/root/bitcoin/src/index/txindex.cpp
Line | Count | Source |
1 | | // Copyright (c) 2017-present The Bitcoin Core developers |
2 | | // Distributed under the MIT software license, see the accompanying |
3 | | // file COPYING or http://www.opensource.org/licenses/mit-license.php. |
4 | | |
5 | | #include <index/txindex.h> |
6 | | |
7 | | #include <chain.h> |
8 | | #include <common/args.h> |
9 | | #include <crypto/siphash.h> |
10 | | #include <dbwrapper.h> |
11 | | #include <flatfile.h> |
12 | | #include <index/base.h> |
13 | | #include <index/disktxpos.h> |
14 | | #include <index/txindex_key.h> |
15 | | #include <interfaces/chain.h> |
16 | | #include <node/blockstorage.h> |
17 | | #include <primitives/block.h> |
18 | | #include <primitives/transaction.h> |
19 | | #include <random.h> |
20 | | #include <serialize.h> |
21 | | #include <streams.h> |
22 | | #include <sync.h> |
23 | | #include <uint256.h> |
24 | | #include <util/fs.h> |
25 | | #include <util/log.h> |
26 | | #include <validation.h> |
27 | | |
28 | | #include <algorithm> |
29 | | #include <array> |
30 | | #include <cassert> |
31 | | #include <cstdint> |
32 | | #include <cstdio> |
33 | | #include <exception> |
34 | | #include <functional> |
35 | | #include <memory> |
36 | | #include <optional> |
37 | | #include <string> |
38 | | #include <utility> |
39 | | #include <vector> |
40 | | |
41 | | std::unique_ptr<TxIndex> g_txindex; |
42 | | |
43 | | namespace { |
44 | | SipHasher13UJ ReadOrCreateTxidHasher(CDBWrapper& db) |
45 | 0 | { |
46 | 0 | std::pair<uint64_t, uint64_t> salt; |
47 | 0 | if (!db.Read(txindex::DB_TXID_HASH_SALT, salt)) { Branch (47:9): [True: 0, False: 0]
|
48 | 0 | FastRandomContext rng{}; |
49 | 0 | salt = {rng.rand64(), rng.rand64()}; |
50 | 0 | db.Write(txindex::DB_TXID_HASH_SALT, salt, /*fSync=*/true); |
51 | 0 | } |
52 | 0 | return SipHasher13UJ{salt.first, salt.second}; |
53 | 0 | } |
54 | | } // namespace |
55 | | |
56 | | /** Access to the txindex database (indexes/txindex/) */ |
57 | | class TxIndex::DB : public BaseIndex::DB |
58 | | { |
59 | | public: |
60 | | explicit DB(size_t n_cache_size, bool f_memory = false, bool f_wipe = false); |
61 | | |
62 | | /// Write a block of transaction positions to the DB. |
63 | | void WriteTxs(const interfaces::BlockInfo& block); |
64 | | |
65 | | /// Used to hash the txid to compute the prefix. |
66 | | const SipHasher13UJ m_hasher; |
67 | | |
68 | | /// Whether the database contains any legacy ('t' + txid) entries. |
69 | | const bool m_has_legacy; |
70 | | |
71 | | CBlockLocator ReadBestBlock() const override; |
72 | | void WriteBestBlock(CDBBatch& batch, const CBlockLocator& locator) override; |
73 | | |
74 | | private: |
75 | | DB(size_t n_cache_size, bool f_memory, bool f_wipe, bool has_legacy); |
76 | | }; |
77 | | |
78 | 0 | static fs::path TxIndexDBPath() { return gArgs.GetDataDirNet() / "indexes" / "txindex"; } |
79 | | |
80 | | TxIndex::DB::DB(size_t n_cache_size, bool f_memory, bool f_wipe) : |
81 | | // Bloom filters are built for every key but only consulted by point reads, |
82 | | // which iterators bypass: the per-tx hashed ('x') lookups seek with an |
83 | | // iterator, and the 's'/'h' point reads are at most one per block against a |
84 | | // tiny keyspace. Only the legacy entries' per-tx point lookups benefit, so |
85 | | // enable the filters only for databases still containing them. |
86 | 0 | DB(n_cache_size, f_memory, f_wipe, |
87 | 0 | /*has_legacy=*/!f_memory && !f_wipe && CDBWrapper::HasKeyStartingWith(TxIndexDBPath(), txindex::DB_TXINDEX)) Branch (87:23): [True: 0, False: 0]
Branch (87:36): [True: 0, False: 0]
Branch (87:47): [True: 0, False: 0]
|
88 | 0 | {} |
89 | | |
90 | | TxIndex::DB::DB(size_t n_cache_size, bool f_memory, bool f_wipe, bool has_legacy) : |
91 | 0 | BaseIndex::DB(TxIndexDBPath(), n_cache_size, f_memory, f_wipe, /*f_obfuscate=*/false, /*f_bloom=*/has_legacy), |
92 | 0 | m_hasher{ReadOrCreateTxidHasher(*this)}, |
93 | 0 | m_has_legacy{has_legacy} |
94 | 0 | {} |
95 | | |
96 | | CBlockLocator TxIndex::DB::ReadBestBlock() const |
97 | 0 | { |
98 | 0 | CBlockLocator locator; |
99 | 0 | if (Read(txindex::DB_BEST_BLOCK_V2, locator)) { Branch (99:9): [True: 0, False: 0]
|
100 | 0 | return locator; |
101 | 0 | } |
102 | | // If we don't have a locator yet, start from the legacy best block. |
103 | 0 | return BaseIndex::DB::ReadBestBlock(); |
104 | 0 | } |
105 | | |
106 | | void TxIndex::DB::WriteBestBlock(CDBBatch& batch, const CBlockLocator& locator) |
107 | 0 | { |
108 | 0 | batch.Write(txindex::DB_BEST_BLOCK_V2, locator); |
109 | 0 | } |
110 | | |
111 | | void TxIndex::DB::WriteTxs(const interfaces::BlockInfo& block) |
112 | 0 | { |
113 | | // A block may be submitted again after it was already indexed, e.g. when it |
114 | | // reconnects after a reorg or is re-processed after an unclean shutdown. It |
115 | | // keeps its original sequence number, so skip it to avoid duplicate entries. |
116 | 0 | if (Exists(txindex::BlockHashKey{block.hash})) return; Branch (116:9): [True: 0, False: 0]
|
117 | | |
118 | 0 | uint32_t block_seq{0}; |
119 | 0 | Read(txindex::DB_NEXT_BLOCK_SEQ, block_seq); |
120 | |
|
121 | 0 | CDBBatch batch(*this); |
122 | 0 | batch.Write(txindex::BlockHashKey{block.hash}, block_seq); |
123 | 0 | batch.Write(txindex::BlockSeqKey{block_seq}, block.hash); |
124 | 0 | batch.Write(txindex::DB_NEXT_BLOCK_SEQ, block_seq + 1); |
125 | 0 | uint32_t tx_offset_in_block{txindex::BLOCK_HEADER_SIZE + GetSizeOfCompactSize(block.data->vtx.size())}; |
126 | 0 | for (const auto& tx : block.data->vtx) { Branch (126:25): [True: 0, False: 0]
|
127 | 0 | const txindex::DBKey key{txindex::CreateKeyPrefix(m_hasher, tx->GetHash()), |
128 | 0 | txindex::BlockTxPosition{block_seq, tx_offset_in_block}}; |
129 | 0 | batch.Write(key, txindex::EMPTY_VALUE); |
130 | 0 | tx_offset_in_block += tx->ComputeTotalSize(); |
131 | 0 | } |
132 | 0 | WriteBatch(batch); |
133 | 0 | } |
134 | | |
135 | | TxIndex::TxIndex(std::unique_ptr<interfaces::Chain> chain, size_t n_cache_size, bool f_memory, bool f_wipe) |
136 | 0 | : BaseIndex(std::move(chain), "txindex", "txidx"), m_db(std::make_unique<TxIndex::DB>(n_cache_size, f_memory, f_wipe)) |
137 | 0 | { |
138 | 0 | if (m_db->m_has_legacy) { Branch (138:9): [True: 0, False: 0]
|
139 | 0 | LogInfo("txindex contains entries in the legacy format, which uses excessive disk space. " |
140 | 0 | "To reclaim disk space, stop the node, delete %s and restart to rebuild the index.", |
141 | 0 | fs::PathToString(TxIndexDBPath())); |
142 | 0 | } |
143 | 0 | } |
144 | | |
145 | 0 | TxIndex::~TxIndex() = default; |
146 | | |
147 | | bool TxIndex::CustomAppend(const interfaces::BlockInfo& block) |
148 | 0 | { |
149 | | // Exclude genesis block transaction because outputs are not spendable. |
150 | 0 | if (block.height == 0) return true; Branch (150:9): [True: 0, False: 0]
|
151 | | |
152 | 0 | assert(block.data); Branch (152:5): [True: 0, False: 0]
|
153 | 0 | m_db->WriteTxs(block); |
154 | 0 | return true; |
155 | 0 | } |
156 | | |
157 | 0 | BaseIndex::DB& TxIndex::GetDB() const { return *m_db; } |
158 | | |
159 | | std::optional<TxIndexResult> TxIndex::FindTx(const Txid& tx_hash) const |
160 | 0 | { |
161 | 0 | struct Candidate { |
162 | 0 | FlatFilePos tx_position; |
163 | 0 | uint256 block_hash; |
164 | 0 | uint32_t block_seq; |
165 | | //! Whether this candidate's block is currently in the active chain. |
166 | | //! Active chain candidates are attempted first, so duplicate entries |
167 | | //! in both active and stale blocks will always return the active block hash. |
168 | 0 | bool in_active_chain; |
169 | 0 | }; |
170 | 0 | std::vector<Candidate> candidates; |
171 | 0 | { |
172 | 0 | std::unique_ptr<CDBIterator> it{m_db->NewIterator()}; |
173 | 0 | const txindex::TxHashKeyPrefix prefix{txindex::CreateKeyPrefix(m_db->m_hasher, tx_hash)}; |
174 | 0 | txindex::DBKey key{prefix, {}}; |
175 | 0 | for (it->Seek(key); it->Valid() && it->GetKey(key) && key.hash_prefix == prefix; it->Next()) { Branch (175:29): [True: 0, False: 0]
Branch (175:44): [True: 0, False: 0]
Branch (175:63): [True: 0, False: 0]
|
176 | 0 | uint256 candidate_block_hash; |
177 | 0 | if (!m_db->Read(txindex::BlockSeqKey{key.pos.block_seq}, candidate_block_hash)) { Branch (177:17): [True: 0, False: 0]
|
178 | 0 | LogWarning("Block sequence %u not found for txid %s", key.pos.block_seq, tx_hash.ToString()); |
179 | 0 | continue; |
180 | 0 | } |
181 | 0 | LOCK(cs_main); |
182 | 0 | const CBlockIndex* block_index{m_chainstate->m_blockman.LookupBlockIndex(candidate_block_hash)}; |
183 | 0 | if (!block_index) { Branch (183:17): [True: 0, False: 0]
|
184 | 0 | LogWarning("Block index entry %s not found for txid %s", candidate_block_hash.ToString(), tx_hash.ToString()); |
185 | 0 | continue; |
186 | 0 | } |
187 | 0 | if (!(block_index->nStatus & BLOCK_HAVE_DATA)) continue; Branch (187:17): [True: 0, False: 0]
|
188 | 0 | const FlatFilePos tx_position{block_index->nFile, block_index->nDataPos + key.pos.tx_offset_in_block}; |
189 | 0 | candidates.emplace_back(tx_position, candidate_block_hash, key.pos.block_seq, m_chainstate->m_chain.Contains(*block_index)); |
190 | 0 | } |
191 | 0 | } |
192 | | |
193 | | // Prefer active-chain matches, then later-connected blocks. |
194 | 0 | std::ranges::sort(candidates, std::greater{}, [](const Candidate& c) { |
195 | 0 | return std::pair{c.in_active_chain, c.block_seq}; |
196 | 0 | }); |
197 | |
|
198 | 0 | for (const auto& candidate : candidates) { Branch (198:32): [True: 0, False: 0]
|
199 | 0 | AutoFile file{m_chainstate->m_blockman.OpenBlockFile(candidate.tx_position, /*fReadOnly=*/true)}; |
200 | 0 | if (file.IsNull()) { Branch (200:13): [True: 0, False: 0]
|
201 | 0 | LogWarning("OpenBlockFile failed for txid %s", tx_hash.ToString()); |
202 | 0 | continue; |
203 | 0 | } |
204 | 0 | CTransactionRef tx; |
205 | 0 | try { |
206 | 0 | file >> TX_WITH_WITNESS(tx); |
207 | 0 | } catch (const std::exception& e) { |
208 | 0 | LogWarning("Deserialize or I/O error - %s", e.what()); |
209 | 0 | continue; |
210 | 0 | } |
211 | 0 | if (tx->GetHash() == tx_hash) { Branch (211:13): [True: 0, False: 0]
|
212 | 0 | return TxIndexResult{candidate.block_hash, std::move(tx)}; |
213 | 0 | } |
214 | 0 | } |
215 | | // Fall back to legacy if no hashed entry matched. This makes misses pay an |
216 | | // extra lookup, but keeps existing full-txid entries readable after upgrade. |
217 | 0 | return m_db->m_has_legacy ? FindLegacyTx(tx_hash) : std::nullopt; Branch (217:12): [True: 0, False: 0]
|
218 | 0 | } |
219 | | |
220 | | std::optional<TxIndexResult> TxIndex::FindLegacyTx(const Txid& tx_hash) const |
221 | 0 | { |
222 | 0 | CDiskTxPos postx; |
223 | 0 | if (!m_db->Read(txindex::LegacyTxKey(tx_hash), postx)) { Branch (223:9): [True: 0, False: 0]
|
224 | 0 | return std::nullopt; |
225 | 0 | } |
226 | | |
227 | 0 | AutoFile file{m_chainstate->m_blockman.OpenBlockFile(postx, /*fReadOnly=*/true)}; |
228 | 0 | if (file.IsNull()) { Branch (228:9): [True: 0, False: 0]
|
229 | 0 | LogError("OpenBlockFile failed"); |
230 | 0 | return std::nullopt; |
231 | 0 | } |
232 | 0 | CBlockHeader header; |
233 | 0 | CTransactionRef tx; |
234 | 0 | try { |
235 | 0 | file >> header; |
236 | 0 | file.seek(postx.nTxOffset, SEEK_CUR); |
237 | 0 | file >> TX_WITH_WITNESS(tx); |
238 | 0 | } catch (const std::exception& e) { |
239 | 0 | LogError("Deserialize or I/O error - %s", e.what()); |
240 | 0 | return std::nullopt; |
241 | 0 | } |
242 | 0 | if (tx->GetHash() != tx_hash) { Branch (242:9): [True: 0, False: 0]
|
243 | 0 | LogError("txid mismatch"); |
244 | 0 | return std::nullopt; |
245 | 0 | } |
246 | 0 | return TxIndexResult{header.GetHash(), std::move(tx)}; |
247 | 0 | } |