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
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
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 /* BTRFS_IOC_ENCODED_WRITE */
78 BTRFS_ORDERED_ENCODED,
79};
80
81/* BTRFS_ORDERED_* flags that specify the type of the extent. */
82#define BTRFS_ORDERED_TYPE_FLAGS ((1UL << BTRFS_ORDERED_REGULAR) | \
83 (1UL << BTRFS_ORDERED_NOCOW) | \
84 (1UL << BTRFS_ORDERED_PREALLOC) | \
85 (1UL << BTRFS_ORDERED_COMPRESSED) | \
86 (1UL << BTRFS_ORDERED_DIRECT) | \
87 (1UL << BTRFS_ORDERED_ENCODED))
88
89struct btrfs_ordered_extent {
90 /* logical offset in the file */
91 u64 file_offset;
92
93 /*
94 * These fields directly correspond to the same fields in
95 * btrfs_file_extent_item.
96 */
97 u64 num_bytes;
98 u64 ram_bytes;
99 u64 disk_bytenr;
100 u64 disk_num_bytes;
101 u64 offset;
102
103 /* number of bytes that still need writing */
104 u64 bytes_left;
105
106 /*
107 * the end of the ordered extent which is behind it but
108 * didn't update disk_i_size. Please see the comment of
109 * btrfs_ordered_update_i_size();
110 */
111 u64 outstanding_isize;
112
113 /*
114 * If we get truncated we need to adjust the file extent we enter for
115 * this ordered extent so that we do not expose stale data.
116 */
117 u64 truncated_len;
118
119 /* flags (described above) */
120 unsigned long flags;
121
122 /* compression algorithm */
123 int compress_type;
124
125 /* Qgroup reserved space */
126 int qgroup_rsv;
127
128 /* reference count */
129 refcount_t refs;
130
131 /* the inode we belong to */
132 struct inode *inode;
133
134 /* list of checksums for insertion when the extent io is done */
135 struct list_head list;
136
137 /* used for fast fsyncs */
138 struct list_head log_list;
139
140 /* used to wait for the BTRFS_ORDERED_COMPLETE bit */
141 wait_queue_head_t wait;
142
143 /* our friendly rbtree entry */
144 struct rb_node rb_node;
145
146 /* a per root list of all the pending ordered extents */
147 struct list_head root_extent_list;
148
149 struct btrfs_work work;
150
151 struct completion completion;
152 struct btrfs_work flush_work;
153 struct list_head work_list;
154
155 /*
156 * Used to reverse-map physical address returned from ZONE_APPEND write
157 * command in a workqueue context
158 */
159 u64 physical;
160 struct block_device *bdev;
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
171int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent);
172
173void btrfs_put_ordered_extent(struct btrfs_ordered_extent *entry);
174void btrfs_remove_ordered_extent(struct btrfs_inode *btrfs_inode,
175 struct btrfs_ordered_extent *entry);
176void btrfs_mark_ordered_io_finished(struct btrfs_inode *inode,
177 struct page *page, u64 file_offset,
178 u64 num_bytes, bool uptodate);
179bool btrfs_dec_test_ordered_pending(struct btrfs_inode *inode,
180 struct btrfs_ordered_extent **cached,
181 u64 file_offset, u64 io_size);
182int btrfs_add_ordered_extent(struct btrfs_inode *inode, u64 file_offset,
183 u64 num_bytes, u64 ram_bytes, u64 disk_bytenr,
184 u64 disk_num_bytes, u64 offset, unsigned flags,
185 int compress_type);
186void btrfs_add_ordered_sum(struct btrfs_ordered_extent *entry,
187 struct btrfs_ordered_sum *sum);
188struct btrfs_ordered_extent *btrfs_lookup_ordered_extent(struct btrfs_inode *inode,
189 u64 file_offset);
190void btrfs_start_ordered_extent(struct btrfs_ordered_extent *entry, int wait);
191int btrfs_wait_ordered_range(struct inode *inode, u64 start, u64 len);
192struct btrfs_ordered_extent *
193btrfs_lookup_first_ordered_extent(struct btrfs_inode *inode, u64 file_offset);
194struct btrfs_ordered_extent *btrfs_lookup_first_ordered_range(
195 struct btrfs_inode *inode, u64 file_offset, u64 len);
196struct btrfs_ordered_extent *btrfs_lookup_ordered_range(
197 struct btrfs_inode *inode,
198 u64 file_offset,
199 u64 len);
200void btrfs_get_ordered_extents_for_logging(struct btrfs_inode *inode,
201 struct list_head *list);
202u64 btrfs_wait_ordered_extents(struct btrfs_root *root, u64 nr,
203 const u64 range_start, const u64 range_len);
204void btrfs_wait_ordered_roots(struct btrfs_fs_info *fs_info, u64 nr,
205 const u64 range_start, const u64 range_len);
206void btrfs_lock_and_flush_ordered_range(struct btrfs_inode *inode, u64 start,
207 u64 end,
208 struct extent_state **cached_state);
209bool btrfs_try_lock_ordered_range(struct btrfs_inode *inode, u64 start, u64 end,
210 struct extent_state **cached_state);
211int btrfs_split_ordered_extent(struct btrfs_ordered_extent *ordered, u64 pre,
212 u64 post);
213int __init ordered_data_init(void);
214void __cold ordered_data_exit(void);
215
216#endif