Loading...
1/*
2 * Copyright (C) 2007 Oracle. All rights reserved.
3 *
4 * This program is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU General Public
6 * License v2 as published by the Free Software Foundation.
7 *
8 * This program is distributed in the hope that it will be useful,
9 * but WITHOUT ANY WARRANTY; without even the implied warranty of
10 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
11 * General Public License for more details.
12 *
13 * You should have received a copy of the GNU General Public
14 * License along with this program; if not, write to the
15 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
16 * Boston, MA 021110-1307, USA.
17 */
18
19#ifndef __BTRFS_ORDERED_DATA__
20#define __BTRFS_ORDERED_DATA__
21
22/* one of these per inode */
23struct btrfs_ordered_inode_tree {
24 spinlock_t lock;
25 struct rb_root tree;
26 struct rb_node *last;
27};
28
29/*
30 * these are used to collect checksums done just before bios submission.
31 * They are attached via a list into the ordered extent, and
32 * checksum items are inserted into the tree after all the blocks in
33 * the ordered extent are on disk
34 */
35struct btrfs_sector_sum {
36 /* bytenr on disk */
37 u64 bytenr;
38 u32 sum;
39};
40
41struct btrfs_ordered_sum {
42 /* bytenr is the start of this extent on disk */
43 u64 bytenr;
44
45 /*
46 * this is the length in bytes covered by the sums array below.
47 */
48 unsigned long len;
49 struct list_head list;
50 /* last field is a variable length array of btrfs_sector_sums */
51 struct btrfs_sector_sum sums[];
52};
53
54/*
55 * bits for the flags field:
56 *
57 * BTRFS_ORDERED_IO_DONE is set when all of the blocks are written.
58 * It is used to make sure metadata is inserted into the tree only once
59 * per extent.
60 *
61 * BTRFS_ORDERED_COMPLETE is set when the extent is removed from the
62 * rbtree, just before waking any waiters. It is used to indicate the
63 * IO is done and any metadata is inserted into the tree.
64 */
65#define BTRFS_ORDERED_IO_DONE 0 /* set when all the pages are written */
66
67#define BTRFS_ORDERED_COMPLETE 1 /* set when removed from the tree */
68
69#define BTRFS_ORDERED_NOCOW 2 /* set when we want to write in place */
70
71#define BTRFS_ORDERED_COMPRESSED 3 /* writing a zlib compressed extent */
72
73#define BTRFS_ORDERED_PREALLOC 4 /* set when writing to prealloced extent */
74
75#define BTRFS_ORDERED_DIRECT 5 /* set when we're doing DIO with this extent */
76
77struct btrfs_ordered_extent {
78 /* logical offset in the file */
79 u64 file_offset;
80
81 /* disk byte number */
82 u64 start;
83
84 /* ram length of the extent in bytes */
85 u64 len;
86
87 /* extent length on disk */
88 u64 disk_len;
89
90 /* number of bytes that still need writing */
91 u64 bytes_left;
92
93 /* flags (described above) */
94 unsigned long flags;
95
96 /* compression algorithm */
97 int compress_type;
98
99 /* reference count */
100 atomic_t refs;
101
102 /* the inode we belong to */
103 struct inode *inode;
104
105 /* list of checksums for insertion when the extent io is done */
106 struct list_head list;
107
108 /* used to wait for the BTRFS_ORDERED_COMPLETE bit */
109 wait_queue_head_t wait;
110
111 /* our friendly rbtree entry */
112 struct rb_node rb_node;
113
114 /* a per root list of all the pending ordered extents */
115 struct list_head root_extent_list;
116};
117
118
119/*
120 * calculates the total size you need to allocate for an ordered sum
121 * structure spanning 'bytes' in the file
122 */
123static inline int btrfs_ordered_sum_size(struct btrfs_root *root,
124 unsigned long bytes)
125{
126 unsigned long num_sectors = (bytes + root->sectorsize - 1) /
127 root->sectorsize;
128 num_sectors++;
129 return sizeof(struct btrfs_ordered_sum) +
130 num_sectors * sizeof(struct btrfs_sector_sum);
131}
132
133static inline void
134btrfs_ordered_inode_tree_init(struct btrfs_ordered_inode_tree *t)
135{
136 spin_lock_init(&t->lock);
137 t->tree = RB_ROOT;
138 t->last = NULL;
139}
140
141int btrfs_put_ordered_extent(struct btrfs_ordered_extent *entry);
142int btrfs_remove_ordered_extent(struct inode *inode,
143 struct btrfs_ordered_extent *entry);
144int btrfs_dec_test_ordered_pending(struct inode *inode,
145 struct btrfs_ordered_extent **cached,
146 u64 file_offset, u64 io_size);
147int btrfs_dec_test_first_ordered_pending(struct inode *inode,
148 struct btrfs_ordered_extent **cached,
149 u64 *file_offset, u64 io_size);
150int btrfs_add_ordered_extent(struct inode *inode, u64 file_offset,
151 u64 start, u64 len, u64 disk_len, int type);
152int btrfs_add_ordered_extent_dio(struct inode *inode, u64 file_offset,
153 u64 start, u64 len, u64 disk_len, int type);
154int btrfs_add_ordered_extent_compress(struct inode *inode, u64 file_offset,
155 u64 start, u64 len, u64 disk_len,
156 int type, int compress_type);
157int btrfs_add_ordered_sum(struct inode *inode,
158 struct btrfs_ordered_extent *entry,
159 struct btrfs_ordered_sum *sum);
160struct btrfs_ordered_extent *btrfs_lookup_ordered_extent(struct inode *inode,
161 u64 file_offset);
162void btrfs_start_ordered_extent(struct inode *inode,
163 struct btrfs_ordered_extent *entry, int wait);
164int btrfs_wait_ordered_range(struct inode *inode, u64 start, u64 len);
165struct btrfs_ordered_extent *
166btrfs_lookup_first_ordered_extent(struct inode * inode, u64 file_offset);
167struct btrfs_ordered_extent *btrfs_lookup_ordered_range(struct inode *inode,
168 u64 file_offset,
169 u64 len);
170int btrfs_ordered_update_i_size(struct inode *inode, u64 offset,
171 struct btrfs_ordered_extent *ordered);
172int btrfs_find_ordered_sum(struct inode *inode, u64 offset, u64 disk_bytenr, u32 *sum);
173int btrfs_run_ordered_operations(struct btrfs_root *root, int wait);
174int btrfs_add_ordered_operation(struct btrfs_trans_handle *trans,
175 struct btrfs_root *root,
176 struct inode *inode);
177int btrfs_wait_ordered_extents(struct btrfs_root *root,
178 int nocow_only, int delay_iput);
179#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