blockstorage.h
1 // Copyright (c) 2011-2022 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 #ifndef BITCOIN_NODE_BLOCKSTORAGE_H 6 #define BITCOIN_NODE_BLOCKSTORAGE_H 7 8 #include <attributes.h> 9 #include <chain.h> 10 #include <dbwrapper.h> 11 #include <flatfile.h> 12 #include <kernel/blockmanager_opts.h> 13 #include <kernel/chainparams.h> 14 #include <kernel/cs_main.h> 15 #include <kernel/messagestartchars.h> 16 #include <primitives/block.h> 17 #include <streams.h> 18 #include <sync.h> 19 #include <uint256.h> 20 #include <util/fs.h> 21 #include <util/hasher.h> 22 23 #include <array> 24 #include <atomic> 25 #include <cstdint> 26 #include <functional> 27 #include <limits> 28 #include <map> 29 #include <memory> 30 #include <optional> 31 #include <set> 32 #include <span> 33 #include <string> 34 #include <unordered_map> 35 #include <utility> 36 #include <vector> 37 38 class BlockValidationState; 39 class CBlockUndo; 40 class Chainstate; 41 class ChainstateManager; 42 namespace Consensus { 43 struct Params; 44 } 45 namespace util { 46 class SignalInterrupt; 47 } // namespace util 48 49 namespace kernel { 50 class CBlockFileInfo 51 { 52 public: 53 uint32_t nBlocks{}; //!< number of blocks stored in file 54 uint32_t nSize{}; //!< number of used bytes of block file 55 uint32_t nUndoSize{}; //!< number of used bytes in the undo file 56 uint32_t nHeightFirst{}; //!< lowest height of block in file 57 uint32_t nHeightLast{}; //!< highest height of block in file 58 uint64_t nTimeFirst{}; //!< earliest time of block in file 59 uint64_t nTimeLast{}; //!< latest time of block in file 60 61 SERIALIZE_METHODS(CBlockFileInfo, obj) 62 { 63 READWRITE(VARINT(obj.nBlocks)); 64 READWRITE(VARINT(obj.nSize)); 65 READWRITE(VARINT(obj.nUndoSize)); 66 READWRITE(VARINT(obj.nHeightFirst)); 67 READWRITE(VARINT(obj.nHeightLast)); 68 READWRITE(VARINT(obj.nTimeFirst)); 69 READWRITE(VARINT(obj.nTimeLast)); 70 } 71 72 CBlockFileInfo() = default; 73 74 std::string ToString() const; 75 76 /** update statistics (does not update nSize) */ 77 void AddBlock(unsigned int nHeightIn, uint64_t nTimeIn) 78 { 79 if (nBlocks == 0 || nHeightFirst > nHeightIn) 80 nHeightFirst = nHeightIn; 81 if (nBlocks == 0 || nTimeFirst > nTimeIn) 82 nTimeFirst = nTimeIn; 83 nBlocks++; 84 if (nHeightIn > nHeightLast) 85 nHeightLast = nHeightIn; 86 if (nTimeIn > nTimeLast) 87 nTimeLast = nTimeIn; 88 } 89 }; 90 91 /** Access to the block database (blocks/index/) */ 92 class BlockTreeDB : public CDBWrapper 93 { 94 public: 95 using CDBWrapper::CDBWrapper; 96 void WriteBatchSync(const std::vector<std::pair<int, const CBlockFileInfo*>>& fileInfo, int nLastFile, const std::vector<const CBlockIndex*>& blockinfo); 97 bool ReadBlockFileInfo(int nFile, CBlockFileInfo& info); 98 bool ReadLastBlockFile(int& nFile); 99 void WriteReindexing(bool fReindexing); 100 void ReadReindexing(bool& fReindexing); 101 void WriteFlag(const std::string& name, bool fValue); 102 bool ReadFlag(const std::string& name, bool& fValue); 103 bool LoadBlockIndexGuts(const Consensus::Params& consensusParams, std::function<CBlockIndex*(const uint256&)> insertBlockIndex, const util::SignalInterrupt& interrupt) 104 EXCLUSIVE_LOCKS_REQUIRED(::cs_main); 105 }; 106 } // namespace kernel 107 108 namespace node { 109 using kernel::CBlockFileInfo; 110 using kernel::BlockTreeDB; 111 112 /** The pre-allocation chunk size for blk?????.dat files (since 0.8) */ 113 static const unsigned int BLOCKFILE_CHUNK_SIZE = 0x1000000; // 16 MiB 114 /** The pre-allocation chunk size for rev?????.dat files (since 0.8) */ 115 static const unsigned int UNDOFILE_CHUNK_SIZE = 0x100000; // 1 MiB 116 /** The maximum size of a blk?????.dat file (since 0.8) */ 117 static const unsigned int MAX_BLOCKFILE_SIZE = 0x8000000; // 128 MiB 118 119 /** Size of header written by WriteBlock before a serialized CBlock (8 bytes) */ 120 static constexpr uint32_t STORAGE_HEADER_BYTES{std::tuple_size_v<MessageStartChars> + sizeof(unsigned int)}; 121 122 /** Total overhead when writing undo data: header (8 bytes) plus checksum (32 bytes) */ 123 static constexpr uint32_t UNDO_DATA_DISK_OVERHEAD{STORAGE_HEADER_BYTES + uint256::size()}; 124 125 // Because validation code takes pointers to the map's CBlockIndex objects, if 126 // we ever switch to another associative container, we need to either use a 127 // container that has stable addressing (true of all std associative 128 // containers), or make the key a `std::unique_ptr<CBlockIndex>` 129 using BlockMap = std::unordered_map<uint256, CBlockIndex, BlockHasher>; 130 131 struct CBlockIndexWorkComparator { 132 bool operator()(const CBlockIndex* pa, const CBlockIndex* pb) const; 133 }; 134 135 struct CBlockIndexHeightOnlyComparator { 136 /* Only compares the height of two block indices, doesn't try to tie-break */ 137 bool operator()(const CBlockIndex* pa, const CBlockIndex* pb) const; 138 }; 139 140 struct PruneLockInfo { 141 int height_first{std::numeric_limits<int>::max()}; //! Height of earliest block that should be kept and not pruned 142 }; 143 144 enum BlockfileType { 145 // Values used as array indexes - do not change carelessly. 146 NORMAL = 0, 147 ASSUMED = 1, 148 NUM_TYPES = 2, 149 }; 150 151 std::ostream& operator<<(std::ostream& os, const BlockfileType& type); 152 153 struct BlockfileCursor { 154 // The latest blockfile number. 155 int file_num{0}; 156 157 // Track the height of the highest block in file_num whose undo 158 // data has been written. Block data is written to block files in download 159 // order, but is written to undo files in validation order, which is 160 // usually in order by height. To avoid wasting disk space, undo files will 161 // be trimmed whenever the corresponding block file is finalized and 162 // the height of the highest block written to the block file equals the 163 // height of the highest block written to the undo file. This is a 164 // heuristic and can sometimes preemptively trim undo files that will write 165 // more data later, and sometimes fail to trim undo files that can't have 166 // more data written later. 167 int undo_height{0}; 168 }; 169 170 std::ostream& operator<<(std::ostream& os, const BlockfileCursor& cursor); 171 172 173 /** 174 * Maintains a tree of blocks (stored in `m_block_index`) which is consulted 175 * to determine where the most-work tip is. 176 * 177 * This data is used mostly in `Chainstate` - information about, e.g., 178 * candidate tips is not maintained here. 179 */ 180 class BlockManager 181 { 182 friend Chainstate; 183 friend ChainstateManager; 184 185 private: 186 const CChainParams& GetParams() const { return m_opts.chainparams; } 187 const Consensus::Params& GetConsensus() const { return m_opts.chainparams.GetConsensus(); } 188 /** 189 * Load the blocktree off disk and into memory. Populate certain metadata 190 * per index entry (nStatus, nChainWork, nTimeMax, etc.) as well as peripheral 191 * collections like m_dirty_blockindex. 192 */ 193 bool LoadBlockIndex(const std::optional<uint256>& snapshot_blockhash) 194 EXCLUSIVE_LOCKS_REQUIRED(cs_main); 195 196 /** Return false if block file or undo file flushing fails. */ 197 [[nodiscard]] bool FlushBlockFile(int blockfile_num, bool fFinalize, bool finalize_undo); 198 199 /** Return false if undo file flushing fails. */ 200 [[nodiscard]] bool FlushUndoFile(int block_file, bool finalize = false); 201 202 /** 203 * Helper function performing various preparations before a block can be saved to disk: 204 * Returns the correct position for the block to be saved, which may be in the current or a new 205 * block file depending on nAddSize. May flush the previous blockfile to disk if full, updates 206 * blockfile info, and checks if there is enough disk space to save the block. 207 * 208 * The nAddSize argument passed to this function should include not just the size of the serialized CBlock, but also the size of 209 * separator fields (STORAGE_HEADER_BYTES). 210 */ 211 [[nodiscard]] FlatFilePos FindNextBlockPos(unsigned int nAddSize, unsigned int nHeight, uint64_t nTime); 212 [[nodiscard]] bool FlushChainstateBlockFile(int tip_height); 213 bool FindUndoPos(BlockValidationState& state, int nFile, FlatFilePos& pos, unsigned int nAddSize); 214 215 AutoFile OpenUndoFile(const FlatFilePos& pos, bool fReadOnly = false) const; 216 217 /* Calculate the block/rev files to delete based on height specified by user with RPC command pruneblockchain */ 218 void FindFilesToPruneManual( 219 std::set<int>& setFilesToPrune, 220 int nManualPruneHeight, 221 const Chainstate& chain, 222 ChainstateManager& chainman); 223 224 /** 225 * Prune block and undo files (blk???.dat and rev???.dat) so that the disk space used is less than a user-defined target. 226 * The user sets the target (in MB) on the command line or in config file. This will be run on startup and whenever new 227 * space is allocated in a block or undo file, staying below the target. Changing back to unpruned requires a reindex 228 * (which in this case means the blockchain must be re-downloaded.) 229 * 230 * Pruning functions are called from FlushStateToDisk when the m_check_for_pruning flag has been set. 231 * Block and undo files are deleted in lock-step (when blk00003.dat is deleted, so is rev00003.dat.) 232 * Pruning cannot take place until the longest chain is at least a certain length (CChainParams::nPruneAfterHeight). 233 * Pruning will never delete a block within a defined distance (currently 288) from the active chain's tip. 234 * The block index is updated by unsetting HAVE_DATA and HAVE_UNDO for any blocks that were stored in the deleted files. 235 * A db flag records the fact that at least some block files have been pruned. 236 * 237 * @param[out] setFilesToPrune The set of file indices that can be unlinked will be returned 238 * @param last_prune The last height we're able to prune, according to the prune locks 239 */ 240 void FindFilesToPrune( 241 std::set<int>& setFilesToPrune, 242 int last_prune, 243 const Chainstate& chain, 244 ChainstateManager& chainman); 245 246 RecursiveMutex cs_LastBlockFile; 247 std::vector<CBlockFileInfo> m_blockfile_info; 248 249 //! Since assumedvalid chainstates may be syncing a range of the chain that is very 250 //! far away from the normal/background validation process, we should segment blockfiles 251 //! for assumed chainstates. Otherwise, we might have wildly different height ranges 252 //! mixed into the same block files, which would impair our ability to prune 253 //! effectively. 254 //! 255 //! This data structure maintains separate blockfile number cursors for each 256 //! BlockfileType. The ASSUMED state is initialized, when necessary, in FindNextBlockPos(). 257 //! 258 //! The first element is the NORMAL cursor, second is ASSUMED. 259 std::array<std::optional<BlockfileCursor>, BlockfileType::NUM_TYPES> 260 m_blockfile_cursors GUARDED_BY(cs_LastBlockFile) = { 261 BlockfileCursor{}, 262 std::nullopt, 263 }; 264 int MaxBlockfileNum() const EXCLUSIVE_LOCKS_REQUIRED(cs_LastBlockFile) 265 { 266 static const BlockfileCursor empty_cursor; 267 const auto& normal = m_blockfile_cursors[BlockfileType::NORMAL].value_or(empty_cursor); 268 const auto& assumed = m_blockfile_cursors[BlockfileType::ASSUMED].value_or(empty_cursor); 269 return std::max(normal.file_num, assumed.file_num); 270 } 271 272 /** Global flag to indicate we should check to see if there are 273 * block/undo files that should be deleted. Set on startup 274 * or if we allocate more file space when we're in prune mode 275 */ 276 bool m_check_for_pruning = false; 277 278 const bool m_prune_mode; 279 280 const Obfuscation m_obfuscation; 281 282 /** Dirty block index entries. */ 283 std::set<CBlockIndex*> m_dirty_blockindex; 284 285 /** Dirty block file entries. */ 286 std::set<int> m_dirty_fileinfo; 287 288 /** 289 * Map from external index name to oldest block that must not be pruned. 290 * 291 * @note Internally, only blocks at height (height_first - PRUNE_LOCK_BUFFER - 1) and 292 * below will be pruned, but callers should avoid assuming any particular buffer size. 293 */ 294 std::unordered_map<std::string, PruneLockInfo> m_prune_locks GUARDED_BY(::cs_main); 295 296 BlockfileType BlockfileTypeForHeight(int height); 297 298 const kernel::BlockManagerOpts m_opts; 299 300 const FlatFileSeq m_block_file_seq; 301 const FlatFileSeq m_undo_file_seq; 302 303 public: 304 using Options = kernel::BlockManagerOpts; 305 306 explicit BlockManager(const util::SignalInterrupt& interrupt, Options opts); 307 308 const util::SignalInterrupt& m_interrupt; 309 std::atomic<bool> m_importing{false}; 310 311 /** 312 * Whether all blockfiles have been added to the block tree database. 313 * Normally true, but set to false when a reindex is requested and the 314 * database is wiped. The value is persisted in the database across restarts 315 * and will be false until reindexing completes. 316 */ 317 std::atomic_bool m_blockfiles_indexed{true}; 318 319 BlockMap m_block_index GUARDED_BY(cs_main); 320 321 /** 322 * The height of the base block of an assumeutxo snapshot, if one is in use. 323 * 324 * This controls how blockfiles are segmented by chainstate type to avoid 325 * comingling different height regions of the chain when an assumedvalid chainstate 326 * is in use. If heights are drastically different in the same blockfile, pruning 327 * suffers. 328 * 329 * This is set during ActivateSnapshot() or upon LoadBlockIndex() if a snapshot 330 * had been previously loaded. After the snapshot is validated, this is unset to 331 * restore normal LoadBlockIndex behavior. 332 */ 333 std::optional<int> m_snapshot_height; 334 335 std::vector<CBlockIndex*> GetAllBlockIndices() EXCLUSIVE_LOCKS_REQUIRED(::cs_main); 336 337 /** 338 * All pairs A->B, where A (or one of its ancestors) misses transactions, but B has transactions. 339 * Pruned nodes may have entries where B is missing data. 340 */ 341 std::multimap<CBlockIndex*, CBlockIndex*> m_blocks_unlinked; 342 343 std::unique_ptr<BlockTreeDB> m_block_tree_db GUARDED_BY(::cs_main); 344 345 void WriteBlockIndexDB() EXCLUSIVE_LOCKS_REQUIRED(::cs_main); 346 bool LoadBlockIndexDB(const std::optional<uint256>& snapshot_blockhash) 347 EXCLUSIVE_LOCKS_REQUIRED(::cs_main); 348 349 /** 350 * Remove any pruned block & undo files that are still on disk. 351 * This could happen on some systems if the file was still being read while unlinked, 352 * or if we crash before unlinking. 353 */ 354 void ScanAndUnlinkAlreadyPrunedFiles() EXCLUSIVE_LOCKS_REQUIRED(::cs_main); 355 356 CBlockIndex* AddToBlockIndex(const CBlockHeader& block, CBlockIndex*& best_header) EXCLUSIVE_LOCKS_REQUIRED(cs_main); 357 /** Create a new block index entry for a given block hash */ 358 CBlockIndex* InsertBlockIndex(const uint256& hash) EXCLUSIVE_LOCKS_REQUIRED(cs_main); 359 360 //! Mark one block file as pruned (modify associated database entries) 361 void PruneOneBlockFile(const int fileNumber) EXCLUSIVE_LOCKS_REQUIRED(cs_main); 362 363 CBlockIndex* LookupBlockIndex(const uint256& hash) EXCLUSIVE_LOCKS_REQUIRED(cs_main); 364 const CBlockIndex* LookupBlockIndex(const uint256& hash) const EXCLUSIVE_LOCKS_REQUIRED(cs_main); 365 366 /** Get block file info entry for one block file */ 367 CBlockFileInfo* GetBlockFileInfo(size_t n); 368 369 bool WriteBlockUndo(const CBlockUndo& blockundo, BlockValidationState& state, CBlockIndex& block) 370 EXCLUSIVE_LOCKS_REQUIRED(::cs_main); 371 372 /** Store block on disk and update block file statistics. 373 * 374 * @param[in] block the block to be stored 375 * @param[in] nHeight the height of the block 376 * 377 * @returns in case of success, the position to which the block was written to 378 * in case of an error, an empty FlatFilePos 379 */ 380 FlatFilePos WriteBlock(const CBlock& block, int nHeight); 381 382 /** Update blockfile info while processing a block during reindex. The block must be available on disk. 383 * 384 * @param[in] block the block being processed 385 * @param[in] nHeight the height of the block 386 * @param[in] pos the position of the serialized CBlock on disk 387 */ 388 void UpdateBlockInfo(const CBlock& block, unsigned int nHeight, const FlatFilePos& pos); 389 390 /** Whether running in -prune mode. */ 391 [[nodiscard]] bool IsPruneMode() const { return m_prune_mode; } 392 393 /** Attempt to stay below this number of bytes of block files. */ 394 [[nodiscard]] uint64_t GetPruneTarget() const { return m_opts.prune_target; } 395 static constexpr auto PRUNE_TARGET_MANUAL{std::numeric_limits<uint64_t>::max()}; 396 397 [[nodiscard]] bool LoadingBlocks() const { return m_importing || !m_blockfiles_indexed; } 398 399 /** Calculate the amount of disk space the block & undo files currently use */ 400 uint64_t CalculateCurrentUsage(); 401 402 //! Check if all blocks in the [upper_block, lower_block] range have data available. 403 //! The caller is responsible for ensuring that lower_block is an ancestor of upper_block 404 //! (part of the same chain). 405 bool CheckBlockDataAvailability(const CBlockIndex& upper_block, const CBlockIndex& lower_block) EXCLUSIVE_LOCKS_REQUIRED(::cs_main); 406 407 /** 408 * @brief Returns the earliest block with specified `status_mask` flags set after 409 * the latest block _not_ having those flags. 410 * 411 * This function starts from `upper_block`, which must have all 412 * `status_mask` flags set, and iterates backwards through its ancestors. It 413 * continues as long as each block has all `status_mask` flags set, until 414 * reaching the oldest ancestor or `lower_block`. 415 * 416 * @pre `upper_block` must have all `status_mask` flags set. 417 * @pre `lower_block` must be null or an ancestor of `upper_block` 418 * 419 * @param upper_block The starting block for the search, which must have all 420 * `status_mask` flags set. 421 * @param status_mask Bitmask specifying required status flags. 422 * @param lower_block The earliest possible block to return. If null, the 423 * search can extend to the genesis block. 424 * 425 * @return A reference to the earliest block between `upper_block` 426 * and `lower_block`, inclusive, such that every block between the 427 * returned block and `upper_block` has `status_mask` flags set. 428 */ 429 const CBlockIndex& GetFirstBlock( 430 const CBlockIndex& upper_block LIFETIMEBOUND, 431 uint32_t status_mask, 432 const CBlockIndex* lower_block LIFETIMEBOUND = nullptr 433 ) const EXCLUSIVE_LOCKS_REQUIRED(::cs_main); 434 435 /** True if any block files have ever been pruned. */ 436 bool m_have_pruned = false; 437 438 //! Check whether the block associated with this index entry is pruned or not. 439 bool IsBlockPruned(const CBlockIndex& block) const EXCLUSIVE_LOCKS_REQUIRED(::cs_main); 440 441 //! Create or update a prune lock identified by its name 442 void UpdatePruneLock(const std::string& name, const PruneLockInfo& lock_info) EXCLUSIVE_LOCKS_REQUIRED(::cs_main); 443 444 /** Open a block file (blk?????.dat) */ 445 AutoFile OpenBlockFile(const FlatFilePos& pos, bool fReadOnly) const; 446 447 /** Translation to a filesystem path */ 448 fs::path GetBlockPosFilename(const FlatFilePos& pos) const; 449 450 /** 451 * Actually unlink the specified files 452 */ 453 void UnlinkPrunedFiles(const std::set<int>& setFilesToPrune) const; 454 455 /** Functions for disk access for blocks */ 456 bool ReadBlock(CBlock& block, const FlatFilePos& pos, const std::optional<uint256>& expected_hash) const; 457 bool ReadBlock(CBlock& block, const CBlockIndex& index) const; 458 bool ReadRawBlock(std::vector<std::byte>& block, const FlatFilePos& pos) const; 459 460 bool ReadBlockUndo(CBlockUndo& blockundo, const CBlockIndex& index) const; 461 462 void CleanupBlockRevFiles() const; 463 }; 464 465 // Calls ActivateBestChain() even if no blocks are imported. 466 void ImportBlocks(ChainstateManager& chainman, std::span<const fs::path> import_paths); 467 } // namespace node 468 469 #endif // BITCOIN_NODE_BLOCKSTORAGE_H