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 
17 struct inode;
18 struct page;
19 struct extent_state;
20 struct btrfs_block_group;
21 struct btrfs_inode;
22 struct btrfs_root;
23 struct btrfs_fs_info;
24 
25 struct btrfs_ordered_sum {
26 	/*
27 	 * Logical start address and length for of the blocks covered by
28 	 * the sums array.
29 	 */
30 	u64 logical;
31 	u32 len;
32 
33 	struct list_head list;
34 	/* last field is a variable length array of csums */
35 	u8 sums[];
36 };
37 
38 /*
39  * Bits for btrfs_ordered_extent::flags.
40  *
41  * BTRFS_ORDERED_IO_DONE is set when all of the blocks are written.
42  * It is used to make sure metadata is inserted into the tree only once
43  * per extent.
44  *
45  * BTRFS_ORDERED_COMPLETE is set when the extent is removed from the
46  * rbtree, just before waking any waiters.  It is used to indicate the
47  * IO is done and any metadata is inserted into the tree.
48  */
49 enum {
50 	/*
51 	 * Different types for ordered extents, one and only one of the 4 types
52 	 * need to be set when creating ordered extent.
53 	 *
54 	 * REGULAR:	For regular non-compressed COW write
55 	 * NOCOW:	For NOCOW write into existing non-hole extent
56 	 * PREALLOC:	For NOCOW write into preallocated extent
57 	 * COMPRESSED:	For compressed COW write
58 	 */
59 	BTRFS_ORDERED_REGULAR,
60 	BTRFS_ORDERED_NOCOW,
61 	BTRFS_ORDERED_PREALLOC,
62 	BTRFS_ORDERED_COMPRESSED,
63 
64 	/*
65 	 * Extra bit for direct io, can only be set for
66 	 * REGULAR/NOCOW/PREALLOC. No direct io for compressed extent.
67 	 */
68 	BTRFS_ORDERED_DIRECT,
69 
70 	/* Extra status bits for ordered extents */
71 
72 	/* set when all the pages are written */
73 	BTRFS_ORDERED_IO_DONE,
74 	/* set when removed from the tree */
75 	BTRFS_ORDERED_COMPLETE,
76 	/* We had an io error when writing this out */
77 	BTRFS_ORDERED_IOERR,
78 	/* Set when we have to truncate an extent */
79 	BTRFS_ORDERED_TRUNCATED,
80 	/* Used during fsync to track already logged extents */
81 	BTRFS_ORDERED_LOGGED,
82 	/* We have already logged all the csums of the ordered extent */
83 	BTRFS_ORDERED_LOGGED_CSUM,
84 	/* We wait for this extent to complete in the current transaction */
85 	BTRFS_ORDERED_PENDING,
86 	/* BTRFS_IOC_ENCODED_WRITE */
87 	BTRFS_ORDERED_ENCODED,
88 };
89 
90 /* BTRFS_ORDERED_* flags that specify the type of the extent. */
91 #define BTRFS_ORDERED_TYPE_FLAGS ((1UL << BTRFS_ORDERED_REGULAR) |	\
92 				  (1UL << BTRFS_ORDERED_NOCOW) |	\
93 				  (1UL << BTRFS_ORDERED_PREALLOC) |	\
94 				  (1UL << BTRFS_ORDERED_COMPRESSED) |	\
95 				  (1UL << BTRFS_ORDERED_DIRECT) |	\
96 				  (1UL << BTRFS_ORDERED_ENCODED))
97 
98 struct btrfs_ordered_extent {
99 	/* logical offset in the file */
100 	u64 file_offset;
101 
102 	/*
103 	 * These fields directly correspond to the same fields in
104 	 * btrfs_file_extent_item.
105 	 */
106 	u64 num_bytes;
107 	u64 ram_bytes;
108 	u64 disk_bytenr;
109 	u64 disk_num_bytes;
110 	u64 offset;
111 
112 	/* number of bytes that still need writing */
113 	u64 bytes_left;
114 
115 	/*
116 	 * If we get truncated we need to adjust the file extent we enter for
117 	 * this ordered extent so that we do not expose stale data.
118 	 */
119 	u64 truncated_len;
120 
121 	/* flags (described above) */
122 	unsigned long flags;
123 
124 	/* compression algorithm */
125 	int compress_type;
126 
127 	/* Qgroup reserved space */
128 	int qgroup_rsv;
129 
130 	/* reference count */
131 	refcount_t refs;
132 
133 	/* the inode we belong to */
134 	struct btrfs_inode *inode;
135 
136 	/* list of checksums for insertion when the extent io is done */
137 	struct list_head list;
138 
139 	/* used for fast fsyncs */
140 	struct list_head log_list;
141 
142 	/* used to wait for the BTRFS_ORDERED_COMPLETE bit */
143 	wait_queue_head_t wait;
144 
145 	/* our friendly rbtree entry */
146 	struct rb_node rb_node;
147 
148 	/* a per root list of all the pending ordered extents */
149 	struct list_head root_extent_list;
150 
151 	struct btrfs_work work;
152 
153 	struct completion completion;
154 	struct btrfs_work flush_work;
155 	struct list_head work_list;
156 
157 	struct list_head bioc_list;
158 };
159 
160 int btrfs_finish_one_ordered(struct btrfs_ordered_extent *ordered_extent);
161 int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent);
162 
163 void btrfs_put_ordered_extent(struct btrfs_ordered_extent *entry);
164 void btrfs_remove_ordered_extent(struct btrfs_inode *btrfs_inode,
165 				struct btrfs_ordered_extent *entry);
166 void btrfs_finish_ordered_extent(struct btrfs_ordered_extent *ordered,
167 				 struct folio *folio, u64 file_offset, u64 len,
168 				 bool uptodate);
169 void btrfs_mark_ordered_io_finished(struct btrfs_inode *inode,
170 				    struct folio *folio, u64 file_offset,
171 				    u64 num_bytes, bool uptodate);
172 bool btrfs_dec_test_ordered_pending(struct btrfs_inode *inode,
173 				    struct btrfs_ordered_extent **cached,
174 				    u64 file_offset, u64 io_size);
175 
176 /*
177  * This represents details about the target file extent item of a write operation.
178  */
179 struct btrfs_file_extent {
180 	u64 disk_bytenr;
181 	u64 disk_num_bytes;
182 	u64 num_bytes;
183 	u64 ram_bytes;
184 	u64 offset;
185 	u8 compression;
186 };
187 
188 struct btrfs_ordered_extent *btrfs_alloc_ordered_extent(
189 			struct btrfs_inode *inode, u64 file_offset,
190 			const struct btrfs_file_extent *file_extent, unsigned long flags);
191 void btrfs_add_ordered_sum(struct btrfs_ordered_extent *entry,
192 			   struct btrfs_ordered_sum *sum);
193 struct btrfs_ordered_extent *btrfs_lookup_ordered_extent(struct btrfs_inode *inode,
194 							 u64 file_offset);
195 void btrfs_start_ordered_extent_nowriteback(struct btrfs_ordered_extent *entry,
196 				u64 nowriteback_start, u32 nowriteback_len);
197 static inline void btrfs_start_ordered_extent(struct btrfs_ordered_extent *entry)
198 {
199 	return btrfs_start_ordered_extent_nowriteback(entry, 0, 0);
200 }
201 
202 int btrfs_wait_ordered_range(struct btrfs_inode *inode, u64 start, u64 len);
203 struct btrfs_ordered_extent *
204 btrfs_lookup_first_ordered_extent(struct btrfs_inode *inode, u64 file_offset);
205 struct btrfs_ordered_extent *btrfs_lookup_first_ordered_range(
206 			struct btrfs_inode *inode, u64 file_offset, u64 len);
207 struct btrfs_ordered_extent *btrfs_lookup_ordered_range(
208 		struct btrfs_inode *inode,
209 		u64 file_offset,
210 		u64 len);
211 void btrfs_get_ordered_extents_for_logging(struct btrfs_inode *inode,
212 					   struct list_head *list);
213 u64 btrfs_wait_ordered_extents(struct btrfs_root *root, u64 nr,
214 			       const struct btrfs_block_group *bg);
215 void btrfs_wait_ordered_roots(struct btrfs_fs_info *fs_info, u64 nr,
216 			      const struct btrfs_block_group *bg);
217 void btrfs_lock_and_flush_ordered_range(struct btrfs_inode *inode, u64 start,
218 					u64 end,
219 					struct extent_state **cached_state);
220 bool btrfs_try_lock_ordered_range(struct btrfs_inode *inode, u64 start, u64 end,
221 				  struct extent_state **cached_state);
222 struct btrfs_ordered_extent *btrfs_split_ordered_extent(
223 			struct btrfs_ordered_extent *ordered, u64 len);
224 void btrfs_mark_ordered_extent_error(struct btrfs_ordered_extent *ordered);
225 int __init ordered_data_init(void);
226 void __cold ordered_data_exit(void);
227 
228 #endif
229