OP Contracts Manager

The OP Contracts Manager is a contract that deploys the L1 contracts for an OP Stack chain in a single transaction. It provides a minimal set of user-configurable parameters to ensure that the resulting chain meets the standard configuration requirements.

The version deployed is always a governance-approved contract release. The set of governance approved contract releases can be found on the Optimism Monorepo releases page, and is the set of releases named op-contracts/vX.Y.Z.

Table of Contents

Overview

The OP Contracts Manager refers to a series of contracts, of which a new singleton is deployed for each new release of the OP Stack contracts.

The OP Contracts Manager corresponding to each release can be used to:

  1. Deploy a new OP chain.
  2. Upgrade the contracts for an existing OP chain from the previous release to the new release.
  3. Orchestrate adding a new game type on a per-chain basis

Upgrades must be performed by the Upgrade Controller Safe for a chain.

Getter Methods

The following interface defines the available getter methods:

/// @notice Returns the latest approved release of the OP Stack contracts are named with the
///         format `op-contracts/vX.Y.Z`.
function release() external view returns (string memory);
/// @notice Represents the interface version so consumers know how to decode the DeployOutput struct
function outputVersion() external view returns (uint256);
/// @notice Addresses of the Blueprint contracts.
function blueprints() external view returns (Blueprints memory);
/// @notice Maps an L2 chain ID to an L1 batch inbox address
function chainIdToBatchInboxAddress(uint256 _l2ChainId) external pure returns (address);
/// @notice Addresses of the latest implementation contracts.
function implementations() external view returns (Implementations memory);
/// @notice Address of the ProtocolVersions contract shared by all chains.
function protocolVersions() external view returns (address);
/// @notice Address of the SuperchainConfig contract shared by all chains.
function superchainConfig() external view returns (address);
/// @notice Maps an L2 Chain ID to the SystemConfig for that chain.
function systemConfigs(uint256 _l2ChainId) external view returns (address);
/// @notice Semver version specific to the OPContractsManager
function version() external view returns (string memory);

Deployment

Interface

deploy

The deploy method is used to deploy the full set of L1 contracts required to setup a new OP Stack chain that complies with the standard configuration. It has the following interface:

/// @notice Deploys a new OP Chain
/// @param _input DeployInput containing chain specific config information.
/// @return DeployOutput containing the new addresses.
function deploy(DeployInput calldata _input) external returns (DeployOutput memory)

The l2ChainId has the following restrictions:

  • It must not be equal to 0.
  • It must not be equal to the chain ID of the chain the OP Contracts Manager is deployed on.
  • It must not be equal to a chain ID that is already present in the ethereum-lists/chains repository. This is not enforced onchain, but may matter for future versions of OP Contracts Manager that handle upgrades.

On success, the following event is emitted:

event Deployed(uint256 indexed outputVersion, uint256 indexed l2ChainId, address indexed deployer, bytes deployOutput);

This method reverts on failure. This occurs when:

  • The input l2ChainId does not comply with the restrictions above.
  • The resulting configuration is not compliant with the standard configuration.

Implementation

Batch Inbox Address

The chain's Batch Inbox address is computed at deploy time using the recommend approach defined in the standard configuration. This improves UX by removing an input, and ensures uniqueness of the batch inbox addresses.

Contract Deployments

All contracts deployed by the OP Contracts Manager are deployed with CREATE2, using the following salt:

keccak256(abi.encode(_l2ChainId, _saltMixer, _contractName));

The saltMixer value is provided as a field in the DeployInput struct.

This provides the following benefits:

  • Contract addresses for a chain can be derived as a function of chain ID without any RPC calls.
  • Chain ID uniqueness is enforced for free, as a deploy using the same chain ID will result in attempting to deploy to the same address, which is prohibited by the EVM.
    • This property is contingent on the proxy and AddressManager code not changing when OP Contracts Manager is upgraded. Both of these are not planned to change.
    • The OP Contracts Manager is not responsible for enforcing chain ID uniqueness, so it is acceptable if this property is not preserved in future versions of the OP Contracts Manager.

Upgrading

Interface

upgrade

The upgrade method is used by the Upgrade Controller to upgrade the full set of L1 contracts for all chains that it controls.

It has the following interface:

struct Roles {
    address opChainProxyAdminOwner;
    address systemConfigOwner;
    address batcher;
    address unsafeBlockSigner;
    address proposer;
    address challenger;
}

struct DeployInput {
    Roles roles;
    uint32 basefeeScalar;
    uint32 blobBasefeeScalar;
    uint256 l2ChainId;
    bytes startingAnchorRoots;
    string saltMixer;
    uint64 gasLimit;
    uint32 disputeGameType;
    bytes32 disputeAbsolutePrestate;
    uint256 disputeMaxGameDepth;
    uint256 disputeSplitDepth;
    uint64 disputeClockExtension;
    uint64 disputeMaxClockDuration;
}

function upgrade(ISystemConfig[] _systemConfigs, IProxyAdmin[] _proxyAdmins, NewChainConfig[] _newConfigs) public;

For each chain successfully upgraded, the following event is emitted:

event Upgraded(uint256 indexed l2ChainId, SystemConfig indexed systemConfig, address indexed upgrader);

This method reverts if the upgrade is not successful for any of the chains.

Implementation

The high level logic of the upgrade method is as follows:

  1. The Upgrade Controller Safe will DELEGATECALL to the OPCM.upgrade() method.
  2. For each _systemConfig, the list of addresses in the chain is retrieved.
  3. For each address, a two step upgrade is used where:
    1. the first upgrade is to an InitializerResetter which resets the initialized value.
    2. the implementation is updated to the final address and upgrade() is called on that address.

This approach requires that all contracts have an upgrade() function which sets the initialized value to true. The upgrade function body should be empty unless it is used to set a new state variable added to that contract since the last upgrade.

IsthmusConfig struct

This struct is used to pass the new chain configuration to the upgrade method, and so it will vary for each release of the OP Contracts Manager, based on what (if any) new parameters are added.

In practice, this struct is likely to be have a unique name for each release of the OP Contracts Manager.

By way of example, if an upgrade is adding a new variable address foo to the SystemConfig contract, for an upgrade named Example, the struct could have the following definition:

struct IsthmusConfig {
    uint32 public operatorFeeScalar;
    uint64 public operatorFeeConstant;
}

Requirements on the OP Chain contracts

In general, all contracts used in an OP Chain SHOULD be proxied with a single shared implementation. This means that all values which are not constant across OP Chains SHOULD be held in storage rather than the bytecode of the implementation.

Any contracts which do not meet this requirement will need to be deployed by the upgrade() function, increasing the cost and reducing the number of OP Chains which can be atomically upgraded.

Adding game types

Because different OP Chains within a Superchain may use different dispute game types, and are expected to move from a permissioned to permissionless game over time, an addGameType() method is provided to enable adding a new game type to multiple games at once.

Interface

addGameType

The addGameType method is used to orchestrate the actions required to add a new game type to one or more chains.

struct PermissionlessGameConfig {
  bytes32 absolutePrestate;
}

function addGameType(ISystemConfig[] _systemConfigs, PermissionlessGameConfig[] _newGames) public;

Implementation

The high level logic of the addGameType method is as follows (for each chain):

  1. Deploy and initialize new DelayedWethProxy for the new game type, reusing the existing implementation
  2. Deploy a new FaultDisputeGame contract. The source of the constructor args is indicated below this list, the value which is not available onchain is the absolutePrestate.
  3. Calls upgrade() on the AnchorStateRegistry to set the new game type to add a new entry to the anchors mapping. The upgrade() method should revert if it would overwrite an existing entry.
  4. Read the DisputeGameFactory address from the SystemConfig.
  5. Call DisputeGameFactory.setImplementation() to register the new game.
NameTypeDescriptionSource
gameTypeuint32Constant value of 1 indicating the game typeHardcoded constant
absolutePrestatebytes32Initial state of the gameInput in PermissionlessGameConfig struct
maxGameDepthuint256Maximum depth of the game treeCopied from existing PermissionedGame
splitDepthuint256Depth at which the game tree splitsCopied from existing PermissionedGame
clockExtensionuint64Time extension granted for movesCopied from existing PermissionedGame
maxClockDurationuint64Maximum duration of the game clockCopied from existing PermissionedGame
vmaddressVirtual machine contract addressCopied from existing PermissionedGame
wethaddressAddress of the newly deployed DelayedWeth contractNewly deployed contract
anchorStateRegistryaddressRegistry contract addressCopied from existing PermissionedGame
l2ChainIduint256Chain ID of the L2 networkCopied from existing PermissionedGame

Security Considerations

Chain ID Source of Truth

One of the implicit restrictions on chain ID is that deploy can only be called once per chain ID, because contract addresses are a function of chain ID. However, future versions of OP Contracts Manager may:

  • Change the Proxy code used, which would allow a duplicate chain ID to be deployed if there is only the implicit check.
  • Manage upgrades, which will require "registering" existing pre-OP Contracts Manager chains in the OP Contracts Manager. Registration will be a privileged action, and the superchain registry will be used as the source of truth for registrations.

This means, for example, if deploying a chain with a chain ID of 10—which is OP Mainnet's chain ID—deployment will execute successfully, but the entry in OP Contracts Manager may be overwritten in a future upgrade. Therefore, chain ID uniqueness is not enforced by the OP Contracts Manager, and it is strongly recommended to only use chain IDs that are not already present in the ethereum-lists/chains repository.

Chain ID Frontrunning

Contract addresses for a chain are a function of chain ID, which implies you can counterfactually compute and use those chain addresses before the chain is deployed. However, this property should not be relied upon—new chain deployments are permissionless, so you cannot guarantee usage of a given chain ID, as deploy transactions can be frontrun.

Chain ID Value

While not specific to OP Contracts Manager, when choosing a chain ID is important to consider that not all chain IDs are well supported by tools. For example, MetaMask only supports chain IDs up to 4503599627370476, well below the max allowable 256-bit value.

OP Contracts Manager does not consider factors such as these. The EVM supports 256-bit chain IDs, so OP Contracts Manager sticks with the full 256-bit range to maximize compatibility.

Proxy Admin Owner

The proxy admin owner is a very powerful role, as it allows upgrading protocol contracts. When choosing the initial proxy admin owner, a Safe is recommended to ensure admin privileges are sufficiently secured.

Safely using DELEGATECALL

Because a Safe will DELEGATECALL to the upgrade() and addGameType() methods, it is critical that no storage writes occur. This should be enforced in multiple ways, including:

  • By static analysis of the upgrade() and addGameType() methods during the development process.
  • By simulating and verifying the state changes which occur in the Upgrade Controller Safe prior to execution.

Atomicity of upgrades

Although atomicity of a superchain upgrade is not essential for many types of upgrade, it will at times be necessary. It is certainly always desirable for operational reasons.

For this reason, efficiency should be kept in mind when designing the upgrade path. When the size of the superchain reaches a size that nears the block gas limit, upgrades may need to be broken up into stages, so that components which must be upgrade atomically can be. For example, all OptimismPortal contracts may need to be upgraded in one transaction, followed by another transaction which upgrades all L1CrossDomainMessenger contracts.