zeekayandHanzo Dev
2f9a315893
The C++ and Rust EVMs are separate luxd plugins. The host spawns every VM with VM_TRANSPORT=zap (node/vms/rpcchainvm/factory.go) and the cevm binary is exercised that way cross-process today. Choosing an EVM is choosing which plugin binary runs -- not a switch inside the Go EVM. This tree carried a second mechanism for the same choice: an in-process cgo bridge that linked C++ into the Go binary behind -tags cevm/revm, with a per-transaction executor registry, an evm_setBackend RPC to flip between backends at runtime, a resident-store config knob, and a startup log naming the chosen backend. Its executor returned nil, nil; its tagged build did not compile (cevm_resident_store.go referenced symbols cevmbridge never defined); and no build produced a binary that used it. 3,961 lines, gone. What stays is what the Go EVM actually has: the BlockExecutor seam and Block-STM behind it (blockstm.go), and the GPU accelerator seam with its darwin/cgo bridge into luxfi/gpu (gpu_bridge.go). The state processor consults both and falls through to sequential execution when neither is registered. Note for the record: nothing in the shipped binary registers a BlockExecutor, so blocks run sequentially on the Go interpreter. The removed startup log said "Block-STM across all cores"; it was not describing this binary. Wiring Block-STM into the production path changes consensus execution and is a separate decision, made deliberately, with a parity gate. Co-authored-by: Hanzo Dev <[email protected]>
Core Package
The core package maintains the backend for the blockchain, transaction pool, and maintains the required indexes for blocks, transactions, logs, and transaction receipts.
Blockchain
The BlockChain struct handles the insertion of blocks into the maintained chain. It maintains a "canonical chain", which is essentially the preferred chain (the chain that ends with the block preferred by the Luxd consensus engine).
When the consensus engine verifies blocks as they are ready to be issued into consensus, it calls Verify() on the ChainVM Block interface implemented here. This calls InsertBlockManual on the BlockChain struct implemented in this package, which is the first entrypoint of a block into the blockchain.
InsertBlockManual verifies the block, inserts it into the state manager to track the merkle trie for the block, and adds it to the canonical chain if it extends the currently preferred chain.
EVM adds functions for Accept and Reject, which take care of marking a block as finalized and performing garbage collection where possible.
The consensus engine can also call SetPreference on a VM to tell the VM that a specific block is preferred by the consensus engine to be accepted. This triggers a call to reorg the blockchain and set the newly preferred block as the preferred chain.
Transaction Pool
The transaction pool maintains the set of transactions that need to be issued into a new block. The VM exposes APIs that allow clients to issue transactions into the transaction pool and also performs gossip across the network in order to send and receive pending transactions that need to be issued into a new block. The transaction pool asynchronously follows the preferred block of the BlockChain struct by subscribing to new head events and updating its state accordingly. When the transaction pool updates, it ensures that any transactions it contains are still valid to be issued on top of the new preferred block.
State Manager
The State Manager manages the TrieDB. The TrieDB tracks a merkle forest of all of the merkle tries for the last accepted block and processing blocks. When a block is processed, the state transition results in a new merkle trie added to the merkle forest. The State Manager can operate in either archival or pruning mode.
Archival Mode
In archival mode, every merkle trie is written to disk so that the node maintains a complete history of all the blocks that it has processed.
Pruning Mode
In pruning mode, the State Manager keeps a reference to merkle tries of processing blocks. When a block gets accepted, it stays in memory. When a block gets rejected, the state manager can dereference and clean up the no longer needed merkle trie. The State Manager does not immediately write the merkle trie to disk of a block when it gets accepted. Instead, at a regular interval (~4096 blocks) it writes the merkle trie to disk, so that it does not add the overhead of storing every accepted block's merkle trie to disk.