mirror of
https://github.com/luxfi/contracts.git
synced 2026-10-30 16:59:12 +00:00
zeekay
cf34e276b0
Canonical Solidity helpers for the Lux EVM's native post-quantum precompiles. One source of truth for L1 contract authors who need to authenticate post-quantum signers. PQAuth wraps: 0x012201 ML-KEM-768 (FIPS 203, key encapsulation) 0x012202 ML-DSA verify (FIPS 204, signatures: 44/65/87) 0x012203 SLH-DSA verify (FIPS 205, hash-based) 0x012204 Corona threshold (lattice threshold verify) 0x002221 X-Wing hybrid KEM (X25519 + ML-KEM-768) API: verifyMLDSA44/65/87(pubkey, msgDigest, sig) -> bool verifySLHDSA (pubkey, msgDigest, sig) -> bool verifyPermit (pubkey, msgDigest, sig) reverts on invalid The msgDigest is 32 bytes, precomputed off-chain as TupleHash256(parts, 32, "LUX_PQ_PERMIT_V1") per NIST SP 800-185 §5. We refuse to implement TupleHash256/cSHAKE256 in Solidity: the EVM lacks the primitive and emulation costs ~200k gas with high risk of subtle field-framing bugs. Reference Go implementation: luxfi/consensus/protocol/auth/hash.go. Gas comparison vs ETHDILITHIUM (LGPL-3.0, vendored ML-DSA in YUL): ETHDILITHIUM verify: ~8,100,000 gas PQAuth.verifyMLDSA65: 100,320 gas (100k base + 320 for 32-byte digest) => ~80x cheaper. Do not vendor or re-implement ML-DSA in user code. IPQVerify.verifyZAuthProof reverts: reserved for the F102/F103 tx-type 0x05 path (separate multi-week project, Option B in the strict-PQ rollout plan). Tests (10 passing under forge): test_VerifyMLDSA65_ValidSignature_ReturnsTrue test_VerifyMLDSA65_InvalidSignature_ReturnsFalse test_VerifyMLDSA65_WrongPubkeyLength_Reverts test_VerifyMLDSA65_WrongSignatureLength_Reverts test_VerifyPermit_ValidSignature_NoRevert test_VerifyPermit_InvalidSignature_Reverts test_VerifySLHDSA_ValidSignature_ReturnsTrue test_VerifySLHDSA_InvalidSignature_ReturnsFalse test_VerifyZAuthProof_Reverts test_CanonicalAddresses Stack: Foundry, solc 0.8.27, evm_version=cancun.