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
77#define BTRFS_ORDERED_IOERR 6 /* We had an io error when writing this out */
78
79#define BTRFS_ORDERED_UPDATED_ISIZE 7 /* indicates wether this ordered extent
80 * has done its due diligence in updating
81 * the isize. */
82
83struct btrfs_ordered_extent {
84 /* logical offset in the file */
85 u64 file_offset;
86
87 /* disk byte number */
88 u64 start;
89
90 /* ram length of the extent in bytes */
91 u64 len;
92
93 /* extent length on disk */
94 u64 disk_len;
95
96 /* number of bytes that still need writing */
97 u64 bytes_left;
98
99 /* flags (described above) */
100 unsigned long flags;
101
102 /* compression algorithm */
103 int compress_type;
104
105 /* reference count */
106 atomic_t refs;
107
108 /* the inode we belong to */
109 struct inode *inode;
110
111 /* list of checksums for insertion when the extent io is done */
112 struct list_head list;
113
114 /* used to wait for the BTRFS_ORDERED_COMPLETE bit */
115 wait_queue_head_t wait;
116
117 /* our friendly rbtree entry */
118 struct rb_node rb_node;
119
120 /* a per root list of all the pending ordered extents */
121 struct list_head root_extent_list;
122
123 struct btrfs_work work;
124};
125
126
127/*
128 * calculates the total size you need to allocate for an ordered sum
129 * structure spanning 'bytes' in the file
130 */
131static inline int btrfs_ordered_sum_size(struct btrfs_root *root,
132 unsigned long bytes)
133{
134 unsigned long num_sectors = (bytes + root->sectorsize - 1) /
135 root->sectorsize;
136 num_sectors++;
137 return sizeof(struct btrfs_ordered_sum) +
138 num_sectors * sizeof(struct btrfs_sector_sum);
139}
140
141static inline void
142btrfs_ordered_inode_tree_init(struct btrfs_ordered_inode_tree *t)
143{
144 spin_lock_init(&t->lock);
145 t->tree = RB_ROOT;
146 t->last = NULL;
147}
148
149void btrfs_put_ordered_extent(struct btrfs_ordered_extent *entry);
150void btrfs_remove_ordered_extent(struct inode *inode,
151 struct btrfs_ordered_extent *entry);
152int btrfs_dec_test_ordered_pending(struct inode *inode,
153 struct btrfs_ordered_extent **cached,
154 u64 file_offset, u64 io_size, int uptodate);
155int btrfs_dec_test_first_ordered_pending(struct inode *inode,
156 struct btrfs_ordered_extent **cached,
157 u64 *file_offset, u64 io_size,
158 int uptodate);
159int btrfs_add_ordered_extent(struct inode *inode, u64 file_offset,
160 u64 start, u64 len, u64 disk_len, int type);
161int btrfs_add_ordered_extent_dio(struct inode *inode, u64 file_offset,
162 u64 start, u64 len, u64 disk_len, int type);
163int btrfs_add_ordered_extent_compress(struct inode *inode, u64 file_offset,
164 u64 start, u64 len, u64 disk_len,
165 int type, int compress_type);
166void btrfs_add_ordered_sum(struct inode *inode,
167 struct btrfs_ordered_extent *entry,
168 struct btrfs_ordered_sum *sum);
169struct btrfs_ordered_extent *btrfs_lookup_ordered_extent(struct inode *inode,
170 u64 file_offset);
171void btrfs_start_ordered_extent(struct inode *inode,
172 struct btrfs_ordered_extent *entry, int wait);
173void btrfs_wait_ordered_range(struct inode *inode, u64 start, u64 len);
174struct btrfs_ordered_extent *
175btrfs_lookup_first_ordered_extent(struct inode * inode, u64 file_offset);
176struct btrfs_ordered_extent *btrfs_lookup_ordered_range(struct inode *inode,
177 u64 file_offset,
178 u64 len);
179int btrfs_ordered_update_i_size(struct inode *inode, u64 offset,
180 struct btrfs_ordered_extent *ordered);
181int btrfs_find_ordered_sum(struct inode *inode, u64 offset, u64 disk_bytenr, u32 *sum);
182void btrfs_run_ordered_operations(struct btrfs_root *root, int wait);
183void btrfs_add_ordered_operation(struct btrfs_trans_handle *trans,
184 struct btrfs_root *root,
185 struct inode *inode);
186void btrfs_wait_ordered_extents(struct btrfs_root *root,
187 int nocow_only, int delay_iput);
188#endif
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
29struct btrfs_ordered_sum {
30 /* bytenr is the start of this extent on disk */
31 u64 bytenr;
32
33 /*
34 * this is the length in bytes covered by the sums array below.
35 */
36 int len;
37 struct list_head list;
38 /* last field is a variable length array of csums */
39 u32 sums[];
40};
41
42/*
43 * bits for the flags field:
44 *
45 * BTRFS_ORDERED_IO_DONE is set when all of the blocks are written.
46 * It is used to make sure metadata is inserted into the tree only once
47 * per extent.
48 *
49 * BTRFS_ORDERED_COMPLETE is set when the extent is removed from the
50 * rbtree, just before waking any waiters. It is used to indicate the
51 * IO is done and any metadata is inserted into the tree.
52 */
53#define BTRFS_ORDERED_IO_DONE 0 /* set when all the pages are written */
54
55#define BTRFS_ORDERED_COMPLETE 1 /* set when removed from the tree */
56
57#define BTRFS_ORDERED_NOCOW 2 /* set when we want to write in place */
58
59#define BTRFS_ORDERED_COMPRESSED 3 /* writing a zlib compressed extent */
60
61#define BTRFS_ORDERED_PREALLOC 4 /* set when writing to prealloced extent */
62
63#define BTRFS_ORDERED_DIRECT 5 /* set when we're doing DIO with this extent */
64
65#define BTRFS_ORDERED_IOERR 6 /* We had an io error when writing this out */
66
67#define BTRFS_ORDERED_UPDATED_ISIZE 7 /* indicates whether this ordered extent
68 * has done its due diligence in updating
69 * the isize. */
70#define BTRFS_ORDERED_LOGGED_CSUM 8 /* We've logged the csums on this ordered
71 ordered extent */
72#define BTRFS_ORDERED_TRUNCATED 9 /* Set when we have to truncate an extent */
73
74#define BTRFS_ORDERED_LOGGED 10 /* Set when we've waited on this ordered extent
75 * in the logging code. */
76#define BTRFS_ORDERED_PENDING 11 /* We are waiting for this ordered extent to
77 * complete in the current transaction. */
78struct btrfs_ordered_extent {
79 /* logical offset in the file */
80 u64 file_offset;
81
82 /* disk byte number */
83 u64 start;
84
85 /* ram length of the extent in bytes */
86 u64 len;
87
88 /* extent length on disk */
89 u64 disk_len;
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 /* reference count */
114 atomic_t refs;
115
116 /* the inode we belong to */
117 struct inode *inode;
118
119 /* list of checksums for insertion when the extent io is done */
120 struct list_head list;
121
122 /* If we need to wait on this to be done */
123 struct list_head log_list;
124
125 /* If the transaction needs to wait on this ordered extent */
126 struct list_head trans_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/*
145 * calculates the total size you need to allocate for an ordered sum
146 * structure spanning 'bytes' in the file
147 */
148static inline int btrfs_ordered_sum_size(struct btrfs_root *root,
149 unsigned long bytes)
150{
151 int num_sectors = (int)DIV_ROUND_UP(bytes, root->sectorsize);
152 return sizeof(struct btrfs_ordered_sum) + num_sectors * sizeof(u32);
153}
154
155static inline void
156btrfs_ordered_inode_tree_init(struct btrfs_ordered_inode_tree *t)
157{
158 spin_lock_init(&t->lock);
159 t->tree = RB_ROOT;
160 t->last = NULL;
161}
162
163void btrfs_put_ordered_extent(struct btrfs_ordered_extent *entry);
164void btrfs_remove_ordered_extent(struct inode *inode,
165 struct btrfs_ordered_extent *entry);
166int btrfs_dec_test_ordered_pending(struct inode *inode,
167 struct btrfs_ordered_extent **cached,
168 u64 file_offset, u64 io_size, int uptodate);
169int btrfs_dec_test_first_ordered_pending(struct inode *inode,
170 struct btrfs_ordered_extent **cached,
171 u64 *file_offset, u64 io_size,
172 int uptodate);
173int btrfs_add_ordered_extent(struct inode *inode, u64 file_offset,
174 u64 start, u64 len, u64 disk_len, int type);
175int btrfs_add_ordered_extent_dio(struct inode *inode, u64 file_offset,
176 u64 start, u64 len, u64 disk_len, int type);
177int btrfs_add_ordered_extent_compress(struct inode *inode, u64 file_offset,
178 u64 start, u64 len, u64 disk_len,
179 int type, int compress_type);
180void btrfs_add_ordered_sum(struct inode *inode,
181 struct btrfs_ordered_extent *entry,
182 struct btrfs_ordered_sum *sum);
183struct btrfs_ordered_extent *btrfs_lookup_ordered_extent(struct inode *inode,
184 u64 file_offset);
185void btrfs_start_ordered_extent(struct inode *inode,
186 struct btrfs_ordered_extent *entry, int wait);
187int btrfs_wait_ordered_range(struct inode *inode, u64 start, u64 len);
188struct btrfs_ordered_extent *
189btrfs_lookup_first_ordered_extent(struct inode * inode, u64 file_offset);
190struct btrfs_ordered_extent *btrfs_lookup_ordered_range(struct inode *inode,
191 u64 file_offset,
192 u64 len);
193bool btrfs_have_ordered_extents_in_range(struct inode *inode,
194 u64 file_offset,
195 u64 len);
196int btrfs_ordered_update_i_size(struct inode *inode, u64 offset,
197 struct btrfs_ordered_extent *ordered);
198int btrfs_find_ordered_sum(struct inode *inode, u64 offset, u64 disk_bytenr,
199 u32 *sum, int len);
200int btrfs_wait_ordered_extents(struct btrfs_root *root, int nr);
201void btrfs_wait_ordered_roots(struct btrfs_fs_info *fs_info, int nr);
202void btrfs_get_logged_extents(struct inode *inode,
203 struct list_head *logged_list,
204 const loff_t start,
205 const loff_t end);
206void btrfs_put_logged_extents(struct list_head *logged_list);
207void btrfs_submit_logged_extents(struct list_head *logged_list,
208 struct btrfs_root *log);
209void btrfs_wait_logged_extents(struct btrfs_trans_handle *trans,
210 struct btrfs_root *log, u64 transid);
211void btrfs_free_logged_extents(struct btrfs_root *log, u64 transid);
212int __init ordered_data_init(void);
213void ordered_data_exit(void);
214#endif