Merge tag 'kvm-s390-next-5.19-2' of git://git.kernel.org/pub/scm/linux/kernel/git...
[linux-block.git] / include / linux / buffer_head.h
CommitLineData
b2441318 1/* SPDX-License-Identifier: GPL-2.0 */
1da177e4
LT
2/*
3 * include/linux/buffer_head.h
4 *
5 * Everything to do with buffer_heads.
6 */
7
8#ifndef _LINUX_BUFFER_HEAD_H
9#define _LINUX_BUFFER_HEAD_H
10
11#include <linux/types.h>
12#include <linux/fs.h>
13#include <linux/linkage.h>
14#include <linux/pagemap.h>
15#include <linux/wait.h>
60063497 16#include <linux/atomic.h>
1da177e4 17
9361401e
DH
18#ifdef CONFIG_BLOCK
19
1da177e4
LT
20enum bh_state_bits {
21 BH_Uptodate, /* Contains valid data */
22 BH_Dirty, /* Is dirty */
23 BH_Lock, /* Is locked */
24 BH_Req, /* Has been submitted for I/O */
25
26 BH_Mapped, /* Has a disk mapping */
27 BH_New, /* Disk mapping was newly created by get_block */
28 BH_Async_Read, /* Is under end_buffer_async_read I/O */
29 BH_Async_Write, /* Is under end_buffer_async_write I/O */
30 BH_Delay, /* Buffer is not yet allocated on disk */
31 BH_Boundary, /* Block is followed by a discontiguity */
32 BH_Write_EIO, /* I/O error on write */
33a266dd 33 BH_Unwritten, /* Buffer is allocated on disk but not written */
08bafc03 34 BH_Quiet, /* Buffer Error Prinks to be quiet */
877f962c
TT
35 BH_Meta, /* Buffer contains metadata */
36 BH_Prio, /* Buffer should be submitted with REQ_PRIO */
7b7a8665 37 BH_Defer_Completion, /* Defer AIO completion to workqueue */
1da177e4
LT
38
39 BH_PrivateStart,/* not a state bit, but the first bit available
40 * for private allocation by other entities
41 */
42};
43
09cbfeaf 44#define MAX_BUF_PER_PAGE (PAGE_SIZE / 512)
1da177e4
LT
45
46struct page;
47struct buffer_head;
48struct address_space;
49typedef void (bh_end_io_t)(struct buffer_head *bh, int uptodate);
50
51/*
205f87f6
BP
52 * Historically, a buffer_head was used to map a single block
53 * within a page, and of course as the unit of I/O through the
54 * filesystem and block layers. Nowadays the basic I/O unit
55 * is the bio, and buffer_heads are used for extracting block
56 * mappings (via a get_block_t call), for tracking state within
57 * a page (via a page_mapping) and for wrapping bio submission
58 * for backward compatibility reasons (e.g. submit_bh).
1da177e4
LT
59 */
60struct buffer_head {
1da177e4
LT
61 unsigned long b_state; /* buffer state bitmap (see above) */
62 struct buffer_head *b_this_page;/* circular list of page's buffers */
63 struct page *b_page; /* the page this bh is mapped to */
1da177e4 64
205f87f6
BP
65 sector_t b_blocknr; /* start block number */
66 size_t b_size; /* size of mapping */
67 char *b_data; /* pointer to data within the page */
1da177e4
LT
68
69 struct block_device *b_bdev;
70 bh_end_io_t *b_end_io; /* I/O completion */
71 void *b_private; /* reserved for b_end_io */
72 struct list_head b_assoc_buffers; /* associated with another mapping */
58ff407b
JK
73 struct address_space *b_assoc_map; /* mapping this buffer is
74 associated with */
205f87f6 75 atomic_t b_count; /* users using this buffer_head */
f1e67e35
TG
76 spinlock_t b_uptodate_lock; /* Used by the first bh in a page, to
77 * serialise IO completion of other
78 * buffers in the page */
1da177e4
LT
79};
80
81/*
82 * macro tricks to expand the set_buffer_foo(), clear_buffer_foo()
83 * and buffer_foo() functions.
60f91826
KW
84 * To avoid reset buffer flags that are already set, because that causes
85 * a costly cache line transition, check the flag first.
1da177e4
LT
86 */
87#define BUFFER_FNS(bit, name) \
ee91ef61 88static __always_inline void set_buffer_##name(struct buffer_head *bh) \
1da177e4 89{ \
60f91826
KW
90 if (!test_bit(BH_##bit, &(bh)->b_state)) \
91 set_bit(BH_##bit, &(bh)->b_state); \
1da177e4 92} \
ee91ef61 93static __always_inline void clear_buffer_##name(struct buffer_head *bh) \
1da177e4
LT
94{ \
95 clear_bit(BH_##bit, &(bh)->b_state); \
96} \
ee91ef61 97static __always_inline int buffer_##name(const struct buffer_head *bh) \
1da177e4
LT
98{ \
99 return test_bit(BH_##bit, &(bh)->b_state); \
100}
101
102/*
103 * test_set_buffer_foo() and test_clear_buffer_foo()
104 */
105#define TAS_BUFFER_FNS(bit, name) \
ee91ef61 106static __always_inline int test_set_buffer_##name(struct buffer_head *bh) \
1da177e4
LT
107{ \
108 return test_and_set_bit(BH_##bit, &(bh)->b_state); \
109} \
ee91ef61 110static __always_inline int test_clear_buffer_##name(struct buffer_head *bh) \
1da177e4
LT
111{ \
112 return test_and_clear_bit(BH_##bit, &(bh)->b_state); \
113} \
114
115/*
116 * Emit the buffer bitops functions. Note that there are also functions
117 * of the form "mark_buffer_foo()". These are higher-level functions which
118 * do something in addition to setting a b_state bit.
119 */
120BUFFER_FNS(Uptodate, uptodate)
121BUFFER_FNS(Dirty, dirty)
122TAS_BUFFER_FNS(Dirty, dirty)
123BUFFER_FNS(Lock, locked)
1da177e4
LT
124BUFFER_FNS(Req, req)
125TAS_BUFFER_FNS(Req, req)
126BUFFER_FNS(Mapped, mapped)
127BUFFER_FNS(New, new)
128BUFFER_FNS(Async_Read, async_read)
129BUFFER_FNS(Async_Write, async_write)
130BUFFER_FNS(Delay, delay)
131BUFFER_FNS(Boundary, boundary)
132BUFFER_FNS(Write_EIO, write_io_error)
33a266dd 133BUFFER_FNS(Unwritten, unwritten)
877f962c
TT
134BUFFER_FNS(Meta, meta)
135BUFFER_FNS(Prio, prio)
7b7a8665 136BUFFER_FNS(Defer_Completion, defer_completion)
1da177e4
LT
137
138#define bh_offset(bh) ((unsigned long)(bh)->b_data & ~PAGE_MASK)
1da177e4
LT
139
140/* If we *know* page->private refers to buffer_heads */
141#define page_buffers(page) \
142 ({ \
4c21e2f2
HD
143 BUG_ON(!PagePrivate(page)); \
144 ((struct buffer_head *)page_private(page)); \
1da177e4
LT
145 })
146#define page_has_buffers(page) PagePrivate(page)
cd1067be 147#define folio_buffers(folio) folio_get_private(folio)
1da177e4 148
520f301c 149void buffer_check_dirty_writeback(struct folio *folio,
b4597226
MG
150 bool *dirty, bool *writeback);
151
1da177e4
LT
152/*
153 * Declarations
154 */
155
b3c97528 156void mark_buffer_dirty(struct buffer_head *bh);
87354e5d 157void mark_buffer_write_io_error(struct buffer_head *bh);
f0059afd 158void touch_buffer(struct buffer_head *bh);
1da177e4
LT
159void set_bh_page(struct buffer_head *bh,
160 struct page *page, unsigned long offset);
68189fef 161bool try_to_free_buffers(struct folio *);
1da177e4 162struct buffer_head *alloc_page_buffers(struct page *page, unsigned long size,
640ab98f 163 bool retry);
1da177e4
LT
164void create_empty_buffers(struct page *, unsigned long,
165 unsigned long b_state);
166void end_buffer_read_sync(struct buffer_head *bh, int uptodate);
167void end_buffer_write_sync(struct buffer_head *bh, int uptodate);
35c80d5f 168void end_buffer_async_write(struct buffer_head *bh, int uptodate);
1da177e4
LT
169
170/* Things to do with buffers at mapping->private_list */
171void mark_buffer_dirty_inode(struct buffer_head *bh, struct inode *inode);
172int inode_has_buffers(struct inode *);
173void invalidate_inode_buffers(struct inode *);
174int remove_inode_buffers(struct inode *inode);
175int sync_mapping_buffers(struct address_space *mapping);
29f3ad7d
JK
176void clean_bdev_aliases(struct block_device *bdev, sector_t block,
177 sector_t len);
e64855c6
JK
178static inline void clean_bdev_bh_alias(struct buffer_head *bh)
179{
180 clean_bdev_aliases(bh->b_bdev, bh->b_blocknr, 1);
181}
1da177e4
LT
182
183void mark_buffer_async_write(struct buffer_head *bh);
1da177e4
LT
184void __wait_on_buffer(struct buffer_head *);
185wait_queue_head_t *bh_waitq_head(struct buffer_head *bh);
3991d3bd
TK
186struct buffer_head *__find_get_block(struct block_device *bdev, sector_t block,
187 unsigned size);
3b5e6454
GK
188struct buffer_head *__getblk_gfp(struct block_device *bdev, sector_t block,
189 unsigned size, gfp_t gfp);
1da177e4
LT
190void __brelse(struct buffer_head *);
191void __bforget(struct buffer_head *);
3991d3bd 192void __breadahead(struct block_device *, sector_t block, unsigned int size);
d87f6392
RG
193void __breadahead_gfp(struct block_device *, sector_t block, unsigned int size,
194 gfp_t gfp);
3b5e6454
GK
195struct buffer_head *__bread_gfp(struct block_device *,
196 sector_t block, unsigned size, gfp_t gfp);
f9a14399 197void invalidate_bh_lrus(void);
243418e3 198void invalidate_bh_lrus_cpu(void);
8cc621d2 199bool has_bh_in_lru(int cpu, void *dummy);
dd0fc66f 200struct buffer_head *alloc_buffer_head(gfp_t gfp_flags);
1da177e4 201void free_buffer_head(struct buffer_head * bh);
b3c97528
HH
202void unlock_buffer(struct buffer_head *bh);
203void __lock_buffer(struct buffer_head *bh);
dfec8a14 204void ll_rw_block(int, int, int, struct buffer_head * bh[]);
1da177e4 205int sync_dirty_buffer(struct buffer_head *bh);
2a222ca9
MC
206int __sync_dirty_buffer(struct buffer_head *bh, int op_flags);
207void write_dirty_buffer(struct buffer_head *bh, int op_flags);
2a222ca9 208int submit_bh(int, int, struct buffer_head *);
1da177e4
LT
209void write_boundary_block(struct block_device *bdev,
210 sector_t bblock, unsigned blocksize);
389d1b08
AK
211int bh_uptodate_or_lock(struct buffer_head *bh);
212int bh_submit_read(struct buffer_head *bh);
1da177e4
LT
213
214extern int buffer_heads_over_limit;
215
216/*
217 * Generic address_space_operations implementations for buffer_head-backed
218 * address_spaces.
219 */
7ba13abb 220void block_invalidate_folio(struct folio *folio, size_t offset, size_t length);
1da177e4
LT
221int block_write_full_page(struct page *page, get_block_t *get_block,
222 struct writeback_control *wbc);
b4bba389
BM
223int __block_write_full_page(struct inode *inode, struct page *page,
224 get_block_t *get_block, struct writeback_control *wbc,
225 bh_end_io_t *handler);
2c69e205 226int block_read_full_folio(struct folio *, get_block_t *);
2e7e80f7 227bool block_is_partially_uptodate(struct folio *, size_t from, size_t count);
155130a4 228int block_write_begin(struct address_space *mapping, loff_t pos, unsigned len,
b3992d1e 229 struct page **pagep, get_block_t *get_block);
6e1db88d
CH
230int __block_write_begin(struct page *page, loff_t pos, unsigned len,
231 get_block_t *get_block);
afddba49
NP
232int block_write_end(struct file *, struct address_space *,
233 loff_t, unsigned, unsigned,
234 struct page *, void *);
235int generic_write_end(struct file *, struct address_space *,
236 loff_t, unsigned, unsigned,
237 struct page *, void *);
238void page_zero_new_buffers(struct page *page, unsigned from, unsigned to);
f892760a 239void clean_page_buffers(struct page *page);
89e10787 240int cont_write_begin(struct file *, struct address_space *, loff_t,
be3bbbc5 241 unsigned, struct page **, void **,
89e10787 242 get_block_t *, loff_t *);
05eb0b51 243int generic_cont_expand_simple(struct inode *inode, loff_t size);
1da177e4 244int block_commit_write(struct page *page, unsigned from, unsigned to);
c2ec175c 245int block_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf,
54171690 246 get_block_t get_block);
24da4fab 247/* Convert errno to return value from ->page_mkwrite() call */
401b25aa 248static inline vm_fault_t block_page_mkwrite_return(int err)
24da4fab
JK
249{
250 if (err == 0)
251 return VM_FAULT_LOCKED;
0911d004 252 if (err == -EFAULT || err == -EAGAIN)
24da4fab
JK
253 return VM_FAULT_NOPAGE;
254 if (err == -ENOMEM)
255 return VM_FAULT_OOM;
256 /* -ENOSPC, -EDQUOT, -EIO ... */
257 return VM_FAULT_SIGBUS;
258}
1da177e4 259sector_t generic_block_bmap(struct address_space *, sector_t, get_block_t *);
1da177e4 260int block_truncate_page(struct address_space *, loff_t, get_block_t *);
8371f30c 261int nobh_write_begin(struct address_space *, loff_t, unsigned len,
03158cd7
NP
262 struct page **, void **, get_block_t*);
263int nobh_write_end(struct file *, struct address_space *,
264 loff_t, unsigned, unsigned,
265 struct page *, void *);
266int nobh_truncate_page(struct address_space *, loff_t, get_block_t *);
1da177e4
LT
267int nobh_writepage(struct page *page, get_block_t *get_block,
268 struct writeback_control *wbc);
269
b6cd0b77 270void buffer_init(void);
1da177e4
LT
271
272/*
273 * inline definitions
274 */
275
1da177e4
LT
276static inline void get_bh(struct buffer_head *bh)
277{
278 atomic_inc(&bh->b_count);
279}
280
281static inline void put_bh(struct buffer_head *bh)
282{
4e857c58 283 smp_mb__before_atomic();
1da177e4
LT
284 atomic_dec(&bh->b_count);
285}
286
287static inline void brelse(struct buffer_head *bh)
288{
289 if (bh)
290 __brelse(bh);
291}
292
293static inline void bforget(struct buffer_head *bh)
294{
295 if (bh)
296 __bforget(bh);
297}
298
299static inline struct buffer_head *
300sb_bread(struct super_block *sb, sector_t block)
301{
3b5e6454
GK
302 return __bread_gfp(sb->s_bdev, block, sb->s_blocksize, __GFP_MOVABLE);
303}
304
305static inline struct buffer_head *
306sb_bread_unmovable(struct super_block *sb, sector_t block)
307{
308 return __bread_gfp(sb->s_bdev, block, sb->s_blocksize, 0);
1da177e4
LT
309}
310
311static inline void
312sb_breadahead(struct super_block *sb, sector_t block)
313{
314 __breadahead(sb->s_bdev, block, sb->s_blocksize);
315}
316
d87f6392
RG
317static inline void
318sb_breadahead_unmovable(struct super_block *sb, sector_t block)
319{
320 __breadahead_gfp(sb->s_bdev, block, sb->s_blocksize, 0);
321}
322
1da177e4
LT
323static inline struct buffer_head *
324sb_getblk(struct super_block *sb, sector_t block)
325{
3b5e6454 326 return __getblk_gfp(sb->s_bdev, block, sb->s_blocksize, __GFP_MOVABLE);
1da177e4
LT
327}
328
bd7ade3c
NB
329
330static inline struct buffer_head *
331sb_getblk_gfp(struct super_block *sb, sector_t block, gfp_t gfp)
332{
333 return __getblk_gfp(sb->s_bdev, block, sb->s_blocksize, gfp);
334}
335
1da177e4
LT
336static inline struct buffer_head *
337sb_find_get_block(struct super_block *sb, sector_t block)
338{
339 return __find_get_block(sb->s_bdev, block, sb->s_blocksize);
340}
341
342static inline void
343map_bh(struct buffer_head *bh, struct super_block *sb, sector_t block)
344{
345 set_buffer_mapped(bh);
346 bh->b_bdev = sb->s_bdev;
347 bh->b_blocknr = block;
b0cf2321 348 bh->b_size = sb->s_blocksize;
1da177e4
LT
349}
350
1da177e4
LT
351static inline void wait_on_buffer(struct buffer_head *bh)
352{
353 might_sleep();
a9877cc2 354 if (buffer_locked(bh))
1da177e4
LT
355 __wait_on_buffer(bh);
356}
357
ca5de404
NP
358static inline int trylock_buffer(struct buffer_head *bh)
359{
51b07fc3 360 return likely(!test_and_set_bit_lock(BH_Lock, &bh->b_state));
ca5de404
NP
361}
362
1da177e4
LT
363static inline void lock_buffer(struct buffer_head *bh)
364{
365 might_sleep();
ca5de404 366 if (!trylock_buffer(bh))
1da177e4
LT
367 __lock_buffer(bh);
368}
369
3b5e6454
GK
370static inline struct buffer_head *getblk_unmovable(struct block_device *bdev,
371 sector_t block,
372 unsigned size)
373{
374 return __getblk_gfp(bdev, block, size, 0);
375}
376
377static inline struct buffer_head *__getblk(struct block_device *bdev,
378 sector_t block,
379 unsigned size)
380{
381 return __getblk_gfp(bdev, block, size, __GFP_MOVABLE);
382}
383
384/**
385 * __bread() - reads a specified block and returns the bh
386 * @bdev: the block_device to read from
387 * @block: number of block
388 * @size: size (in bytes) to read
389 *
390 * Reads a specified block, and returns buffer head that contains it.
391 * The page cache is allocated from movable area so that it can be migrated.
392 * It returns NULL if the block was unreadable.
393 */
394static inline struct buffer_head *
395__bread(struct block_device *bdev, sector_t block, unsigned size)
396{
397 return __bread_gfp(bdev, block, size, __GFP_MOVABLE);
398}
399
e621900a 400bool block_dirty_folio(struct address_space *mapping, struct folio *folio);
9361401e
DH
401
402#else /* CONFIG_BLOCK */
403
404static inline void buffer_init(void) {}
68189fef 405static inline bool try_to_free_buffers(struct folio *folio) { return true; }
9361401e
DH
406static inline int inode_has_buffers(struct inode *inode) { return 0; }
407static inline void invalidate_inode_buffers(struct inode *inode) {}
408static inline int remove_inode_buffers(struct inode *inode) { return 1; }
409static inline int sync_mapping_buffers(struct address_space *mapping) { return 0; }
243418e3 410static inline void invalidate_bh_lrus_cpu(void) {}
6de522d1 411static inline bool has_bh_in_lru(int cpu, void *dummy) { return false; }
d2de7ea4 412#define buffer_heads_over_limit 0
9361401e
DH
413
414#endif /* CONFIG_BLOCK */
1da177e4 415#endif /* _LINUX_BUFFER_HEAD_H */