This PR introduces powerful code size optimizations. This is achieved
with an additional IR layer in between Yul and LLVM IR. It works quite
well (see below for a comparison) because it captures optimization gains
that neither solc nor LLVM can realize.
The feature is gated behind `--newyork` or a corresponding standard JSON
input setting and **off by default**.
### How it works
We implemented an SSA IR modelled after or inspired by the [MLIR SCF
dialect](https://mlir.llvm.org/docs/Dialects/SCFDialect/). Yul is
translated into this IR and the IR implements mainly the following
optimizations:
- **Type narrowing**: narrows 256-bit values to I1..I160 where provable
(bidirectional + interprocedural), so LLVM emits native 32/64-bit ops
instead
of software 256-bit arithmetic.
- **FMP range proof**: exposes the `mload(0x40) < heap_size` bound to
LLVM
IPSCCP, eliminating trailing overflow checks on FMP-derived offsets.
Single
largest code-size contributor.
- **Heap analysis**: skips big-endian byte-swaps where native
little-endian
layout is safe.
- **mem_opt / keccak fusion+folding**, **compound outlining** (mapping
sload/sstore), **guard narrowing**, **inlining** tuned for PolkaVM, and
**exact + fuzzy function dedup**.
### Results
The optimizations implemented with the newyork IR bring substantial code
size improvements.
- Integration contracts: approximately **-20% to -50%** code size.
- OpenZeppelin Wizard corpus: around **-47%** code size.
This table shows the code size improvements on various OpenZeppelin ERC
implementations, taken from their [contracts
wizard](https://wizard.openzeppelin.com/), with almost all features
enabled.
| Contract | Default: (bytes) | With `--newyork`: (bytes) | Reduction |
|---|---:|---:|---:|
| erc1155 | 59,206 | 30,649 | -48.2% |
| erc20 | 81,697 | 42,893 | -47.5% |
| erc721 | 92,052 | 49,423 | -46.3% |
| oz_gov | 141,861 | 81,159 | -42.8% |
| oz_rwa | 77,274 | 37,975 | -50.9% |
| oz_simple_erc20 | 29,950 | 16,554 | -44.7% |
| oz_stable | 82,117 | 39,032 | -52.5% |
| **Total** | **564,157** | **297,685** | **-47.2%** |
We believe this to be a reasonably representative dataset and conclude
that the newyork IR achieves drastic code size reductions on real world
dApp implementations.
### Methodology, security and runtime performance
Since the main contributors to these improvements are the FMP range
proofs and the type narrowing, we don't expect big runtime performance
regressions (it may even benefit runtime performance in some cases).
The PR also includes docs
(`book/src/developer_guide/newyork_{optimizer,ir}.md`) and many new
integration tests, including regression tests for known FMP soundness
traps.
The PR was authored collaboratively and reviewed extensively by the
`revive` maintainers, as well as Claude Opus, Claude Fable, Qwen,
Minimax and Deepseek LLMs, over a span of many months.
The `newyork` compiler pipeline introduces no new regressions over our
integration and differential test suites (containing over 30000 test
cases). This was achieved by careful manual reviews and many LLM bughunt
loops. Additionally, a final security review by Anthropic's Fable 5 LLM
found no remaining soundness issues. Nonetheless, like any brand new
software, please consider this an experimental feature upon release.
---------
Signed-off-by: xermicus <bigcyrill@hotmail.com>
Signed-off-by: xermicus <cyrill@parity.io>
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Co-authored-by: neo <me@90f9.org>
Co-authored-by: LJ <81748770+elle-j@users.noreply.github.com>
Separate between compilation and linker phases to allow deploy time
linking and back-porting era compiler changes to fix#91. Unlinked
contract binaries (caused by missing libraries or missing factory
dependencies in turn) are emitted as raw ELF object.
Few drive by fixes:
- #98
- A compiler panic on missing libraries definitions.
- Fixes some incosistent type forwarding in JSON output (empty string
vs. null object).
- Remove the unused fallback for size optimization setting.
- Remove the broken `--lvm-ir` mode.
- CI workflow fixes.
---------
Signed-off-by: Cyrill Leutwiler <bigcyrill@hotmail.com>
Signed-off-by: xermicus <bigcyrill@hotmail.com>
Signed-off-by: xermicus <cyrill@parity.io>
Exposes the following PolkaVM specific options via the standard json
interface:
- Heap size
- Stack size
- Whether to emit source level debug information
Additionally it is now forbidden to specify those as CLI option in
standard JSON mode. They are bytecode altering options and having
multiple ways to specify them creates unnecessary room for confusion:
The standard JSON input description should be sufficient and succint for
reproducible builds.
Closes#290
---------
Signed-off-by: xermicus <bigcyrill@hotmail.com>
- Allow configuration of the maximum heap and stack size via CLI flags
and JSON input settings.
- Increase the default value for the stack size to 32kb.
---------
Signed-off-by: Cyrill Leutwiler <bigcyrill@hotmail.com>
This PR changes the implementation of the emulated EVM heap memory:
Instead of linking in a C implementation the code is emitted directly
into the contract module. Which allows making it configurable via a
compiler parameter (a follow up PR to this).
---------
Signed-off-by: Cyrill Leutwiler <bigcyrill@hotmail.com>
Storage keys and values are big endian. Keeping them LE was a pre-mature
optimization because for the contract itself it this is a no-op and thus
not observable. However we should consider the storage layout as part of
the contract ABI. The endianness of transient storage values are still
kept as-is.
---------
Signed-off-by: Cyrill Leutwiler <bigcyrill@hotmail.com>
- Support for passing LLVM command line options via the prcoess input or
providing one or more `--llvm-arg='..'` resolc CLI flag. This allows
more fine-grained control over the LLVM backend configuration.
- Make LLVM initialization idempotent.
---------
Signed-off-by: Cyrill Leutwiler <bigcyrill@hotmail.com>
The differential testing framework will make a second consumer. There
seems to be no re-usable Rust crate for this. But we already have
everything here, just needs a small refactor to make it fully re-usable.
- Mostly decouple the solc JSON-input-output interface types from the
`solidity` frontend crate
- Expose the JSON-input-output interface types in a dedicated crate
---------
Signed-off-by: Cyrill Leutwiler <bigcyrill@hotmail.com>
- Add the revive runtime function interface to minimize boiler plate
code.
- Outline heavily repeated code into dedicated functions to bring down
code size.
- The code size tests builds optimized for size.
- Function attributes are passed as slices.
This significantly brings down the code size for all OpenZeppelin wizard
contracts (using all possible features) compiled against OpenZeppelin
`v5.0.0` with size optimizations.
|contract|| `-Oz` main | `-Oz` PR || `-O3` main | `-O3` PR |
|-|-|-|-|-|-|-|
|erc1155.sol||100K|67K||114K|147K|
|erc20.sol||120K|90K||160K|191K|
|erc721.sol||128K|101K||178K|214K|
|governor.sol||226K|165K||293K|349K|
|rwa.sol||116K|85K||154K|185K|
|stable.sol||116K|86K||155K|192K|
On the flip side this introduces a heavy penalty for cycle optimized
builds. Setting the no-inline attributes for cycle optimized builds
helps a lot but heavily penalizes runtime speed (LLVM does not yet
inline everything properly - to be investigated later on).
Next steps:
- Modularize more functions
- Refactor the YUL function arguments to use pointers instead of values
- Afterwards check if LLVM still has trouble inline-ing properly on O3
or set the no-inline attribute if it does not penalize runtime
performance too bad.
