAN_SHORTEN (1 << 1)
1976 * These flags signal that the caller is ok with altering various aspects of
1977 * the behavior of the remap operation. The changes must be made by the
1978 * implementation; the vfs remap helper functions can take advantage of them.
1979 * Flags in this category exist to preserve the quirky behavior of the hoisted
1980 * btrfs clone/dedupe ioctls.
1982 #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN)
1986 struct file_operations {
1987 struct module *owner;
1988 loff_t (*llseek) (struct file *, loff_t, int);
1989 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1990 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1991 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1992 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1993 int (*iopoll)(struct kiocb *kiocb, struct io_comp_batch *,
1994 unsigned int flags);
1995 int (*iterate) (struct file *, struct dir_context *);
1996 int (*iterate_shared) (struct file *, struct dir_context *);
1997 __poll_t (*poll) (struct file *, struct poll_table_struct *);
1998 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1999 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
2000 int (*mmap) (struct file *, struct vm_area_struct *);
2001 unsigned long mmap_supported_flags;
2002 int (*open) (struct inode *, struct file *);
2003 int (*flush) (struct file *, fl_owner_t id);
2004 int (*release) (struct inode *, struct file *);
2005 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
2006 int (*fasync) (int, struct file *, int);
2007 int (*lock) (struct file *, int, struct file_lock *);
2008 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
2009 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
2010 int (*check_flags)(int);
2011 int (*flock) (struct file *, int, struct file_lock *);
2012 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
2013 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
2014 int (*setlease)(struct file *, long, struct file_lock **, void **);
2015 long (*fallocate)(struct file *file, int mode, loff_t offset,
2017 void (*show_fdinfo)(struct seq_file *m, struct file *f);
2019 unsigned (*mmap_capabilities)(struct file *);
2021 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
2022 loff_t, size_t, unsigned int);
2023 loff_t (*remap_file_range)(struct file *file_in, loff_t pos_in,
2024 struct file *file_out, loff_t pos_out,
2025 loff_t len, unsigned int remap_flags);
2026 int (*fadvise)(struct file *, loff_t, loff_t, int);
2027 } __randomize_layout;
2029 struct inode_operations {
2030 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
2031 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
2032 int (*permission) (struct user_namespace *, struct inode *, int);
2033 struct posix_acl * (*get_acl)(struct inode *, int, bool);
2035 int (*readlink) (struct dentry *, char __user *,int);
2037 int (*create) (struct user_namespace *, struct inode *,struct dentry *,
2039 int (*link) (struct dentry *,struct inode *,struct dentry *);
2040 int (*unlink) (struct inode *,struct dentry *);
2041 int (*symlink) (struct user_namespace *, struct inode *,struct dentry *,
2043 int (*mkdir) (struct user_namespace *, struct inode *,struct dentry *,
2045 int (*rmdir) (struct inode *,struct dentry *);
2046 int (*mknod) (struct user_namespace *, struct inode *,struct dentry *,
2048 int (*rename) (struct user_namespace *, struct inode *, struct dentry *,
2049 struct inode *, struct dentry *, unsigned int);
2050 int (*setattr) (struct user_namespace *, struct dentry *,
2052 int (*getattr) (struct user_namespace *, const struct path *,
2053 struct kstat *, u32, unsigned int);
2054 ssize_t (*listxattr) (struct dentry *, char *, size_t);
2055 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
2057 int (*update_time)(struct inode *, struct timespec64 *, int);
2058 int (*atomic_open)(struct inode *, struct dentry *,
2059 struct file *, unsigned open_flag,
2060 umode_t create_mode);
2061 int (*tmpfile) (struct user_namespace *, struct inode *,
2062 struct dentry *, umode_t);
2063 int (*set_acl)(struct user_namespace *, struct inode *,
2064 struct posix_acl *, int);
2065 int (*fileattr_set)(struct user_namespace *mnt_userns,
2066 struct dentry *dentry, struct fileattr *fa);
2067 int (*fileattr_get)(struct dentry *dentry, struct fileattr *fa);
2068 } ____cacheline_aligned;
2070 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
2071 struct iov_iter *iter)
2073 return file->f_op->read_iter(kio, iter);
2076 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
2077 struct iov_iter *iter)
2079 return file->f_op->write_iter(kio, iter);
2082 static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
2084 return file->f_op->mmap(file, vma);
2087 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
2088 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
2089 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
2090 loff_t, size_t, unsigned int);
2091 extern ssize_t generic_copy_file_range(struct file *file_in, loff_t pos_in,
2092 struct file *file_out, loff_t pos_out,
2093 size_t len, unsigned int flags);
2094 extern int generic_remap_file_range_prep(struct file *file_in, loff_t pos_in,
2095 struct file *file_out, loff_t pos_out,
2097 unsigned int remap_flags);
2098 extern loff_t do_clone_file_range(struct file *file_in, loff_t pos_in,
2099 struct file *file_out, loff_t pos_out,
2100 loff_t len, unsigned int remap_flags);
2101 extern loff_t vfs_clone_file_range(struct file *file_in, loff_t pos_in,
2102 struct file *file_out, loff_t pos_out,
2103 loff_t len, unsigned int remap_flags);
2104 extern int vfs_dedupe_file_range(struct file *file,
2105 struct file_dedupe_range *same);
2106 extern loff_t vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos,
2107 struct file *dst_file, loff_t dst_pos,
2108 loff_t len, unsigned int remap_flags);
2111 struct super_operations {
2112 struct inode *(*alloc_inode)(struct super_block *sb);
2113 void (*destroy_inode)(struct inode *);
2114 void (*free_inode)(struct inode *);
2116 void (*dirty_inode) (struct inode *, int flags);
2117 int (*write_inode) (struct inode *, struct writeback_control *wbc);
2118 int (*drop_inode) (struct inode *);
2119 void (*evict_inode) (struct inode *);
2120 void (*put_super) (struct super_block *);
2121 int (*sync_fs)(struct super_block *sb, int wait);
2122 int (*freeze_super) (struct super_block *);
2123 int (*freeze_fs) (struct super_block *);
2124 int (*thaw_super) (struct super_block *);
2125 int (*unfreeze_fs) (struct super_block *);
2126 int (*statfs) (struct dentry *, struct kstatfs *);
2127 int (*remount_fs) (struct super_block *, int *, char *);
2128 void (*umount_begin) (struct super_block *);
2130 int (*show_options)(struct seq_file *, struct dentry *);
2131 int (*show_devname)(struct seq_file *, struct dentry *);
2132 int (*show_path)(struct seq_file *, struct dentry *);
2133 int (*show_stats)(struct seq_file *, struct dentry *);
2135 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
2136 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
2137 struct dquot **(*get_dquots)(struct inode *);
2139 long (*nr_cached_objects)(struct super_block *,
2140 struct shrink_control *);
2141 long (*free_cached_objects)(struct super_block *,
2142 struct shrink_control *);
2146 * Inode flags - they have no relation to superblock flags now
2148 #define S_SYNC (1 << 0) /* Writes are synced at once */
2149 #define S_NOATIME (1 << 1) /* Do not update access times */
2150 #define S_APPEND (1 << 2) /* Append-only file */
2151 #define S_IMMUTABLE (1 << 3) /* Immutable file */
2152 #define S_DEAD (1 << 4) /* removed, but still open directory */
2153 #define S_NOQUOTA (1 << 5) /* Inode is not counted to quota */
2154 #define S_DIRSYNC (1 << 6) /* Directory modifications are synchronous */
2155 #define S_NOCMTIME (1 << 7) /* Do not update file c/mtime */
2156 #define S_SWAPFILE (1 << 8) /* Do not truncate: swapon got its bmaps */
2157 #define S_PRIVATE (1 << 9) /* Inode is fs-internal */
2158 #define S_IMA (1 << 10) /* Inode has an associated IMA struct */
2159 #define S_AUTOMOUNT (1 << 11) /* Automount/referral quasi-directory */
2160 #define S_NOSEC (1 << 12) /* no suid or xattr security attributes */
2161 #ifdef CONFIG_FS_DAX
2162 #define S_DAX (1 << 13) /* Direct Access, avoiding the page cache */
2164 #define S_DAX 0 /* Make all the DAX code disappear */
2166 #define S_ENCRYPTED (1 << 14) /* Encrypted file (using fs/crypto/) */
2167 #define S_CASEFOLD (1 << 15) /* Casefolded file */
2168 #define S_VERITY (1 << 16) /* Verity file (using fs/verity/) */
2169 #define S_KERNEL_FILE (1 << 17) /* File is in use by the kernel (eg. fs/cachefiles) */
2172 * Note that nosuid etc flags are inode-specific: setting some file-system
2173 * flags just means all the inodes inherit those flags by default. It might be
2174 * possible to override it selectively if you really wanted to with some
2175 * ioctl() that is not currently implemented.
2177 * Exception: SB_RDONLY is always applied to the entire file system.
2179 * Unfortunately, it is possible to change a filesystems flags with it mounted
2180 * with files in use. This means that all of the inodes will not have their
2181 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
2182 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
2184 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
2186 static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; }
2187 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
2188 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
2189 ((inode)->i_flags & S_SYNC))
2190 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
2191 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
2192 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
2193 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
2194 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
2196 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
2197 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
2198 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
2199 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
2201 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
2202 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
2203 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
2204 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
2205 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
2206 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
2207 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
2208 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
2209 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
2210 #define IS_CASEFOLDED(inode) ((inode)->i_flags & S_CASEFOLD)
2211 #define IS_VERITY(inode) ((inode)->i_flags & S_VERITY)
2213 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
2214 (inode)->i_rdev == WHITEOUT_DEV)
2216 static inline bool HAS_UNMAPPED_ID(struct user_namespace *mnt_userns,
2217 struct inode *inode)
2219 return !uid_valid(i_uid_into_mnt(mnt_userns, inode)) ||
2220 !gid_valid(i_gid_into_mnt(mnt_userns, inode));
2223 static inline int iocb_flags(struct file *file);
2225 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
2227 *kiocb = (struct kiocb) {
2229 .ki_flags = iocb_flags(filp),
2230 .ki_ioprio = get_current_ioprio(),
2234 static inline void kiocb_clone(struct kiocb *kiocb, struct kiocb *kiocb_src,
2237 *kiocb = (struct kiocb) {
2239 .ki_flags = kiocb_src->ki_flags,
2240 .ki_ioprio = kiocb_src->ki_ioprio,
2241 .ki_pos = kiocb_src->ki_pos,
2246 * Inode state bits. Protected by inode->i_lock
2248 * Four bits determine the dirty state of the inode: I_DIRTY_SYNC,
2249 * I_DIRTY_DATASYNC, I_DIRTY_PAGES, and I_DIRTY_TIME.
2251 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
2252 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
2253 * various stages of removing an inode.
2255 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2257 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
2258 * fdatasync() (unless I_DIRTY_DATASYNC is also set).
2259 * Timestamp updates are the usual cause.
2260 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
2261 * these changes separately from I_DIRTY_SYNC so that we
2262 * don't have to write inode on fdatasync() when only
2263 * e.g. the timestamps have changed.
2264 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
2265 * I_DIRTY_TIME The inode itself only has dirty timestamps, and the
2266 * lazytime mount option is enabled. We keep track of this
2267 * separately from I_DIRTY_SYNC in order to implement
2268 * lazytime. This gets cleared if I_DIRTY_INODE
2269 * (I_DIRTY_SYNC and/or I_DIRTY_DATASYNC) gets set. I.e.
2270 * either I_DIRTY_TIME *or* I_DIRTY_INODE can be set in
2271 * i_state, but not both. I_DIRTY_PAGES may still be set.
2272 * I_NEW Serves as both a mutex and completion notification.
2273 * New inodes set I_NEW. If two processes both create
2274 * the same inode, one of them will release its inode and
2275 * wait for I_NEW to be released before returning.
2276 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2277 * also cause waiting on I_NEW, without I_NEW actually
2278 * being set. find_inode() uses this to prevent returning
2279 * nearly-dead inodes.
2280 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2281 * is zero. I_FREEING must be set when I_WILL_FREE is
2283 * I_FREEING Set when inode is about to be freed but still has dirty
2284 * pages or buffers attached or the inode itself is still
2286 * I_CLEAR Added by clear_inode(). In this state the inode is
2287 * clean and can be destroyed. Inode keeps I_FREEING.
2289 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2290 * prohibited for many purposes. iget() must wait for
2291 * the inode to be completely released, then create it
2292 * anew. Other functions will just ignore such inodes,
2293 * if appropriate. I_NEW is used for waiting.
2295 * I_SYNC Writeback of inode is running. The bit is set during
2296 * data writeback, and cleared with a wakeup on the bit
2297 * address once it is done. The bit is also used to pin
2298 * the inode in memory for flusher thread.
2300 * I_REFERENCED Marks the inode as recently references on the LRU list.
2302 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2304 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2305 * synchronize competing switching instances and to tell
2306 * wb stat updates to grab the i_pages lock. See
2307 * inode_switch_wbs_work_fn() for details.
2309 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2310 * and work dirs among overlayfs mounts.
2312 * I_CREATING New object's inode in the middle of setting up.
2314 * I_DONTCACHE Evict inode as soon as it is not used anymore.
2316 * I_SYNC_QUEUED Inode is queued in b_io or b_more_io writeback lists.
2317 * Used to detect that mark_inode_dirty() should not move
2318 * inode between dirty lists.
2320 * I_PINNING_FSCACHE_WB Inode is pinning an fscache object for writeback.
2322 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2324 #define I_DIRTY_SYNC (1 << 0)
2325 #define I_DIRTY_DATASYNC (1 << 1)
2326 #define I_DIRTY_PAGES (1 << 2)
2328 #define I_NEW (1 << __I_NEW)
2329 #define I_WILL_FREE (1 << 4)
2330 #define I_FREEING (1 << 5)
2331 #define I_CLEAR (1 << 6)
2333 #define I_SYNC (1 << __I_SYNC)
2334 #define I_REFERENCED (1 << 8)
2335 #define __I_DIO_WAKEUP 9
2336 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2337 #define I_LINKABLE (1 << 10)
2338 #define I_DIRTY_TIME (1 << 11)
2339 #define I_WB_SWITCH (1 << 13)
2340 #define I_OVL_INUSE (1 << 14)
2341 #define I_CREATING (1 << 15)
2342 #define I_DONTCACHE (1 << 16)
2343 #define I_SYNC_QUEUED (1 << 17)
2344 #define I_PINNING_FSCACHE_WB (1 << 18)
2346 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2347 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2348 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2350 extern void __mark_inode_dirty(struct inode *, int);
2351 static inline void mark_inode_dirty(struct inode *inode)
2353 __mark_inode_dirty(inode, I_DIRTY);
2356 static inline void mark_inode_dirty_sync(struct inode *inode)
2358 __mark_inode_dirty(inode, I_DIRTY_SYNC);
2362 * Returns true if the given inode itself only has dirty timestamps (its pages
2363 * may still be dirty) and isn't currently being allocated or freed.
2364 * Filesystems should call this if when writing an inode when lazytime is
2365 * enabled, they want to opportunistically write the timestamps of other inodes
2366 * located very nearby on-disk, e.g. in the same inode block. This returns true
2367 * if the given inode is in need of such an opportunistic update. Requires
2368 * i_lock, or at least later re-checking under i_lock.
2370 static inline bool inode_is_dirtytime_only(struct inode *inode)
2372 return (inode->i_state & (I_DIRTY_TIME | I_NEW |
2373 I_FREEING | I_WILL_FREE)) == I_DIRTY_TIME;
2376 extern void inc_nlink(struct inode *inode);
2377 extern void drop_nlink(struct inode *inode);
2378 extern void clear_nlink(struct inode *inode);
2379 extern void set_nlink(struct inode *inode, unsigned int nlink);
2381 static inline void inode_inc_link_count(struct inode *inode)
2384 mark_inode_dirty(inode);
2387 static inline void inode_dec_link_count(struct inode *inode)
2390 mark_inode_dirty(inode);
2393 enum file_time_flags {
2400 extern bool atime_needs_update(const struct path *, struct inode *);
2401 extern void touch_atime(const struct path *);
2402 int inode_update_time(struct inode *inode, struct timespec64 *time, int flags);
2404 static inline void file_accessed(struct file *file)
2406 if (!(file->f_flags & O_NOATIME))
2407 touch_atime(&file->f_path);
2410 extern int file_modified(struct file *file);
2412 int sync_inode_metadata(struct inode *inode, int wait);
2414 struct file_system_type {
2417 #define FS_REQUIRES_DEV 1
2418 #define FS_BINARY_MOUNTDATA 2
2419 #define FS_HAS_SUBTYPE 4
2420 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2421 #define FS_DISALLOW_NOTIFY_PERM 16 /* Disable fanotify permission events */
2422 #define FS_ALLOW_IDMAP 32 /* FS has been updated to handle vfs idmappings. */
2423 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2424 int (*init_fs_context)(struct fs_context *);
2425 const struct fs_parameter_spec *parameters;
2426 struct dentry *(*mount) (struct file_system_type *, int,
2427 const char *, void *);
2428 void (*kill_sb) (struct super_block *);
2429 struct module *owner;
2430 struct file_system_type * next;
2431 struct hlist_head fs_supers;
2433 struct lock_class_key s_lock_key;
2434 struct lock_class_key s_umount_key;
2435 struct lock_class_key s_vfs_rename_key;
2436 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2438 struct lock_class_key i_lock_key;
2439 struct lock_class_key i_mutex_key;
2440 struct lock_class_key invalidate_lock_key;
2441 struct lock_class_key i_mutex_dir_key;
2444 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2446 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2447 int flags, const char *dev_name, void *data,
2448 int (*fill_super)(struct super_block *, void *, int));
2449 extern struct dentry *mount_single(struct file_system_type *fs_type,
2450 int flags, void *data,
2451 int (*fill_super)(struct super_block *, void *, int));
2452 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2453 int flags, void *data,
2454 int (*fill_super)(struct super_block *, void *, int));
2455 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2456 void generic_shutdown_super(struct super_block *sb);
2457 void kill_block_super(struct super_block *sb);
2458 void kill_anon_super(struct super_block *sb);
2459 void kill_litter_super(struct super_block *sb);
2460 void deactivate_super(struct super_block *sb);
2461 void deactivate_locked_super(struct super_block *sb);
2462 int set_anon_super(struct super_block *s, void *data);
2463 int set_anon_super_fc(struct super_block *s, struct fs_context *fc);
2464 int get_anon_bdev(dev_t *);
2465 void free_anon_bdev(dev_t);
2466 struct super_block *sget_fc(struct fs_context *fc,
2467 int (*test)(struct super_block *, struct fs_context *),
2468 int (*set)(struct super_block *, struct fs_context *));
2469 struct super_block *sget(struct file_system_type *type,
2470 int (*test)(struct super_block *,void *),
2471 int (*set)(struct super_block *,void *),
2472 int flags, void *data);
2474 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2475 #define fops_get(fops) \
2476 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2477 #define fops_put(fops) \
2478 do { if (fops) module_put((fops)->owner); } while(0)
2480 * This one is to be used *ONLY* from ->open() instances.
2481 * fops must be non-NULL, pinned down *and* module dependencies
2482 * should be sufficient to pin the caller down as well.
2484 #define replace_fops(f, fops) \
2486 struct file *__file = (f); \
2487 fops_put(__file->f_op); \
2488 BUG_ON(!(__file->f_op = (fops))); \
2491 extern int register_filesystem(struct file_system_type *);
2492 extern int unregister_filesystem(struct file_system_type *);
2493 extern struct vfsmount *kern_mount(struct file_system_type *);
2494 extern void kern_unmount(struct vfsmount *mnt);
2495 extern int may_umount_tree(struct vfsmount *);
2496 extern int may_umount(struct vfsmount *);
2497 extern long do_mount(const char *, const char __user *,
2498 const char *, unsigned long, void *);
2499 extern struct vfsmount *collect_mounts(const struct path *);
2500 extern void drop_collected_mounts(struct vfsmount *);
2501 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2503 extern int vfs_statfs(const struct path *, struct kstatfs *);
2504 extern int user_statfs(const char __user *, struct kstatfs *);
2505 extern int fd_statfs(int, struct kstatfs *);
2506 extern int freeze_super(struct super_block *super);
2507 extern int thaw_super(struct super_block *super);
2508 extern bool our_mnt(struct vfsmount *mnt);
2509 extern __printf(2, 3)
2510 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2511 extern int super_setup_bdi(struct super_block *sb);
2513 extern int current_umask(void);
2515 extern void ihold(struct inode * inode);
2516 extern void iput(struct inode *);
2517 extern int generic_update_time(struct inode *, struct timespec64 *, int);
2520 extern struct kobject *fs_kobj;
2522 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2524 #ifdef CONFIG_FILE_LOCKING
2525 static inline int break_lease(struct inode *inode, unsigned int mode)
2528 * Since this check is lockless, we must ensure that any refcounts
2529 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2530 * could end up racing with tasks trying to set a new lease on this
2534 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2535 return __break_lease(inode, mode, FL_LEASE);
2539 static inline int break_deleg(struct inode *inode, unsigned int mode)
2542 * Since this check is lockless, we must ensure that any refcounts
2543 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2544 * could end up racing with tasks trying to set a new lease on this
2548 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2549 return __break_lease(inode, mode, FL_DELEG);
2553 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2557 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2558 if (ret == -EWOULDBLOCK && delegated_inode) {
2559 *delegated_inode = inode;
2565 static inline int break_deleg_wait(struct inode **delegated_inode)
2569 ret = break_deleg(*delegated_inode, O_WRONLY);
2570 iput(*delegated_inode);
2571 *delegated_inode = NULL;
2575 static inline int break_layout(struct inode *inode, bool wait)
2578 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2579 return __break_lease(inode,
2580 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2585 #else /* !CONFIG_FILE_LOCKING */
2586 static inline int break_lease(struct inode *inode, unsigned int mode)
2591 static inline int break_deleg(struct inode *inode, unsigned int mode)
2596 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2601 static inline int break_deleg_wait(struct inode **delegated_inode)
2607 static inline int break_layout(struct inode *inode, bool wait)
2612 #endif /* CONFIG_FILE_LOCKING */
2617 const char *name; /* pointer to actual string */
2618 const __user char *uptr; /* original userland pointer */
2620 struct audit_names *aname;
2623 static_assert(offsetof(struct filename, iname) % sizeof(long) == 0);
2625 static inline struct user_namespace *file_mnt_user_ns(struct file *file)
2627 return mnt_user_ns(file->f_path.mnt);
2631 * is_idmapped_mnt - check whether a mount is mapped
2632 * @mnt: the mount to check
2634 * If @mnt has an idmapping attached different from the
2635 * filesystem's idmapping then @mnt is mapped.
2637 * Return: true if mount is mapped, false if not.
2639 static inline bool is_idmapped_mnt(const struct vfsmount *mnt)
2641 return mnt_user_ns(mnt) != mnt->mnt_sb->s_user_ns;
2644 extern long vfs_truncate(const struct path *, loff_t);
2645 int do_truncate(struct user_namespace *, struct dentry *, loff_t start,
2646 unsigned int time_attrs, struct file *filp);
2647 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2649 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2651 extern struct file *file_open_name(struct filename *, int, umode_t);
2652 extern struct file *filp_open(const char *, int, umode_t);
2653 extern struct file *file_open_root(const struct path *,
2654 const char *, int, umode_t);
2655 static inline struct file *file_open_root_mnt(struct vfsmount *mnt,
2656 const char *name, int flags, umode_t mode)
2658 return file_open_root(&(struct path){.mnt = mnt, .dentry = mnt->mnt_root},
2661 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2662 extern struct file * open_with_fake_path(const struct path *, int,
2663 struct inode*, const struct cred *);
2664 static inline struct file *file_clone_open(struct file *file)
2666 return dentry_open(&file->f_path, file->f_flags, file->f_cred);
2668 extern int filp_close(struct file *, fl_owner_t id);
2670 extern struct filename *getname_flags(const char __user *, int, int *);
2671 extern struct filename *getname_uflags(const char __user *, int);
2672 extern struct filename *getname(const char __user *);
2673 extern struct filename *getname_kernel(const char *);
2674 extern void putname(struct filename *name);
2676 extern int finish_open(struct file *file, struct dentry *dentry,
2677 int (*open)(struct inode *, struct file *));
2678 extern int finish_no_open(struct file *file, struct dentry *dentry);
2681 extern void __init vfs_caches_init_early(void);
2682 extern void __init vfs_caches_init(void);
2684 extern struct kmem_cache *names_cachep;
2686 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2687 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2689 extern struct super_block *blockdev_superblock;
2690 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2692 return IS_ENABLED(CONFIG_BLOCK) && sb == blockdev_superblock;
2695 void emergency_thaw_all(void);
2696 extern int sync_filesystem(struct super_block *);
2697 extern const struct file_operations def_blk_fops;
2698 extern const struct file_operations def_chr_fops;
2701 #define CHRDEV_MAJOR_MAX 512
2702 /* Marks the bottom of the first segment of free char majors */
2703 #define CHRDEV_MAJOR_DYN_END 234
2704 /* Marks the top and bottom of the second segment of free char majors */
2705 #define CHRDEV_MAJOR_DYN_EXT_START 511
2706 #define CHRDEV_MAJOR_DYN_EXT_END 384
2708 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2709 extern int register_chrdev_region(dev_t, unsigned, const char *);
2710 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2711 unsigned int count, const char *name,
2712 const struct file_operations *fops);
2713 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2714 unsigned int count, const char *name);
2715 extern void unregister_chrdev_region(dev_t, unsigned);
2716 extern void chrdev_show(struct seq_file *,off_t);
2718 static inline int register_chrdev(unsigned int major, const char *name,
2719 const struct file_operations *fops)
2721 return __register_chrdev(major, 0, 256, name, fops);
2724 static inline void unregister_chrdev(unsigned int major, const char *name)
2726 __unregister_chrdev(major, 0, 256, name);
2729 extern void init_special_inode(struct inode *, umode_t, dev_t);
2731 /* Invalid inode operations -- fs/bad_inode.c */
2732 extern void make_bad_inode(struct inode *);
2733 extern bool is_bad_inode(struct inode *);
2735 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2737 extern int __must_check file_check_and_advance_wb_err(struct file *file);
2738 extern int __must_check file_write_and_wait_range(struct file *file,
2739 loff_t start, loff_t end);
2741 static inline int file_write_and_wait(struct file *file)
2743 return file_write_and_wait_range(file, 0, LLONG_MAX);
2746 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2748 extern int vfs_fsync(struct file *file, int datasync);
2750 extern int sync_file_range(struct file *file, loff_t offset, loff_t nbytes,
2751 unsigned int flags);
2754 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2755 * to already be updated for the write, and will return either the amount
2756 * of bytes passed in, or an error if syncing the file failed.
2758 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2760 if (iocb->ki_flags & IOCB_DSYNC) {
2761 int ret = vfs_fsync_range(iocb->ki_filp,
2762 iocb->ki_pos - count, iocb->ki_pos - 1,
2763 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2771 extern void emergency_sync(void);
2772 extern void emergency_remount(void);
2775 extern int bmap(struct inode *inode, sector_t *block);
2777 static inline int bmap(struct inode *inode, sector_t *block)
2783 int notify_change(struct user_namespace *, struct dentry *,
2784 struct iattr *, struct inode **);
2785 int inode_permission(struct user_namespace *, struct inode *, int);
2786 int generic_permission(struct user_namespace *, struct inode *, int);
2787 static inline int file_permission(struct file *file, int mask)
2789 return inode_permission(file_mnt_user_ns(file),
2790 file_inode(file), mask);
2792 static inline int path_permission(const struct path *path, int mask)
2794 return inode_permission(mnt_user_ns(path->mnt),
2795 d_inode(path->dentry), mask);
2797 int __check_sticky(struct user_namespace *mnt_userns, struct inode *dir,
2798 struct inode *inode);
2800 static inline bool execute_ok(struct inode *inode)
2802 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2805 static inline bool inode_wrong_type(const struct inode *inode, umode_t mode)
2807 return (inode->i_mode ^ mode) & S_IFMT;
2810 static inline void file_start_write(struct file *file)
2812 if (!S_ISREG(file_inode(file)->i_mode))
2814 sb_start_write(file_inode(file)->i_sb);
2817 static inline bool file_start_write_trylock(struct file *file)
2819 if (!S_ISREG(file_inode(file)->i_mode))
2821 return sb_start_write_trylock(file_inode(file)->i_sb);
2824 static inline void file_end_write(struct file *file)
2826 if (!S_ISREG(file_inode(file)->i_mode))
2828 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2832 * This is used for regular files where some users -- especially the
2833 * currently executed binary in a process, previously handled via
2834 * VM_DENYWRITE -- cannot handle concurrent write (and maybe mmap
2835 * read-write shared) accesses.
2837 * get_write_access() gets write permission for a file.
2838 * put_write_access() releases this write permission.
2839 * deny_write_access() denies write access to a file.
2840 * allow_write_access() re-enables write access to a file.
2842 * The i_writecount field of an inode can have the following values:
2843 * 0: no write access, no denied write access
2844 * < 0: (-i_writecount) users that denied write access to the file.
2845 * > 0: (i_writecount) users that have write access to the file.
2847 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2848 * except for the cases where we don't hold i_writecount yet. Then we need to
2849 * use {get,deny}_write_access() - these functions check the sign and refuse
2850 * to do the change if sign is wrong.
2852 static inline int get_write_access(struct inode *inode)
2854 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2856 static inline int deny_write_access(struct file *file)
2858 struct inode *inode = file_inode(file);
2859 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2861 static inline void put_write_access(struct inode * inode)
2863 atomic_dec(&inode->i_writecount);
2865 static inline void allow_write_access(struct file *file)
2868 atomic_inc(&file_inode(file)->i_writecount);
2870 static inline bool inode_is_open_for_write(const struct inode *inode)
2872 return atomic_read(&inode->i_writecount) > 0;
2875 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
2876 static inline void i_readcount_dec(struct inode *inode)
2878 BUG_ON(!atomic_read(&inode->i_readcount));
2879 atomic_dec(&inode->i_readcount);
2881 static inline void i_readcount_inc(struct inode *inode)
2883 atomic_inc(&inode->i_readcount);
2886 static inline void i_readcount_dec(struct inode *inode)
2890 static inline void i_readcount_inc(struct inode *inode)
2895 extern int do_pipe_flags(int *, int);
2897 extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
2898 ssize_t __kernel_read(struct file *file, void *buf, size_t count, loff_t *pos);
2899 extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
2900 extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
2901 extern struct file * open_exec(const char *);
2903 /* fs/dcache.c -- generic fs support functions */
2904 extern bool is_subdir(struct dentry *, struct dentry *);
2905 extern bool path_is_under(const struct path *, const struct path *);
2907 extern char *file_path(struct file *, char *, int);
2909 #include <linux/err.h>
2911 /* needed for stackable file system support */
2912 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2914 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2916 extern int inode_init_always(struct super_block *, struct inode *);
2917 extern void inode_init_once(struct inode *);
2918 extern void address_space_init_once(struct address_space *mapping);
2919 extern struct inode * igrab(struct inode *);
2920 extern ino_t iunique(struct super_block *, ino_t);
2921 extern int inode_needs_sync(struct inode *inode);
2922 extern int generic_delete_inode(struct inode *inode);
2923 static inline int generic_drop_inode(struct inode *inode)
2925 return !inode->i_nlink || inode_unhashed(inode);
2927 extern void d_mark_dontcache(struct inode *inode);
2929 extern struct inode *ilookup5_nowait(struct super_block *sb,
2930 unsigned long hashval, int (*test)(struct inode *, void *),
2932 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2933 int (*test)(struct inode *, void *), void *data);
2934 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2936 extern struct inode *inode_insert5(struct inode *inode, unsigned long hashval,
2937 int (*test)(struct inode *, void *),
2938 int (*set)(struct inode *, void *),
2940 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2941 extern struct inode * iget_locked(struct super_block *, unsigned long);
2942 extern struct inode *find_inode_nowait(struct super_block *,
2944 int (*match)(struct inode *,
2945 unsigned long, void *),
2947 extern struct inode *find_inode_rcu(struct super_block *, unsigned long,
2948 int (*)(struct inode *, void *), void *);
2949 extern struct inode *find_inode_by_ino_rcu(struct super_block *, unsigned long);
2950 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2951 extern int insert_inode_locked(struct inode *);
2952 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2953 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2955 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2957 extern void unlock_new_inode(struct inode *);
2958 extern void discard_new_inode(struct inode *);
2959 extern unsigned int get_next_ino(void);
2960 extern void evict_inodes(struct super_block *sb);
2961 void dump_mapping(const struct address_space *);
2964 * Userspace may rely on the the inode number being non-zero. For example, glibc
2965 * simply ignores files with zero i_ino in unlink() and other places.
2967 * As an additional complication, if userspace was compiled with
2968 * _FILE_OFFSET_BITS=32 on a 64-bit kernel we'll only end up reading out the
2969 * lower 32 bits, so we need to check that those aren't zero explicitly. With
2970 * _FILE_OFFSET_BITS=64, this may cause some harmless false-negatives, but
2971 * better safe than sorry.
2973 static inline bool is_zero_ino(ino_t ino)
2975 return (u32)ino == 0;
2978 extern void __iget(struct inode * inode);
2979 extern void iget_failed(struct inode *);
2980 extern void clear_inode(struct inode *);
2981 extern void __destroy_inode(struct inode *);
2982 extern struct inode *new_inode_pseudo(struct super_block *sb);
2983 extern struct inode *new_inode(struct super_block *sb);
2984 extern void free_inode_nonrcu(struct inode *inode);
2985 extern int should_remove_suid(struct dentry *);
2986 extern int file_remove_privs(struct file *);
2989 * This must be used for allocating filesystems specific inodes to set
2990 * up the inode reclaim context correctly.
2992 static inline void *
2993 alloc_inode_sb(struct super_block *sb, struct kmem_cache *cache, gfp_t gfp)
2995 return kmem_cache_alloc_lru(cache, &sb->s_inode_lru, gfp);
2998 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2999 static inline void insert_inode_hash(struct inode *inode)
3001 __insert_inode_hash(inode, inode->i_ino);
3004 extern void __remove_inode_hash(struct inode *);
3005 static inline void remove_inode_hash(struct inode *inode)
3007 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
3008 __remove_inode_hash(inode);
3011 extern void inode_sb_list_add(struct inode *inode);
3012 extern void inode_add_lru(struct inode *inode);
3014 extern int sb_set_blocksize(struct super_block *, int);
3015 extern int sb_min_blocksize(struct super_block *, int);
3017 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
3018 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
3019 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
3020 int generic_write_checks_count(struct kiocb *iocb, loff_t *count);
3021 extern int generic_write_check_limits(struct file *file, loff_t pos,
3023 extern int generic_file_rw_checks(struct file *file_in, struct file *file_out);
3024 ssize_t filemap_read(struct kiocb *iocb, struct iov_iter *to,
3025 ssize_t already_read);
3026 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
3027 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
3028 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
3029 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
3030 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
3032 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
3034 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
3036 ssize_t vfs_iocb_iter_read(struct file *file, struct kiocb *iocb,
3037 struct iov_iter *iter);
3038 ssize_t vfs_iocb_iter_write(struct file *file, struct kiocb *iocb,
3039 struct iov_iter *iter);
3042 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
3043 struct pipe_inode_info *, size_t, unsigned int);
3044 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
3045 struct file *, loff_t *, size_t, unsigned int);
3046 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
3047 struct file *out, loff_t *, size_t len, unsigned int flags);
3048 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
3049 loff_t *opos, size_t len, unsigned int flags);
3053 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
3054 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
3055 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
3056 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
3057 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
3058 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
3059 int whence, loff_t maxsize, loff_t eof);
3060 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
3061 int whence, loff_t size);
3062 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
3063 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
3064 int rw_verify_area(int, struct file *, const loff_t *, size_t);
3065 extern int generic_file_open(struct inode * inode, struct file * filp);
3066 extern int nonseekable_open(struct inode * inode, struct file * filp);
3067 extern int stream_open(struct inode * inode, struct file * filp);
3070 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
3071 loff_t file_offset);
3074 /* need locking between buffered and direct access */
3077 /* filesystem does not support filling holes */
3078 DIO_SKIP_HOLES = 0x02,
3081 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
3082 struct block_device *bdev, struct iov_iter *iter,
3083 get_block_t get_block,
3084 dio_iodone_t end_io, dio_submit_t submit_io,
3087 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
3088 struct inode *inode,
3089 struct iov_iter *iter,
3090 get_block_t get_block)
3092 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
3093 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
3097 void inode_dio_wait(struct inode *inode);
3100 * inode_dio_begin - signal start of a direct I/O requests
3101 * @inode: inode the direct I/O happens on
3103 * This is called once we've finished processing a direct I/O request,
3104 * and is used to wake up callers waiting for direct I/O to be quiesced.
3106 static inline void inode_dio_begin(struct inode *inode)
3108 atomic_inc(&inode->i_dio_count);
3112 * inode_dio_end - signal finish of a direct I/O requests
3113 * @inode: inode the direct I/O happens on
3115 * This is called once we've finished processing a direct I/O request,
3116 * and is used to wake up callers waiting for direct I/O to be quiesced.
3118 static inline void inode_dio_end(struct inode *inode)
3120 if (atomic_dec_and_test(&inode->i_dio_count))
3121 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
3125 * Warn about a page cache invalidation failure diring a direct I/O write.
3127 void dio_warn_stale_pagecache(struct file *filp);
3129 extern void inode_set_flags(struct inode *inode, unsigned int flags,
3132 extern const struct file_operations generic_ro_fops;
3134 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3136 extern int readlink_copy(char __user *, int, const char *);
3137 extern int page_readlink(struct dentry *, char __user *, int);
3138 extern const char *page_get_link(struct dentry *, struct inode *,
3139 struct delayed_call *);
3140 extern void page_put_link(void *);
3141 extern int __page_symlink(struct inode *inode, const char *symname, int len,
3143 extern int page_symlink(struct inode *inode, const char *symname, int len);
3144 extern const struct inode_operations page_symlink_inode_operations;
3145 extern void kfree_link(void *);
3146 void generic_fillattr(struct user_namespace *, struct inode *, struct kstat *);
3147 void generic_fill_statx_attr(struct inode *inode, struct kstat *stat);
3148 extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
3149 extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
3150 void __inode_add_bytes(struct inode *inode, loff_t bytes);
3151 void inode_add_bytes(struct inode *inode, loff_t bytes);
3152 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
3153 void inode_sub_bytes(struct inode *inode, loff_t bytes);
3154 static inline loff_t __inode_get_bytes(struct inode *inode)
3156 return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes;
3158 loff_t inode_get_bytes(struct inode *inode);
3159 void inode_set_bytes(struct inode *inode, loff_t bytes);
3160 const char *simple_get_link(struct dentry *, struct inode *,
3161 struct delayed_call *);
3162 extern const struct inode_operations simple_symlink_inode_operations;
3164 extern int iterate_dir(struct file *, struct dir_context *);
3166 int vfs_fstatat(int dfd, const char __user *filename, struct kstat *stat,
3168 int vfs_fstat(int fd, struct kstat *stat);
3170 static inline int vfs_stat(const char __user *filename, struct kstat *stat)
3172 return vfs_fstatat(AT_FDCWD, filename, stat, 0);
3174 static inline int vfs_lstat(const char __user *name, struct kstat *stat)
3176 return vfs_fstatat(AT_FDCWD, name, stat, AT_SYMLINK_NOFOLLOW);
3179 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
3180 extern int vfs_readlink(struct dentry *, char __user *, int);
3182 extern struct file_system_type *get_filesystem(struct file_system_type *fs);
3183 extern void put_filesystem(struct file_system_type *fs);
3184 extern struct file_system_type *get_fs_type(const char *name);
3185 extern struct super_block *get_super(struct block_device *);
3186 extern struct super_block *get_active_super(struct block_device *bdev);
3187 extern void drop_super(struct super_block *sb);
3188 extern void drop_super_exclusive(struct super_block *sb);
3189 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
3190 extern void iterate_supers_type(struct file_system_type *,
3191 void (*)(struct super_block *, void *), void *);
3193 extern int dcache_dir_open(struct inode *, struct file *);
3194 extern int dcache_dir_close(struct inode *, struct file *);
3195 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
3196 extern int dcache_readdir(struct file *, struct dir_context *);
3197 extern int simple_setattr(struct user_namespace *, struct dentry *,
3199 extern int simple_getattr(struct user_namespace *, const struct path *,
3200 struct kstat *, u32, unsigned int);
3201 extern int simple_statfs(struct dentry *, struct kstatfs *);
3202 extern int simple_open(struct inode *inode, struct file *file);
3203 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3204 extern int simple_unlink(struct inode *, struct dentry *);
3205 extern int simple_rmdir(struct inode *, struct dentry *);
3206 extern int simple_rename_exchange(struct inode *old_dir, struct dentry *old_dentry,
3207 struct inode *new_dir, struct dentry *new_dentry);
3208 extern int simple_rename(struct user_namespace *, struct inode *,
3209 struct dentry *, struct inode *, struct dentry *,
3211 extern void simple_recursive_removal(struct dentry *,
3212 void (*callback)(struct dentry *));
3213 extern int noop_fsync(struct file *, loff_t, loff_t, int);
3214 extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter);
3215 extern int simple_empty(struct dentry *);
3216 extern int simple_write_begin(struct file *file, struct address_space *mapping,
3217 loff_t pos, unsigned len, unsigned flags,
3218 struct page **pagep, void **fsdata);
3219 extern const struct address_space_operations ram_aops;
3220 extern int always_delete_dentry(const struct dentry *);
3221 extern struct inode *alloc_anon_inode(struct super_block *);
3222 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
3223 extern const struct dentry_operations simple_dentry_operations;
3225 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3226 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3227 extern const struct file_operations simple_dir_operations;
3228 extern const struct inode_operations simple_dir_inode_operations;
3229 extern void make_empty_dir_inode(struct inode *inode);
3230 extern bool is_empty_dir_inode(struct inode *inode);
3231 struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
3232 struct dentry *d_alloc_name(struct dentry *, const char *);
3233 extern int simple_fill_super(struct super_block *, unsigned long,
3234 const struct tree_descr *);
3235 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3236 extern void simple_release_fs(struct vfsmount **mount, int *count);
3238 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3239 loff_t *ppos, const void *from, size_t available);
3240 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3241 const void __user *from, size_t count);
3243 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3244 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3246 extern int generic_check_addressable(unsigned, u64);
3248 extern void generic_set_encrypted_ci_d_ops(struct dentry *dentry);
3250 #ifdef CONFIG_MIGRATION
3251 extern int buffer_migrate_page(struct address_space *,
3252 struct page *, struct page *,
3254 extern int buffer_migrate_page_norefs(struct address_space *,
3255 struct page *, struct page *,
3258 #define buffer_migrate_page NULL
3259 #define buffer_migrate_page_norefs NULL
3262 int may_setattr(struct user_namespace *mnt_userns, struct inode *inode,
3263 unsigned int ia_valid);
3264 int setattr_prepare(struct user_namespace *, struct dentry *, struct iattr *);
3265 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3266 void setattr_copy(struct user_namespace *, struct inode *inode,
3267 const struct iattr *attr);
3269 extern int file_update_time(struct file *file);
3271 static inline bool vma_is_dax(const struct vm_area_struct *vma)
3273 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3276 static inline bool vma_is_fsdax(struct vm_area_struct *vma)
3278 struct inode *inode;
3280 if (!IS_ENABLED(CONFIG_FS_DAX) || !vma->vm_file)
3282 if (!vma_is_dax(vma))
3284 inode = file_inode(vma->vm_file);
3285 if (S_ISCHR(inode->i_mode))
3286 return false; /* device-dax */
3290 static inline int iocb_flags(struct file *file)
3293 if (file->f_flags & O_APPEND)
3295 if (file->f_flags & O_DIRECT)
3297 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3299 if (file->f_flags & __O_SYNC)
3304 static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
3306 int kiocb_flags = 0;
3308 /* make sure there's no overlap between RWF and private IOCB flags */
3309 BUILD_BUG_ON((__force int) RWF_SUPPORTED & IOCB_EVENTFD);
3313 if (unlikely(flags & ~RWF_SUPPORTED))
3316 if (flags & RWF_NOWAIT) {
3317 if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
3319 kiocb_flags |= IOCB_NOIO;
3321 kiocb_flags |= (__force int) (flags & RWF_SUPPORTED);
3322 if (flags & RWF_SYNC)
3323 kiocb_flags |= IOCB_DSYNC;
3325 ki->ki_flags |= kiocb_flags;
3329 static inline ino_t parent_ino(struct dentry *dentry)
3334 * Don't strictly need d_lock here? If the parent ino could change
3335 * then surely we'd have a deeper race in the caller?
3337 spin_lock(&dentry->d_lock);
3338 res = dentry->d_parent->d_inode->i_ino;
3339 spin_unlock(&dentry->d_lock);
3343 /* Transaction based IO helpers */
3346 * An argresp is stored in an allocated page and holds the
3347 * size of the argument or response, along with its content
3349 struct simple_transaction_argresp {
3354 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3356 char *simple_transaction_get(struct file *file, const char __user *buf,
3358 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3359 size_t size, loff_t *pos);
3360 int simple_transaction_release(struct inode *inode, struct file *file);
3362 void simple_transaction_set(struct file *file, size_t n);
3365 * simple attribute files
3367 * These attributes behave similar to those in sysfs:
3369 * Writing to an attribute immediately sets a value, an open file can be
3370 * written to multiple times.
3372 * Reading from an attribute creates a buffer from the value that might get
3373 * read with multiple read calls. When the attribute has been read
3374 * completely, no further read calls are possible until the file is opened
3377 * All attributes contain a text representation of a numeric value
3378 * that are accessed with the get() and set() functions.
3380 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3381 static int __fops ## _open(struct inode *inode, struct file *file) \
3383 __simple_attr_check_format(__fmt, 0ull); \
3384 return simple_attr_open(inode, file, __get, __set, __fmt); \
3386 static const struct file_operations __fops = { \
3387 .owner = THIS_MODULE, \
3388 .open = __fops ## _open, \
3389 .release = simple_attr_release, \
3390 .read = simple_attr_read, \
3391 .write = simple_attr_write, \
3392 .llseek = generic_file_llseek, \
3395 static inline __printf(1, 2)
3396 void __simple_attr_check_format(const char *fmt, ...)
3398 /* don't do anything, just let the compiler check the arguments; */
3401 int simple_attr_open(struct inode *inode, struct file *file,
3402 int (*get)(void *, u64 *), int (*set)(void *, u64),
3404 int simple_attr_release(struct inode *inode, struct file *file);
3405 ssize_t simple_attr_read(struct file *file, char __user *buf,
3406 size_t len, loff_t *ppos);
3407 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3408 size_t len, loff_t *ppos);
3411 int __init list_bdev_fs_names(char *buf, size_t size);
3413 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3414 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3416 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3417 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3418 (flag & __FMODE_NONOTIFY)))
3420 static inline bool is_sxid(umode_t mode)
3422 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3425 static inline int check_sticky(struct user_namespace *mnt_userns,
3426 struct inode *dir, struct inode *inode)
3428 if (!(dir->i_mode & S_ISVTX))
3431 return __check_sticky(mnt_userns, dir, inode);
3434 static inline void inode_has_no_xattr(struct inode *inode)
3436 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
3437 inode->i_flags |= S_NOSEC;
3440 static inline bool is_root_inode(struct inode *inode)
3442 return inode == inode->i_sb->s_root->d_inode;
3445 static inline bool dir_emit(struct dir_context *ctx,
3446 const char *name, int namelen,
3447 u64 ino, unsigned type)
3449 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3451 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3453 return ctx->actor(ctx, ".", 1, ctx->pos,
3454 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3456 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3458 return ctx->actor(ctx, "..", 2, ctx->pos,
3459 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3461 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3463 if (ctx->pos == 0) {
3464 if (!dir_emit_dot(file, ctx))
3468 if (ctx->pos == 1) {
3469 if (!dir_emit_dotdot(file, ctx))
3475 static inline bool dir_relax(struct inode *inode)
3477 inode_unlock(inode);
3479 return !IS_DEADDIR(inode);
3482 static inline bool dir_relax_shared(struct inode *inode)
3484 inode_unlock_shared(inode);
3485 inode_lock_shared(inode);
3486 return !IS_DEADDIR(inode);
3489 extern bool path_noexec(const struct path *path);
3490 extern void inode_nohighmem(struct inode *inode);
3493 extern int vfs_fadvise(struct file *file, loff_t offset, loff_t len,
3495 extern int generic_fadvise(struct file *file, loff_t offset, loff_t len,
3498 #endif /* _LINUX_FS_H */