[execution/vm] Add in-tree `monad_message` type by khordadi · Pull Request #2540 · category-labs/monad

khordadi · 2026-09-05T01:48:58Z

Part of EXE-172 (EXE-162), reduced to the type. monad_message, with monad_call_kind and monad_call_flags, mirrors the fork's evmc_message; message.cpp asserts layout, field offsets, field types and enumerator values. Purely additive, nothing references the type yet.

The call-site migration (formerly commit 2 here) moves into the vm::Host port, EXE-173. With the vtable slot and the runtime on monad_message there is no per-call evmc_message conversion, which the adapter in EvmcHost::call needed (one out-of-line call and a 192-byte copy per nested call).

Verification. gcc-15 RelWithDebInfo build clean; clang-format-19 and clang-tidy-19 on the new files clean. message.cpp includes <evmc/evmc.h> exactly as revision.cpp and status_code.cpp already do, so the zkVM guest sees nothing new.

Conflicts trivially with #2539 on category/vm/evm/CMakeLists.txt; whichever lands first, the other rebases.

🤖 Generated with Claude Code

khordadi

Notes for the reviewer on the non-obvious lines.

khordadi · 2026-09-05T01:48:59Z

+};
+
+// Mirrors evmc_flags 1:1; values asserted equal in message.cpp.
+enum monad_call_flags


Not in the issue title, but without it every reader of flags still needs evmc.h.

khordadi · 2026-09-05T01:48:59Z

+    monad::Address code_address;
+    uint8_t *memory_handle;
+    uint8_t *memory;
+    uint32_t memory_capacity;


Fork layout: evmc 663a1c2 dropped code/code_size, 8533622 added the memory-pool fields. A mirror of upstream evmc_message fails the asserts in message.cpp.

khordadi · 2026-09-05T01:48:59Z

+static_assert(sizeof(monad_message) == sizeof(evmc_message));
+static_assert(alignof(monad_message) == alignof(evmc_message));
+
+#define MONAD_ASSERT_MESSAGE_FIELD_EQ(field)                                   \


Offsets only: sizeof on the pointer members trips bugprone-sizeof-expression. The type asserts below cover the narrowed-field case that offsets alone would miss.

Copilot

🔵 Needs a closer look

It changes core VM/execution message plumbing across many call sites and relies on layout-dependent bit_cast adapters, warranting final human verification.

Pull request overview

This PR continues the execution/vm “evmc leaf types” migration by introducing an in-tree monad_message (plus monad_call_kind / monad_call_flags) and switching the execution layer + VM entry points to use it, keeping EVMC types only at the ABI seams via std::bit_cast adapters with compile-time layout verification.

Changes:

Add category/vm/evm/message.{hpp,cpp} defining monad_message and to_evmc_message / from_evmc_message, with static assertions for enum values and struct layout.
Replace evmc_message with monad_message in VM APIs (VM::execute, VM::execute_bytecode) and in execution-layer message plumbing (precompiles, execute_{message,transaction}, tracing).
Update tests, fuzzing, and benchmarking tooling to construct and pass monad_message, bridging back to EVMC only where required.

Verdict: CORRECT

Generated with Claude Code

File summaries

File	Description
`category/vm/evm/message.hpp`	Introduces `monad_message` + call kind/flags mirrors and conversion function declarations.
`category/vm/evm/message.cpp`	Proves enum/value + layout equivalence vs `evmc_message` and implements `bit_cast` conversions.
`category/vm/evm/CMakeLists.txt`	Adds the new message mirror sources to the EVM VM target.
`category/vm/vm.hpp`	Updates VM public template entrypoints to accept `monad_message`.
`category/vm/vm.cpp`	Adapts execute-override path to convert `monad_message` to `evmc_message` for the EVMC ABI hook.
`category/vm/runtime/types.hpp`	Changes `runtime::Context::from` to accept `monad_message`.
`category/vm/runtime/context.cpp`	Wires `monad_message` fields into runtime context construction.
`category/vm/fuzzing/generator/generator.hpp`	Switches fuzz-message generation and flags/kind selection to `monad_message` types.
`category/execution/ethereum/execute_transaction.hpp`	Updates transaction→message construction to return `monad_message`.
`category/execution/ethereum/execute_transaction.cpp`	Replaces message kinds/flags and value encoding to populate `monad_message`.
`category/execution/ethereum/execute_message.hpp`	Updates execute_{call,create}_message signatures to take `monad_message`.
`category/execution/ethereum/execute_message.cpp`	Ports runtime message handling logic from `evmc_message` to `monad_message`.
`category/execution/ethereum/precompiles.hpp`	Ports precompile checks to `monad_message`.
`category/execution/ethereum/precompiles.cpp`	Updates flag checks (delegated/static) and signatures to `monad_message`.
`category/execution/ethereum/evmc_host.hpp`	Adds `call(monad_message const&)` and makes the EVMC vtable slot an adapter.
`category/execution/ethereum/trace/call_tracer.hpp`	Updates tracer interface `on_enter` to accept `monad_message`.
`category/execution/ethereum/trace/call_tracer.cpp`	Ports call-kind handling to `MONAD_*` enumerators.
`category/execution/ethereum/trace/call_frame.cpp`	Ports static-call flag checks to `MONAD_STATIC`.
`category/execution/ethereum/process_requests.cpp`	Constructs system-call messages as `monad_message`.
`category/execution/ethereum/block_hash_history_test.cpp`	Updates test message construction to `monad_message`.
`category/execution/ethereum/trace/state_tracer_test.cpp`	Updates message construction to `monad_message` for tracing tests.
`category/execution/ethereum/test/test_call_trace.cpp`	Updates trace test message type to `monad_message`.
`category/execution/ethereum/precompiles_test.cpp`	Updates precompile tests to use `monad_message`.
`category/execution/ethereum/execute_message_test.cpp`	Updates execution tests to use `monad_message` and `bytes32_t` values directly.
`category/execution/monad/monad_precompiles.hpp`	Ports Monad precompile entrypoint to accept `monad_message`.
`category/execution/monad/monad_precompiles.cpp`	Ports kind checks to `MONAD_CALL` and signature to `monad_message`.
`category/execution/monad/execute_system_transaction.hpp`	Changes system-tx `to_message()` to return `monad_message`.
`category/execution/monad/execute_system_transaction.cpp`	Constructs `monad_message` for system tx execution and encodes value into `bytes32_t`.
`category/execution/monad/reserve_balance/reserve_balance_contract_test.cpp`	Updates reserve-balance tests and helper signatures to `monad_message`.
`cmd/vm/mce/src/instrumentable_vm.hpp`	Switches microbenchmark message allocation and kinds to `monad_message`.
`test/vm/utils/test_message.hpp`	Updates `TestMessage` helper to store a `monad_message`.
`test/vm/vm/test_vm.cpp`	Adapts EVMC ABI `BlockchainTestVM::execute` to convert EVMC→Monad message before `Context::from`.
`test/vm/unit/runtime/memory_tests.cpp`	Uses `monad_message` in runtime memory unit tests.
`test/vm/unit/monad_vm_interface_tests.cpp`	Updates host mock plumbing and test lambdas to accept `monad_message`.
`test/vm/unit/evm_fixture.hpp`	Keeps EVMC ABI boundary by converting `monad_message` back to `evmc_message` for evmone.
`test/vm/fuzzer/fuzzer.cpp`	Ports fuzz transition helpers to accept `monad_message`.

Review details

Files reviewed: 36/36 changed files
Comments generated: 0
Review effort level: Lite

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

github-actions

Reviewed per REVIEW.md: traits usage, FFI/ABI boundaries, C++ correctness (lifetimes, bit_cast legality, overload/name-hiding, temporary binding at every conversion seam), testing, style, and hygiene.

What I verified:

Layout proof is sound. message.cpp pins size (I independently recomputed the 192-byte layout: 4+4+4+pad+8+20+20+8+8+32+32+20+pad+8+8+4+pad = 192), alignment, all 14 field offsets, exact types of the 8 integer/pointer fields, and the enumerator values. Combined with is_standard_layout / is_trivially_copyable, this makes the two std::bit_casts well-defined; a mirror written from upstream evmc (or a swapped enumerator) fails to compile, as the PR description demonstrates.
No overload/dispatch hazards. EvmcHost is final, so the non-virtual in-tree call(monad_message const &) alongside the virtual evmc adapter cannot be bypassed by a derived override, and the adapter's from_evmc_message argument selects the monad overload with no recursion risk.
Temporary lifetimes are correct at every seam: to_evmc_message locals in VM::execute/execute_bytecode outlive the synchronous override call; the temporaries bound to const-ref parameters in evm_fixture.hpp, monad_vm_interface_tests.cpp, and test_vm.cpp live through the full expression / enclosing scope of their synchronous use.
Remaining evmc_message sites are the intended ABI boundaries (runtime/call.cpp/create.cpp feeding the evmc host table, evmc_vm callbacks, benchmark harnesses, ExecuteOverride), matching the EXE-173/EXE-161/EXE-174 scoping in the description.
Traits/instantiation rules respected: no macro changes, no new is_monad_trait_v branching in ethereum/, and check-trait-instantiations passes.
FFI: monad_message is not exposed to the Rust bindgen headers; conversions happen before crossing the evmone C ABI.

Testing via the compile-time assertion suite is appropriate for a pure layout mirror. CI builds/tests were still pending at review time; the PR's own verification record (full ctest, ASAN+UBSAN with compiler testing, fuzzer, spec tests) is thorough.

One non-blocking nit posted inline (P3: noexcept on the conversion functions).

Verdict: CORRECT

🤖 Generated with Claude Code

EXE-172, part of EXE-162. monad_message, with monad_call_kind and monad_call_flags, mirrors the fork's evmc_message field-for-field (no code/code_size, evmc 663a1c2; memory_handle/memory/memory_capacity, evmc 8533622). message.cpp asserts standard layout, sizeof, alignof, the offsetof of all 14 fields, the exact type of the 8 integer/pointer fields and the enumerator values. Purely additive; nothing references the type yet. The call-site migration goes with the vm::Host port (EXE-173), where the runtime builds the message itself and no per-call evmc conversion is needed. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

Copilot AI lite review requested due to automatic review settings September 5, 2026 01:48

khordadi requested review from Baltoli and andreaslyn as code owners September 5, 2026 01:48

khordadi commented Sep 5, 2026

View reviewed changes

Copilot started reviewing on behalf of khordadi September 5, 2026 01:49 View session

Copilot AI reviewed Sep 5, 2026

View reviewed changes

github-actions Bot reviewed Sep 5, 2026

View reviewed changes

Comment thread category/vm/evm/message.hpp Outdated

github-actions Bot reviewed Sep 5, 2026

View reviewed changes

khordadi force-pushed the akhordadi/exe-172-port-evmc_message-and-evmc_call_kind-to-in-tree-monad-types branch 4 times, most recently from 9c4b807 to 1184605 Compare September 15, 2026 12:07

khordadi force-pushed the akhordadi/exe-172-port-evmc_message-and-evmc_call_kind-to-in-tree-monad-types branch from 1184605 to 8fe6bf2 Compare September 15, 2026 13:38

khordadi changed the title ~~[execution/vm] Port evmc_message and evmc_call_kind to in-tree Monad types~~ [execution/vm] Add in-tree monad_message type Sep 15, 2026

khordadi mentioned this pull request Sep 15, 2026

[execution/vm] Add in-tree monad_tx_context type #2539

Open

Navigation Menu

[execution/vm] Add in-tree `monad_message` type - #2540

[execution/vm] Add in-tree `monad_message` type#2540
khordadi wants to merge 1 commit into
mainfrom
akhordadi/exe-172-port-evmc_message-and-evmc_call_kind-to-in-tree-monad-types

khordadi commented Sep 5, 2026 •

edited

Loading

Uh oh!

khordadi left a comment

Uh oh!

khordadi Sep 5, 2026 •

edited

Loading

Uh oh!

khordadi Sep 5, 2026 •

edited

Loading

Uh oh!

khordadi Sep 5, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

Uh oh!

github-actions Bot left a comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

khordadi commented Sep 5, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

khordadi left a comment

Choose a reason for hiding this comment

Uh oh!

khordadi Sep 5, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

khordadi Sep 5, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

khordadi Sep 5, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

🔵 Needs a closer look

Uh oh!

Uh oh!

github-actions Bot left a comment

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

khordadi commented Sep 5, 2026 •

edited

Loading

khordadi Sep 5, 2026 •

edited

Loading

khordadi Sep 5, 2026 •

edited

Loading