xref: /openbmc/linux/fs/btrfs/extent_io.h (revision 987a6485)
1 /* SPDX-License-Identifier: GPL-2.0 */
2 
3 #ifndef BTRFS_EXTENT_IO_H
4 #define BTRFS_EXTENT_IO_H
5 
6 #include <linux/rbtree.h>
7 #include <linux/refcount.h>
8 #include <linux/fiemap.h>
9 #include "ulist.h"
10 
11 /*
12  * flags for bio submission. The high bits indicate the compression
13  * type for this bio
14  */
15 #define EXTENT_BIO_COMPRESSED 1
16 #define EXTENT_BIO_FLAG_SHIFT 16
17 
18 enum {
19 	EXTENT_BUFFER_UPTODATE,
20 	EXTENT_BUFFER_DIRTY,
21 	EXTENT_BUFFER_CORRUPT,
22 	/* this got triggered by readahead */
23 	EXTENT_BUFFER_READAHEAD,
24 	EXTENT_BUFFER_TREE_REF,
25 	EXTENT_BUFFER_STALE,
26 	EXTENT_BUFFER_WRITEBACK,
27 	/* read IO error */
28 	EXTENT_BUFFER_READ_ERR,
29 	EXTENT_BUFFER_UNMAPPED,
30 	EXTENT_BUFFER_IN_TREE,
31 	/* write IO error */
32 	EXTENT_BUFFER_WRITE_ERR,
33 };
34 
35 /* these are flags for __process_pages_contig */
36 #define PAGE_UNLOCK		(1 << 0)
37 #define PAGE_CLEAR_DIRTY	(1 << 1)
38 #define PAGE_SET_WRITEBACK	(1 << 2)
39 #define PAGE_END_WRITEBACK	(1 << 3)
40 #define PAGE_SET_PRIVATE2	(1 << 4)
41 #define PAGE_SET_ERROR		(1 << 5)
42 #define PAGE_LOCK		(1 << 6)
43 
44 /*
45  * page->private values.  Every page that is controlled by the extent
46  * map has page->private set to one.
47  */
48 #define EXTENT_PAGE_PRIVATE 1
49 
50 /*
51  * The extent buffer bitmap operations are done with byte granularity instead of
52  * word granularity for two reasons:
53  * 1. The bitmaps must be little-endian on disk.
54  * 2. Bitmap items are not guaranteed to be aligned to a word and therefore a
55  *    single word in a bitmap may straddle two pages in the extent buffer.
56  */
57 #define BIT_BYTE(nr) ((nr) / BITS_PER_BYTE)
58 #define BYTE_MASK ((1 << BITS_PER_BYTE) - 1)
59 #define BITMAP_FIRST_BYTE_MASK(start) \
60 	((BYTE_MASK << ((start) & (BITS_PER_BYTE - 1))) & BYTE_MASK)
61 #define BITMAP_LAST_BYTE_MASK(nbits) \
62 	(BYTE_MASK >> (-(nbits) & (BITS_PER_BYTE - 1)))
63 
64 struct btrfs_root;
65 struct btrfs_inode;
66 struct btrfs_io_bio;
67 struct io_failure_record;
68 struct extent_io_tree;
69 
70 typedef blk_status_t (submit_bio_hook_t)(struct inode *inode, struct bio *bio,
71 					 int mirror_num,
72 					 unsigned long bio_flags);
73 
74 typedef blk_status_t (extent_submit_bio_start_t)(void *private_data,
75 		struct bio *bio, u64 bio_offset);
76 
77 #define INLINE_EXTENT_BUFFER_PAGES 16
78 #define MAX_INLINE_EXTENT_BUFFER_SIZE (INLINE_EXTENT_BUFFER_PAGES * PAGE_SIZE)
79 struct extent_buffer {
80 	u64 start;
81 	unsigned long len;
82 	unsigned long bflags;
83 	struct btrfs_fs_info *fs_info;
84 	spinlock_t refs_lock;
85 	atomic_t refs;
86 	atomic_t io_pages;
87 	int read_mirror;
88 	struct rcu_head rcu_head;
89 	pid_t lock_owner;
90 
91 	int blocking_writers;
92 	atomic_t blocking_readers;
93 	bool lock_recursed;
94 	/* >= 0 if eb belongs to a log tree, -1 otherwise */
95 	short log_index;
96 
97 	/* protects write locks */
98 	rwlock_t lock;
99 
100 	/* readers use lock_wq while they wait for the write
101 	 * lock holders to unlock
102 	 */
103 	wait_queue_head_t write_lock_wq;
104 
105 	/* writers use read_lock_wq while they wait for readers
106 	 * to unlock
107 	 */
108 	wait_queue_head_t read_lock_wq;
109 	struct page *pages[INLINE_EXTENT_BUFFER_PAGES];
110 #ifdef CONFIG_BTRFS_DEBUG
111 	int spinning_writers;
112 	atomic_t spinning_readers;
113 	atomic_t read_locks;
114 	int write_locks;
115 	struct list_head leak_list;
116 #endif
117 };
118 
119 /*
120  * Structure to record how many bytes and which ranges are set/cleared
121  */
122 struct extent_changeset {
123 	/* How many bytes are set/cleared in this operation */
124 	unsigned int bytes_changed;
125 
126 	/* Changed ranges */
127 	struct ulist range_changed;
128 };
129 
130 static inline void extent_changeset_init(struct extent_changeset *changeset)
131 {
132 	changeset->bytes_changed = 0;
133 	ulist_init(&changeset->range_changed);
134 }
135 
136 static inline struct extent_changeset *extent_changeset_alloc(void)
137 {
138 	struct extent_changeset *ret;
139 
140 	ret = kmalloc(sizeof(*ret), GFP_KERNEL);
141 	if (!ret)
142 		return NULL;
143 
144 	extent_changeset_init(ret);
145 	return ret;
146 }
147 
148 static inline void extent_changeset_release(struct extent_changeset *changeset)
149 {
150 	if (!changeset)
151 		return;
152 	changeset->bytes_changed = 0;
153 	ulist_release(&changeset->range_changed);
154 }
155 
156 static inline void extent_changeset_free(struct extent_changeset *changeset)
157 {
158 	if (!changeset)
159 		return;
160 	extent_changeset_release(changeset);
161 	kfree(changeset);
162 }
163 
164 static inline void extent_set_compress_type(unsigned long *bio_flags,
165 					    int compress_type)
166 {
167 	*bio_flags |= compress_type << EXTENT_BIO_FLAG_SHIFT;
168 }
169 
170 static inline int extent_compress_type(unsigned long bio_flags)
171 {
172 	return bio_flags >> EXTENT_BIO_FLAG_SHIFT;
173 }
174 
175 struct extent_map_tree;
176 
177 typedef struct extent_map *(get_extent_t)(struct btrfs_inode *inode,
178 					  struct page *page, size_t pg_offset,
179 					  u64 start, u64 len);
180 
181 int try_release_extent_mapping(struct page *page, gfp_t mask);
182 int try_release_extent_buffer(struct page *page);
183 
184 int __must_check submit_one_bio(struct bio *bio, int mirror_num,
185 				unsigned long bio_flags);
186 int btrfs_do_readpage(struct page *page, struct extent_map **em_cached,
187 		      struct bio **bio, unsigned long *bio_flags,
188 		      unsigned int read_flags, u64 *prev_em_start);
189 int extent_write_full_page(struct page *page, struct writeback_control *wbc);
190 int extent_write_locked_range(struct inode *inode, u64 start, u64 end,
191 			      int mode);
192 int extent_writepages(struct address_space *mapping,
193 		      struct writeback_control *wbc);
194 int btree_write_cache_pages(struct address_space *mapping,
195 			    struct writeback_control *wbc);
196 void extent_readahead(struct readahead_control *rac);
197 int extent_fiemap(struct btrfs_inode *inode, struct fiemap_extent_info *fieinfo,
198 		  u64 start, u64 len);
199 void set_page_extent_mapped(struct page *page);
200 
201 struct extent_buffer *alloc_extent_buffer(struct btrfs_fs_info *fs_info,
202 					  u64 start);
203 struct extent_buffer *__alloc_dummy_extent_buffer(struct btrfs_fs_info *fs_info,
204 						  u64 start, unsigned long len);
205 struct extent_buffer *alloc_dummy_extent_buffer(struct btrfs_fs_info *fs_info,
206 						u64 start);
207 struct extent_buffer *btrfs_clone_extent_buffer(const struct extent_buffer *src);
208 struct extent_buffer *find_extent_buffer(struct btrfs_fs_info *fs_info,
209 					 u64 start);
210 void free_extent_buffer(struct extent_buffer *eb);
211 void free_extent_buffer_stale(struct extent_buffer *eb);
212 #define WAIT_NONE	0
213 #define WAIT_COMPLETE	1
214 #define WAIT_PAGE_LOCK	2
215 int read_extent_buffer_pages(struct extent_buffer *eb, int wait,
216 			     int mirror_num);
217 void wait_on_extent_buffer_writeback(struct extent_buffer *eb);
218 
219 static inline int num_extent_pages(const struct extent_buffer *eb)
220 {
221 	return (round_up(eb->start + eb->len, PAGE_SIZE) >> PAGE_SHIFT) -
222 	       (eb->start >> PAGE_SHIFT);
223 }
224 
225 static inline int extent_buffer_uptodate(const struct extent_buffer *eb)
226 {
227 	return test_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
228 }
229 
230 int memcmp_extent_buffer(const struct extent_buffer *eb, const void *ptrv,
231 			 unsigned long start, unsigned long len);
232 void read_extent_buffer(const struct extent_buffer *eb, void *dst,
233 			unsigned long start,
234 			unsigned long len);
235 int read_extent_buffer_to_user_nofault(const struct extent_buffer *eb,
236 				       void __user *dst, unsigned long start,
237 				       unsigned long len);
238 void write_extent_buffer_fsid(const struct extent_buffer *eb, const void *src);
239 void write_extent_buffer_chunk_tree_uuid(const struct extent_buffer *eb,
240 		const void *src);
241 void write_extent_buffer(const struct extent_buffer *eb, const void *src,
242 			 unsigned long start, unsigned long len);
243 void copy_extent_buffer_full(const struct extent_buffer *dst,
244 			     const struct extent_buffer *src);
245 void copy_extent_buffer(const struct extent_buffer *dst,
246 			const struct extent_buffer *src,
247 			unsigned long dst_offset, unsigned long src_offset,
248 			unsigned long len);
249 void memcpy_extent_buffer(const struct extent_buffer *dst,
250 			  unsigned long dst_offset, unsigned long src_offset,
251 			  unsigned long len);
252 void memmove_extent_buffer(const struct extent_buffer *dst,
253 			   unsigned long dst_offset, unsigned long src_offset,
254 			   unsigned long len);
255 void memzero_extent_buffer(const struct extent_buffer *eb, unsigned long start,
256 			   unsigned long len);
257 int extent_buffer_test_bit(const struct extent_buffer *eb, unsigned long start,
258 			   unsigned long pos);
259 void extent_buffer_bitmap_set(const struct extent_buffer *eb, unsigned long start,
260 			      unsigned long pos, unsigned long len);
261 void extent_buffer_bitmap_clear(const struct extent_buffer *eb,
262 				unsigned long start, unsigned long pos,
263 				unsigned long len);
264 void clear_extent_buffer_dirty(const struct extent_buffer *eb);
265 bool set_extent_buffer_dirty(struct extent_buffer *eb);
266 void set_extent_buffer_uptodate(struct extent_buffer *eb);
267 void clear_extent_buffer_uptodate(struct extent_buffer *eb);
268 int extent_buffer_under_io(const struct extent_buffer *eb);
269 void extent_range_clear_dirty_for_io(struct inode *inode, u64 start, u64 end);
270 void extent_range_redirty_for_io(struct inode *inode, u64 start, u64 end);
271 void extent_clear_unlock_delalloc(struct btrfs_inode *inode, u64 start, u64 end,
272 				  struct page *locked_page,
273 				  unsigned bits_to_clear,
274 				  unsigned long page_ops);
275 struct bio *btrfs_bio_alloc(u64 first_byte);
276 struct bio *btrfs_io_bio_alloc(unsigned int nr_iovecs);
277 struct bio *btrfs_bio_clone(struct bio *bio);
278 struct bio *btrfs_bio_clone_partial(struct bio *orig, int offset, int size);
279 
280 struct btrfs_fs_info;
281 struct btrfs_inode;
282 
283 int repair_io_failure(struct btrfs_fs_info *fs_info, u64 ino, u64 start,
284 		      u64 length, u64 logical, struct page *page,
285 		      unsigned int pg_offset, int mirror_num);
286 void end_extent_writepage(struct page *page, int err, u64 start, u64 end);
287 int btrfs_repair_eb_io_failure(const struct extent_buffer *eb, int mirror_num);
288 
289 /*
290  * When IO fails, either with EIO or csum verification fails, we
291  * try other mirrors that might have a good copy of the data.  This
292  * io_failure_record is used to record state as we go through all the
293  * mirrors.  If another mirror has good data, the page is set up to date
294  * and things continue.  If a good mirror can't be found, the original
295  * bio end_io callback is called to indicate things have failed.
296  */
297 struct io_failure_record {
298 	struct page *page;
299 	u64 start;
300 	u64 len;
301 	u64 logical;
302 	unsigned long bio_flags;
303 	int this_mirror;
304 	int failed_mirror;
305 	int in_validation;
306 };
307 
308 
309 blk_status_t btrfs_submit_read_repair(struct inode *inode,
310 				      struct bio *failed_bio, u64 phy_offset,
311 				      struct page *page, unsigned int pgoff,
312 				      u64 start, u64 end, int failed_mirror,
313 				      submit_bio_hook_t *submit_bio_hook);
314 
315 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
316 bool find_lock_delalloc_range(struct inode *inode,
317 			     struct page *locked_page, u64 *start,
318 			     u64 *end);
319 #endif
320 struct extent_buffer *alloc_test_extent_buffer(struct btrfs_fs_info *fs_info,
321 					       u64 start);
322 
323 #ifdef CONFIG_BTRFS_DEBUG
324 void btrfs_extent_buffer_leak_debug_check(struct btrfs_fs_info *fs_info);
325 #else
326 #define btrfs_extent_buffer_leak_debug_check(fs_info)	do {} while (0)
327 #endif
328 
329 #endif
330