Loading...
1/* SPDX-License-Identifier: GPL-2.0 */
2/*
3 * Copyright (C) 2007 Oracle. All rights reserved.
4 */
5
6#ifndef BTRFS_ORDERED_DATA_H
7#define BTRFS_ORDERED_DATA_H
8
9#include <linux/types.h>
10#include <linux/list.h>
11#include <linux/refcount.h>
12#include <linux/completion.h>
13#include <linux/rbtree.h>
14#include <linux/wait.h>
15#include "async-thread.h"
16
17struct inode;
18struct page;
19struct extent_state;
20struct btrfs_inode;
21struct btrfs_root;
22struct btrfs_fs_info;
23
24struct btrfs_ordered_sum {
25 /*
26 * Logical start address and length for of the blocks covered by
27 * the sums array.
28 */
29 u64 logical;
30 u32 len;
31
32 struct list_head list;
33 /* last field is a variable length array of csums */
34 u8 sums[];
35};
36
37/*
38 * Bits for btrfs_ordered_extent::flags.
39 *
40 * BTRFS_ORDERED_IO_DONE is set when all of the blocks are written.
41 * It is used to make sure metadata is inserted into the tree only once
42 * per extent.
43 *
44 * BTRFS_ORDERED_COMPLETE is set when the extent is removed from the
45 * rbtree, just before waking any waiters. It is used to indicate the
46 * IO is done and any metadata is inserted into the tree.
47 */
48enum {
49 /*
50 * Different types for ordered extents, one and only one of the 4 types
51 * need to be set when creating ordered extent.
52 *
53 * REGULAR: For regular non-compressed COW write
54 * NOCOW: For NOCOW write into existing non-hole extent
55 * PREALLOC: For NOCOW write into preallocated extent
56 * COMPRESSED: For compressed COW write
57 */
58 BTRFS_ORDERED_REGULAR,
59 BTRFS_ORDERED_NOCOW,
60 BTRFS_ORDERED_PREALLOC,
61 BTRFS_ORDERED_COMPRESSED,
62
63 /*
64 * Extra bit for direct io, can only be set for
65 * REGULAR/NOCOW/PREALLOC. No direct io for compressed extent.
66 */
67 BTRFS_ORDERED_DIRECT,
68
69 /* Extra status bits for ordered extents */
70
71 /* set when all the pages are written */
72 BTRFS_ORDERED_IO_DONE,
73 /* set when removed from the tree */
74 BTRFS_ORDERED_COMPLETE,
75 /* We had an io error when writing this out */
76 BTRFS_ORDERED_IOERR,
77 /* Set when we have to truncate an extent */
78 BTRFS_ORDERED_TRUNCATED,
79 /* Used during fsync to track already logged extents */
80 BTRFS_ORDERED_LOGGED,
81 /* We have already logged all the csums of the ordered extent */
82 BTRFS_ORDERED_LOGGED_CSUM,
83 /* We wait for this extent to complete in the current transaction */
84 BTRFS_ORDERED_PENDING,
85 /* BTRFS_IOC_ENCODED_WRITE */
86 BTRFS_ORDERED_ENCODED,
87};
88
89/* BTRFS_ORDERED_* flags that specify the type of the extent. */
90#define BTRFS_ORDERED_TYPE_FLAGS ((1UL << BTRFS_ORDERED_REGULAR) | \
91 (1UL << BTRFS_ORDERED_NOCOW) | \
92 (1UL << BTRFS_ORDERED_PREALLOC) | \
93 (1UL << BTRFS_ORDERED_COMPRESSED) | \
94 (1UL << BTRFS_ORDERED_DIRECT) | \
95 (1UL << BTRFS_ORDERED_ENCODED))
96
97struct btrfs_ordered_extent {
98 /* logical offset in the file */
99 u64 file_offset;
100
101 /*
102 * These fields directly correspond to the same fields in
103 * btrfs_file_extent_item.
104 */
105 u64 num_bytes;
106 u64 ram_bytes;
107 u64 disk_bytenr;
108 u64 disk_num_bytes;
109 u64 offset;
110
111 /* number of bytes that still need writing */
112 u64 bytes_left;
113
114 /*
115 * If we get truncated we need to adjust the file extent we enter for
116 * this ordered extent so that we do not expose stale data.
117 */
118 u64 truncated_len;
119
120 /* flags (described above) */
121 unsigned long flags;
122
123 /* compression algorithm */
124 int compress_type;
125
126 /* Qgroup reserved space */
127 int qgroup_rsv;
128
129 /* reference count */
130 refcount_t refs;
131
132 /* the inode we belong to */
133 struct btrfs_inode *inode;
134
135 /* list of checksums for insertion when the extent io is done */
136 struct list_head list;
137
138 /* used for fast fsyncs */
139 struct list_head log_list;
140
141 /* used to wait for the BTRFS_ORDERED_COMPLETE bit */
142 wait_queue_head_t wait;
143
144 /* our friendly rbtree entry */
145 struct rb_node rb_node;
146
147 /* a per root list of all the pending ordered extents */
148 struct list_head root_extent_list;
149
150 struct btrfs_work work;
151
152 struct completion completion;
153 struct btrfs_work flush_work;
154 struct list_head work_list;
155
156 struct list_head bioc_list;
157};
158
159int btrfs_finish_one_ordered(struct btrfs_ordered_extent *ordered_extent);
160int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent);
161
162void btrfs_put_ordered_extent(struct btrfs_ordered_extent *entry);
163void btrfs_remove_ordered_extent(struct btrfs_inode *btrfs_inode,
164 struct btrfs_ordered_extent *entry);
165void btrfs_finish_ordered_extent(struct btrfs_ordered_extent *ordered,
166 struct folio *folio, u64 file_offset, u64 len,
167 bool uptodate);
168void btrfs_mark_ordered_io_finished(struct btrfs_inode *inode,
169 struct folio *folio, u64 file_offset,
170 u64 num_bytes, bool uptodate);
171bool btrfs_dec_test_ordered_pending(struct btrfs_inode *inode,
172 struct btrfs_ordered_extent **cached,
173 u64 file_offset, u64 io_size);
174
175/*
176 * This represents details about the target file extent item of a write operation.
177 */
178struct btrfs_file_extent {
179 u64 disk_bytenr;
180 u64 disk_num_bytes;
181 u64 num_bytes;
182 u64 ram_bytes;
183 u64 offset;
184 u8 compression;
185};
186
187struct btrfs_ordered_extent *btrfs_alloc_ordered_extent(
188 struct btrfs_inode *inode, u64 file_offset,
189 const struct btrfs_file_extent *file_extent, unsigned long flags);
190void btrfs_add_ordered_sum(struct btrfs_ordered_extent *entry,
191 struct btrfs_ordered_sum *sum);
192struct btrfs_ordered_extent *btrfs_lookup_ordered_extent(struct btrfs_inode *inode,
193 u64 file_offset);
194void btrfs_start_ordered_extent(struct btrfs_ordered_extent *entry);
195int btrfs_wait_ordered_range(struct btrfs_inode *inode, u64 start, u64 len);
196struct btrfs_ordered_extent *
197btrfs_lookup_first_ordered_extent(struct btrfs_inode *inode, u64 file_offset);
198struct btrfs_ordered_extent *btrfs_lookup_first_ordered_range(
199 struct btrfs_inode *inode, u64 file_offset, u64 len);
200struct btrfs_ordered_extent *btrfs_lookup_ordered_range(
201 struct btrfs_inode *inode,
202 u64 file_offset,
203 u64 len);
204void btrfs_get_ordered_extents_for_logging(struct btrfs_inode *inode,
205 struct list_head *list);
206u64 btrfs_wait_ordered_extents(struct btrfs_root *root, u64 nr,
207 const struct btrfs_block_group *bg);
208void btrfs_wait_ordered_roots(struct btrfs_fs_info *fs_info, u64 nr,
209 const struct btrfs_block_group *bg);
210void btrfs_lock_and_flush_ordered_range(struct btrfs_inode *inode, u64 start,
211 u64 end,
212 struct extent_state **cached_state);
213bool btrfs_try_lock_ordered_range(struct btrfs_inode *inode, u64 start, u64 end,
214 struct extent_state **cached_state);
215struct btrfs_ordered_extent *btrfs_split_ordered_extent(
216 struct btrfs_ordered_extent *ordered, u64 len);
217void btrfs_mark_ordered_extent_error(struct btrfs_ordered_extent *ordered);
218int __init ordered_data_init(void);
219void __cold ordered_data_exit(void);
220
221#endif
1/* SPDX-License-Identifier: GPL-2.0 */
2/*
3 * Copyright (C) 2007 Oracle. All rights reserved.
4 */
5
6#ifndef BTRFS_ORDERED_DATA_H
7#define BTRFS_ORDERED_DATA_H
8
9/* one of these per inode */
10struct btrfs_ordered_inode_tree {
11 spinlock_t lock;
12 struct rb_root tree;
13 struct rb_node *last;
14};
15
16struct btrfs_ordered_sum {
17 /* bytenr is the start of this extent on disk */
18 u64 bytenr;
19
20 /*
21 * this is the length in bytes covered by the sums array below.
22 */
23 int len;
24 struct list_head list;
25 /* last field is a variable length array of csums */
26 u8 sums[];
27};
28
29/*
30 * Bits for btrfs_ordered_extent::flags.
31 *
32 * BTRFS_ORDERED_IO_DONE is set when all of the blocks are written.
33 * It is used to make sure metadata is inserted into the tree only once
34 * per extent.
35 *
36 * BTRFS_ORDERED_COMPLETE is set when the extent is removed from the
37 * rbtree, just before waking any waiters. It is used to indicate the
38 * IO is done and any metadata is inserted into the tree.
39 */
40enum {
41 /*
42 * Different types for ordered extents, one and only one of the 4 types
43 * need to be set when creating ordered extent.
44 *
45 * REGULAR: For regular non-compressed COW write
46 * NOCOW: For NOCOW write into existing non-hole extent
47 * PREALLOC: For NOCOW write into preallocated extent
48 * COMPRESSED: For compressed COW write
49 */
50 BTRFS_ORDERED_REGULAR,
51 BTRFS_ORDERED_NOCOW,
52 BTRFS_ORDERED_PREALLOC,
53 BTRFS_ORDERED_COMPRESSED,
54
55 /*
56 * Extra bit for direct io, can only be set for
57 * REGULAR/NOCOW/PREALLOC. No direct io for compressed extent.
58 */
59 BTRFS_ORDERED_DIRECT,
60
61 /* Extra status bits for ordered extents */
62
63 /* set when all the pages are written */
64 BTRFS_ORDERED_IO_DONE,
65 /* set when removed from the tree */
66 BTRFS_ORDERED_COMPLETE,
67 /* We had an io error when writing this out */
68 BTRFS_ORDERED_IOERR,
69 /* Set when we have to truncate an extent */
70 BTRFS_ORDERED_TRUNCATED,
71 /* Used during fsync to track already logged extents */
72 BTRFS_ORDERED_LOGGED,
73 /* We have already logged all the csums of the ordered extent */
74 BTRFS_ORDERED_LOGGED_CSUM,
75 /* We wait for this extent to complete in the current transaction */
76 BTRFS_ORDERED_PENDING,
77};
78
79struct btrfs_ordered_extent {
80 /* logical offset in the file */
81 u64 file_offset;
82
83 /*
84 * These fields directly correspond to the same fields in
85 * btrfs_file_extent_item.
86 */
87 u64 disk_bytenr;
88 u64 num_bytes;
89 u64 disk_num_bytes;
90
91 /* number of bytes that still need writing */
92 u64 bytes_left;
93
94 /*
95 * the end of the ordered extent which is behind it but
96 * didn't update disk_i_size. Please see the comment of
97 * btrfs_ordered_update_i_size();
98 */
99 u64 outstanding_isize;
100
101 /*
102 * If we get truncated we need to adjust the file extent we enter for
103 * this ordered extent so that we do not expose stale data.
104 */
105 u64 truncated_len;
106
107 /* flags (described above) */
108 unsigned long flags;
109
110 /* compression algorithm */
111 int compress_type;
112
113 /* Qgroup reserved space */
114 int qgroup_rsv;
115
116 /* reference count */
117 refcount_t refs;
118
119 /* the inode we belong to */
120 struct inode *inode;
121
122 /* list of checksums for insertion when the extent io is done */
123 struct list_head list;
124
125 /* used for fast fsyncs */
126 struct list_head log_list;
127
128 /* used to wait for the BTRFS_ORDERED_COMPLETE bit */
129 wait_queue_head_t wait;
130
131 /* our friendly rbtree entry */
132 struct rb_node rb_node;
133
134 /* a per root list of all the pending ordered extents */
135 struct list_head root_extent_list;
136
137 struct btrfs_work work;
138
139 struct completion completion;
140 struct btrfs_work flush_work;
141 struct list_head work_list;
142
143 /*
144 * Used to reverse-map physical address returned from ZONE_APPEND write
145 * command in a workqueue context
146 */
147 u64 physical;
148 struct block_device *bdev;
149};
150
151/*
152 * calculates the total size you need to allocate for an ordered sum
153 * structure spanning 'bytes' in the file
154 */
155static inline int btrfs_ordered_sum_size(struct btrfs_fs_info *fs_info,
156 unsigned long bytes)
157{
158 int num_sectors = (int)DIV_ROUND_UP(bytes, fs_info->sectorsize);
159
160 return sizeof(struct btrfs_ordered_sum) + num_sectors * fs_info->csum_size;
161}
162
163static inline void
164btrfs_ordered_inode_tree_init(struct btrfs_ordered_inode_tree *t)
165{
166 spin_lock_init(&t->lock);
167 t->tree = RB_ROOT;
168 t->last = NULL;
169}
170
171void btrfs_put_ordered_extent(struct btrfs_ordered_extent *entry);
172void btrfs_remove_ordered_extent(struct btrfs_inode *btrfs_inode,
173 struct btrfs_ordered_extent *entry);
174void btrfs_mark_ordered_io_finished(struct btrfs_inode *inode,
175 struct page *page, u64 file_offset,
176 u64 num_bytes, btrfs_func_t finish_func,
177 bool uptodate);
178bool btrfs_dec_test_ordered_pending(struct btrfs_inode *inode,
179 struct btrfs_ordered_extent **cached,
180 u64 file_offset, u64 io_size, int uptodate);
181int btrfs_add_ordered_extent(struct btrfs_inode *inode, u64 file_offset,
182 u64 disk_bytenr, u64 num_bytes, u64 disk_num_bytes,
183 int type);
184int btrfs_add_ordered_extent_dio(struct btrfs_inode *inode, u64 file_offset,
185 u64 disk_bytenr, u64 num_bytes,
186 u64 disk_num_bytes, int type);
187int btrfs_add_ordered_extent_compress(struct btrfs_inode *inode, u64 file_offset,
188 u64 disk_bytenr, u64 num_bytes,
189 u64 disk_num_bytes, int compress_type);
190void btrfs_add_ordered_sum(struct btrfs_ordered_extent *entry,
191 struct btrfs_ordered_sum *sum);
192struct btrfs_ordered_extent *btrfs_lookup_ordered_extent(struct btrfs_inode *inode,
193 u64 file_offset);
194void btrfs_start_ordered_extent(struct btrfs_ordered_extent *entry, int wait);
195int btrfs_wait_ordered_range(struct inode *inode, u64 start, u64 len);
196struct btrfs_ordered_extent *
197btrfs_lookup_first_ordered_extent(struct btrfs_inode *inode, u64 file_offset);
198struct btrfs_ordered_extent *btrfs_lookup_first_ordered_range(
199 struct btrfs_inode *inode, u64 file_offset, u64 len);
200struct btrfs_ordered_extent *btrfs_lookup_ordered_range(
201 struct btrfs_inode *inode,
202 u64 file_offset,
203 u64 len);
204void btrfs_get_ordered_extents_for_logging(struct btrfs_inode *inode,
205 struct list_head *list);
206u64 btrfs_wait_ordered_extents(struct btrfs_root *root, u64 nr,
207 const u64 range_start, const u64 range_len);
208void btrfs_wait_ordered_roots(struct btrfs_fs_info *fs_info, u64 nr,
209 const u64 range_start, const u64 range_len);
210void btrfs_lock_and_flush_ordered_range(struct btrfs_inode *inode, u64 start,
211 u64 end,
212 struct extent_state **cached_state);
213int btrfs_split_ordered_extent(struct btrfs_ordered_extent *ordered, u64 pre,
214 u64 post);
215int __init ordered_data_init(void);
216void __cold ordered_data_exit(void);
217
218#endif