2 * include/linux/buffer_head.h
4 * Everything to do with buffer_heads.
7 #ifndef _LINUX_BUFFER_HEAD_H
8 #define _LINUX_BUFFER_HEAD_H
10 #include <linux/types.h>
12 #include <linux/linkage.h>
13 #include <linux/pagemap.h>
14 #include <linux/wait.h>
15 #include <linux/atomic.h>
20 BH_Uptodate, /* Contains valid data */
21 BH_Dirty, /* Is dirty */
22 BH_Lock, /* Is locked */
23 BH_Req, /* Has been submitted for I/O */
24 BH_Uptodate_Lock,/* Used by the first bh in a page, to serialise
25 * IO completion of other buffers in the page
28 BH_Mapped, /* Has a disk mapping */
29 BH_New, /* Disk mapping was newly created by get_block */
30 BH_Async_Read, /* Is under end_buffer_async_read I/O */
31 BH_Async_Write, /* Is under end_buffer_async_write I/O */
32 BH_Delay, /* Buffer is not yet allocated on disk */
33 BH_Boundary, /* Block is followed by a discontiguity */
34 BH_Write_EIO, /* I/O error on write */
35 BH_Unwritten, /* Buffer is allocated on disk but not written */
36 BH_Quiet, /* Buffer Error Prinks to be quiet */
37 BH_Meta, /* Buffer contains metadata */
38 BH_Prio, /* Buffer should be submitted with REQ_PRIO */
39 BH_Defer_Completion, /* Defer AIO completion to workqueue */
41 BH_PrivateStart,/* not a state bit, but the first bit available
42 * for private allocation by other entities
46 #define MAX_BUF_PER_PAGE (PAGE_SIZE / 512)
51 typedef void (bh_end_io_t)(struct buffer_head *bh, int uptodate);
54 * Historically, a buffer_head was used to map a single block
55 * within a page, and of course as the unit of I/O through the
56 * filesystem and block layers. Nowadays the basic I/O unit
57 * is the bio, and buffer_heads are used for extracting block
58 * mappings (via a get_block_t call), for tracking state within
59 * a page (via a page_mapping) and for wrapping bio submission
60 * for backward compatibility reasons (e.g. submit_bh).
63 unsigned long b_state; /* buffer state bitmap (see above) */
64 struct buffer_head *b_this_page;/* circular list of page's buffers */
65 struct page *b_page; /* the page this bh is mapped to */
67 sector_t b_blocknr; /* start block number */
68 size_t b_size; /* size of mapping */
69 char *b_data; /* pointer to data within the page */
71 struct block_device *b_bdev;
72 bh_end_io_t *b_end_io; /* I/O completion */
73 void *b_private; /* reserved for b_end_io */
74 struct list_head b_assoc_buffers; /* associated with another mapping */
75 struct address_space *b_assoc_map; /* mapping this buffer is
77 atomic_t b_count; /* users using this buffer_head */
81 * macro tricks to expand the set_buffer_foo(), clear_buffer_foo()
82 * and buffer_foo() functions.
84 #define BUFFER_FNS(bit, name) \
85 static __always_inline void set_buffer_##name(struct buffer_head *bh) \
87 set_bit(BH_##bit, &(bh)->b_state); \
89 static __always_inline void clear_buffer_##name(struct buffer_head *bh) \
91 clear_bit(BH_##bit, &(bh)->b_state); \
93 static __always_inline int buffer_##name(const struct buffer_head *bh) \
95 return test_bit(BH_##bit, &(bh)->b_state); \
99 * test_set_buffer_foo() and test_clear_buffer_foo()
101 #define TAS_BUFFER_FNS(bit, name) \
102 static __always_inline int test_set_buffer_##name(struct buffer_head *bh) \
104 return test_and_set_bit(BH_##bit, &(bh)->b_state); \
106 static __always_inline int test_clear_buffer_##name(struct buffer_head *bh) \
108 return test_and_clear_bit(BH_##bit, &(bh)->b_state); \
112 * Emit the buffer bitops functions. Note that there are also functions
113 * of the form "mark_buffer_foo()". These are higher-level functions which
114 * do something in addition to setting a b_state bit.
116 BUFFER_FNS(Dirty, dirty)
117 TAS_BUFFER_FNS(Dirty, dirty)
118 BUFFER_FNS(Lock, locked)
120 TAS_BUFFER_FNS(Req, req)
121 BUFFER_FNS(Mapped, mapped)
123 BUFFER_FNS(Async_Read, async_read)
124 BUFFER_FNS(Async_Write, async_write)
125 BUFFER_FNS(Delay, delay)
126 BUFFER_FNS(Boundary, boundary)
127 BUFFER_FNS(Write_EIO, write_io_error)
128 BUFFER_FNS(Unwritten, unwritten)
129 BUFFER_FNS(Meta, meta)
130 BUFFER_FNS(Prio, prio)
131 BUFFER_FNS(Defer_Completion, defer_completion)
133 static __always_inline void set_buffer_uptodate(struct buffer_head *bh)
136 * If somebody else already set this uptodate, they will
137 * have done the memory barrier, and a reader will thus
138 * see *some* valid buffer state.
140 * Any other serialization (with IO errors or whatever that
141 * might clear the bit) has to come from other state (eg BH_Lock).
143 if (test_bit(BH_Uptodate, &bh->b_state))
147 * make it consistent with folio_mark_uptodate
148 * pairs with smp_load_acquire in buffer_uptodate
150 smp_mb__before_atomic();
151 set_bit(BH_Uptodate, &bh->b_state);
154 static __always_inline void clear_buffer_uptodate(struct buffer_head *bh)
156 clear_bit(BH_Uptodate, &bh->b_state);
159 static __always_inline int buffer_uptodate(const struct buffer_head *bh)
162 * make it consistent with folio_test_uptodate
163 * pairs with smp_mb__before_atomic in set_buffer_uptodate
165 return (smp_load_acquire(&bh->b_state) & (1UL << BH_Uptodate)) != 0;
168 #define bh_offset(bh) ((unsigned long)(bh)->b_data & ~PAGE_MASK)
170 /* If we *know* page->private refers to buffer_heads */
171 #define page_buffers(page) \
173 BUG_ON(!PagePrivate(page)); \
174 ((struct buffer_head *)page_private(page)); \
176 #define page_has_buffers(page) PagePrivate(page)
178 void buffer_check_dirty_writeback(struct page *page,
179 bool *dirty, bool *writeback);
185 void mark_buffer_dirty(struct buffer_head *bh);
186 void init_buffer(struct buffer_head *, bh_end_io_t *, void *);
187 void touch_buffer(struct buffer_head *bh);
188 void set_bh_page(struct buffer_head *bh,
189 struct page *page, unsigned long offset);
190 int try_to_free_buffers(struct page *);
191 struct buffer_head *alloc_page_buffers(struct page *page, unsigned long size,
193 void create_empty_buffers(struct page *, unsigned long,
194 unsigned long b_state);
195 void end_buffer_read_sync(struct buffer_head *bh, int uptodate);
196 void end_buffer_write_sync(struct buffer_head *bh, int uptodate);
197 void end_buffer_async_write(struct buffer_head *bh, int uptodate);
199 /* Things to do with buffers at mapping->private_list */
200 void mark_buffer_dirty_inode(struct buffer_head *bh, struct inode *inode);
201 int inode_has_buffers(struct inode *);
202 void invalidate_inode_buffers(struct inode *);
203 int remove_inode_buffers(struct inode *inode);
204 int sync_mapping_buffers(struct address_space *mapping);
205 void unmap_underlying_metadata(struct block_device *bdev, sector_t block);
207 void mark_buffer_async_write(struct buffer_head *bh);
208 void __wait_on_buffer(struct buffer_head *);
209 wait_queue_head_t *bh_waitq_head(struct buffer_head *bh);
210 struct buffer_head *__find_get_block(struct block_device *bdev, sector_t block,
212 struct buffer_head *__getblk_gfp(struct block_device *bdev, sector_t block,
213 unsigned size, gfp_t gfp);
214 void __brelse(struct buffer_head *);
215 void __bforget(struct buffer_head *);
216 void __breadahead(struct block_device *, sector_t block, unsigned int size);
217 struct buffer_head *__bread_gfp(struct block_device *,
218 sector_t block, unsigned size, gfp_t gfp);
219 void invalidate_bh_lrus(void);
220 struct buffer_head *alloc_buffer_head(gfp_t gfp_flags);
221 void free_buffer_head(struct buffer_head * bh);
222 void unlock_buffer(struct buffer_head *bh);
223 void __lock_buffer(struct buffer_head *bh);
224 void ll_rw_block(int, int, int, struct buffer_head * bh[]);
225 int sync_dirty_buffer(struct buffer_head *bh);
226 int __sync_dirty_buffer(struct buffer_head *bh, int op_flags);
227 void write_dirty_buffer(struct buffer_head *bh, int op_flags);
228 int _submit_bh(int op, int op_flags, struct buffer_head *bh,
229 unsigned long bio_flags);
230 int submit_bh(int, int, struct buffer_head *);
231 void write_boundary_block(struct block_device *bdev,
232 sector_t bblock, unsigned blocksize);
233 int bh_uptodate_or_lock(struct buffer_head *bh);
234 int bh_submit_read(struct buffer_head *bh);
236 extern int buffer_heads_over_limit;
239 * Generic address_space_operations implementations for buffer_head-backed
242 void block_invalidatepage(struct page *page, unsigned int offset,
243 unsigned int length);
244 int block_write_full_page(struct page *page, get_block_t *get_block,
245 struct writeback_control *wbc);
246 int __block_write_full_page(struct inode *inode, struct page *page,
247 get_block_t *get_block, struct writeback_control *wbc,
248 bh_end_io_t *handler);
249 int block_read_full_page(struct page*, get_block_t*);
250 int block_is_partially_uptodate(struct page *page, unsigned long from,
251 unsigned long count);
252 int block_write_begin(struct address_space *mapping, loff_t pos, unsigned len,
253 unsigned flags, struct page **pagep, get_block_t *get_block);
254 int __block_write_begin(struct page *page, loff_t pos, unsigned len,
255 get_block_t *get_block);
256 int block_write_end(struct file *, struct address_space *,
257 loff_t, unsigned, unsigned,
258 struct page *, void *);
259 int generic_write_end(struct file *, struct address_space *,
260 loff_t, unsigned, unsigned,
261 struct page *, void *);
262 void page_zero_new_buffers(struct page *page, unsigned from, unsigned to);
263 void clean_page_buffers(struct page *page);
264 int cont_write_begin(struct file *, struct address_space *, loff_t,
265 unsigned, unsigned, struct page **, void **,
266 get_block_t *, loff_t *);
267 int generic_cont_expand_simple(struct inode *inode, loff_t size);
268 int block_commit_write(struct page *page, unsigned from, unsigned to);
269 int block_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf,
270 get_block_t get_block);
271 /* Convert errno to return value from ->page_mkwrite() call */
272 static inline int block_page_mkwrite_return(int err)
275 return VM_FAULT_LOCKED;
276 if (err == -EFAULT || err == -EAGAIN)
277 return VM_FAULT_NOPAGE;
280 /* -ENOSPC, -EDQUOT, -EIO ... */
281 return VM_FAULT_SIGBUS;
283 sector_t generic_block_bmap(struct address_space *, sector_t, get_block_t *);
284 int block_truncate_page(struct address_space *, loff_t, get_block_t *);
285 int nobh_write_begin(struct address_space *, loff_t, unsigned, unsigned,
286 struct page **, void **, get_block_t*);
287 int nobh_write_end(struct file *, struct address_space *,
288 loff_t, unsigned, unsigned,
289 struct page *, void *);
290 int nobh_truncate_page(struct address_space *, loff_t, get_block_t *);
291 int nobh_writepage(struct page *page, get_block_t *get_block,
292 struct writeback_control *wbc);
294 void buffer_init(void);
300 static inline void attach_page_buffers(struct page *page,
301 struct buffer_head *head)
304 SetPagePrivate(page);
305 set_page_private(page, (unsigned long)head);
308 static inline void get_bh(struct buffer_head *bh)
310 atomic_inc(&bh->b_count);
313 static inline void put_bh(struct buffer_head *bh)
315 smp_mb__before_atomic();
316 atomic_dec(&bh->b_count);
319 static inline void brelse(struct buffer_head *bh)
325 static inline void bforget(struct buffer_head *bh)
331 static inline struct buffer_head *
332 sb_bread(struct super_block *sb, sector_t block)
334 return __bread_gfp(sb->s_bdev, block, sb->s_blocksize, __GFP_MOVABLE);
337 static inline struct buffer_head *
338 sb_bread_unmovable(struct super_block *sb, sector_t block)
340 return __bread_gfp(sb->s_bdev, block, sb->s_blocksize, 0);
344 sb_breadahead(struct super_block *sb, sector_t block)
346 __breadahead(sb->s_bdev, block, sb->s_blocksize);
349 static inline struct buffer_head *
350 sb_getblk(struct super_block *sb, sector_t block)
352 return __getblk_gfp(sb->s_bdev, block, sb->s_blocksize, __GFP_MOVABLE);
356 static inline struct buffer_head *
357 sb_getblk_gfp(struct super_block *sb, sector_t block, gfp_t gfp)
359 return __getblk_gfp(sb->s_bdev, block, sb->s_blocksize, gfp);
362 static inline struct buffer_head *
363 sb_find_get_block(struct super_block *sb, sector_t block)
365 return __find_get_block(sb->s_bdev, block, sb->s_blocksize);
369 map_bh(struct buffer_head *bh, struct super_block *sb, sector_t block)
371 set_buffer_mapped(bh);
372 bh->b_bdev = sb->s_bdev;
373 bh->b_blocknr = block;
374 bh->b_size = sb->s_blocksize;
377 static inline void wait_on_buffer(struct buffer_head *bh)
380 if (buffer_locked(bh))
381 __wait_on_buffer(bh);
384 static inline int trylock_buffer(struct buffer_head *bh)
386 return likely(!test_and_set_bit_lock(BH_Lock, &bh->b_state));
389 static inline void lock_buffer(struct buffer_head *bh)
392 if (!trylock_buffer(bh))
396 static inline struct buffer_head *getblk_unmovable(struct block_device *bdev,
400 return __getblk_gfp(bdev, block, size, 0);
403 static inline struct buffer_head *__getblk(struct block_device *bdev,
407 return __getblk_gfp(bdev, block, size, __GFP_MOVABLE);
411 * __bread() - reads a specified block and returns the bh
412 * @bdev: the block_device to read from
413 * @block: number of block
414 * @size: size (in bytes) to read
416 * Reads a specified block, and returns buffer head that contains it.
417 * The page cache is allocated from movable area so that it can be migrated.
418 * It returns NULL if the block was unreadable.
420 static inline struct buffer_head *
421 __bread(struct block_device *bdev, sector_t block, unsigned size)
423 return __bread_gfp(bdev, block, size, __GFP_MOVABLE);
426 extern int __set_page_dirty_buffers(struct page *page);
428 #else /* CONFIG_BLOCK */
430 static inline void buffer_init(void) {}
431 static inline int try_to_free_buffers(struct page *page) { return 1; }
432 static inline int inode_has_buffers(struct inode *inode) { return 0; }
433 static inline void invalidate_inode_buffers(struct inode *inode) {}
434 static inline int remove_inode_buffers(struct inode *inode) { return 1; }
435 static inline int sync_mapping_buffers(struct address_space *mapping) { return 0; }
437 #endif /* CONFIG_BLOCK */
438 #endif /* _LINUX_BUFFER_HEAD_H */