Migrating to permissionless fault proofs on OP Stack

A high-level guide for transitioning from permissioned to permissionless fault proofs on an OP Stack.

This guide provides a high-level overview for chain operators looking to transition their OP Stack from permissioned to permissionless fault proofs. It’s designed to be accessible for technical decision makers while providing sufficient detail for implementation teams.

Overview

The OP Stack architecture uses Fault Proofs to ensure the validity of withdrawals from L2 to L1. Transitioning from permissioned to permissionless proofs represents a significant security upgrade, allowing any participant to propose and challenge state output roots. Permissioned games previously relied on a single trusted validator, this is typically the proposer which is configured in the PermissionedDisputeGame, and is usually the network’s only sequencer.

This migration involves several key components:

• Configuring security-critical dispute monitoring services
• Deploying and configuring smart contracts using op-deployer
• Testing the new system before activation
• Setting the respected game type to permissionless fault proofs, specifically using the FaultDisputeGame

Prerequisites

Before beginning this transition, your chain should:

• Be running a standard OP Stack implementation
• It’s recommended to use the latest contracts version, minimum required v5.0.0.
• Be operating with the required infrastructure services including op-challenger and op-dispute-mon.

Migration steps

The process of migrating from permissioned to permissionless fault proofs involves four main phases: configuring off-chain dispute components, deploying the necessary smart contracts, testing the system thoroughly, and finally switching the chain to use permissionless proofs. Each step builds on the previous one to ensure a smooth and secure transition.

Let’s begin with configuring the dispute components that will interact with the new permissionless game.

1. Configure the dispute components

The op-challenger and op-dispute-mon services are critical security components that participate in the dispute game process to challenge invalid proposals and monitor active games.

Upgrade to the latest op-challenger

Upgrade to the latest release, which contains important improvements to simplify the upgrade process.

You can use the official Docker image for reliability and ease of deployment:

# Pull a specific stable release version
docker pull us-docker.pkg.dev/oplabs-tools-artifacts/images/op-challenger:<latest_version>

Then run the image, for example:

# Run with all required flags
docker run -d --name op-challenger \
  -e OP_CHALLENGER_TRACE_TYPE=permissioned,cannon \
  -e OP_CHALLENGER_PRESTATES_URL=<YOUR_PRESTATES_URL> \
  -e OP_CHALLENGER_L1_ETH_RPC=<YOUR_L1_RPC_URL> \
  -e OP_CHALLENGER_GAME_FACTORY_ADDRESS=<YOUR_DISPUTE_GAME_FACTORY> \
  -e OP_CHALLENGER_PRIVATE_KEY=<YOUR_PRIVATE_KEY_OR_USE_WALLET_SIGNER> \
  -e OP_CHALLENGER_NETWORK=<YOUR_NETWORK> \
  -v /path/to/local/prestates:/prestates \
  us-docker.pkg.dev/oplabs-tools-artifacts/images/op-challenger:latest

# Replace all placeholder values with your actual configuration

Replace <YOUR_PRESTATES_URL> with your actual prestates URL.

If your deployment requires building from source, you can alternatively use:

git clone https://github.com/ethereum-optimism/optimism -b op-challenger/v1.3.3 --recurse-submodules
cd optimism
make op-challenger

Update network configuration

Configure op-challenger to load your chain configuration. Even if your chain is not included in the superchain-registry, you can specify a custom configuration:

# For chains in the registry
--network <chain-name>

# For chains not in the registry, provide a path to your rollup configuration
- `rollup.json` - Your rollup configuration
- `genesis-l2.json` - Your L2 genesis file

Enable cannon trace type

Configure op-challenger to support both permissioned and permissionless games by setting:

--trace-type permissioned,cannon

Or by setting the environment variable:

OP_CHALLENGER_TRACE_TYPE=permissioned,cannon

Configure prestates access

Replace the --cannon-prestate flag with --prestates-url, which points to a source containing all required prestates:

--prestates-url <URL_TO_PRESTATES_DIRECTORY>

The URL can use http, https, or file protocols. Each prestate should be named as <PRESTATE_HASH>.json or <PRESTATE_HASH>.bin.gz.

Building required prestates for chains not in the Superchain Registry

You’ll need to deploy two new dispute game contracts with the new absolute prestate:

1. FaultDisputeGame
2. PermissionedDisputeGame

The initial prestate used for permissioned games doesn’t include the necessary chain configuration for the Fault Proof System. The assumption is that the chain operator, the single permissioned actor, will not challenge their own games. So the absolute prestate on the initial PermissionedDisputeGame will never be used. When deploying a new chain, you must first deploy the L1 contracts and then generate the chain genesis file and rollup configuration files. These are inputs to the creation of the absolute prestate and this circular dependency is the reason chains cannot be deployed directly to the permissionless Fault Proof System.

For chains not in the Superchain Registry, you need to build custom prestates with your chain’s configuration:

The prestate will typically be generated at:

• op-program/bin/prestate.json (for older versions)
• op-program/bin/prestate.bin.gz (for intermediate versions)
• op-program/bin/prestate-mt64.bin.gz (for chains upgraded to Cannon MT64, starting from upgrade 14)

Post-upgrade 14, chains are expected to use prestate-mt64.bin.gz due to the Fault Proof VM contract upgrade to cannon-mt64. The older prestate.bin.gz will eventually be deprecated but is temporarily retained until all chains complete the upgrade.

Ensure sufficient funds for bonds

Bonds are required for both permissioned and permissionless games. However, with permissioned games, you typically don’t post claims regularly, making bond requirements less noticeable. In contrast, the challenger in permissionless games will frequently need to post bonds with each claim it makes. Therefore, ensure your challenger has sufficient funds available.

As a general guideline:

• Maintain a minimum balance of 50 ETH
• Have access to a large pool of ETH for potential attack scenarios
• Implement monitoring to ensure sufficient funds are always available

Ensure there is disk space for the challenger to use

op-challenger, particularly in permissionless mode, should have access to disk. About 50GB is enough to cover a couple of invalid_ games. Though the storage requirements will increase if the challenger needs to respond to more invalid dispute games. Nominally, the challenger does not use disk space as long as there aren’t any invalid proposals being made.

One way to gain more confidence that the op-challenger was configured correctly is to operate the op-challenger in “runner” mode. Using the same op-challenger configuration, invoke the op-challenger run-trace -run cannon subcommand. This will run op-challenger in a mode where it runs the op-program in a Cannon VM on live L2 blocks to ensure those were configured correctly. This command runs forever, checking every couple of L2 blocks. But it suffices to let it run until it has completed one loop and kill it. Wait you see Successfully verified output root in the logs before shutting it down. Any errors indicate a misconfiguration.

Set up op-dispute-mon

Ensure op-dispute-mon is properly configured by following the steps in the documentation.

2. Deploy and configure smart contracts using OPCM

This section requires privileged actions by the ProxyAdminOwner and the Guardian role.

Understanding ProxyAdmin Owner and Guardian roles

This migration requires actions by privileged roles in your system:

• The ProxyAdmin Owner has the authority to upgrade proxy contracts.
• The Guardian has emergency powers like pausing withdrawals and changing the respected game type.

For detailed information about privileged roles and their security implications, refer to the privileged roles documentation.

Adding the PermissionlessDisputeGame to a chain

To enable the permissionless dispute game, you must call the addGameType()function on the OPChainManager (OPCM) contract.

This function orchestrates all the necessary steps to register the new dispute game type for your chain.

This method will:

1. Deploy the FaultDisputeGame contract
2. Setup the DelayedWethProxy for the new game
3. Reinitialize the AnchorStateRegistry to add the new game type.

See a high‐level implementation from this docs or this superchain-ops template.

3. Testing off-chain agents

After you’ve set the permissionless FaultDisputeContract implementations on the DisputeGameFactory and before you set the respected game type to it (game type 0), you can test op-challenger and op-dispute-mon to ensure they are working correctly with permissionless games.

There are a number of useful op-challenger subcommands that can be used for testing, particularly list-games, list-claims and create-game. See the README and op-challenger --help output for further details. The two tests below are basic sanity tests:

Test defending valid proposals

Create a valid proposal using the permissionless game type 0:

1. Ensure the proposal is from a block at or before the safe head:

   cast block --rpc-url <OP_GETH_ENDPOINT> safe
   
   

2. Get a valid output root (from op-node):

   cast rpc --rpc-url <OP_NODE_ENDPOINT> optimism_outputAtBlock \
     $(cast 2h <BLOCK_NUMBER>) | jq -r .outputRoot
   
   

3. Create a test game:

   ./op-challenger/bin/op-challenger create-game \
     --l1-eth-rpc=<L1_RPC_ENDPOINT> \
     --game-factory-address <DISPUTE_GAME_FACTORY_ADDR> \
     --l2-block-num <BLOCK_NUMBER> \
     --output-root <OUTPUT_ROOT> \
     <SIGNER_OPTIONS>
   
   

4. Verify:

   • op-challenger logs a message showing the game is in progress
   • op-challenger doesn’t post a counter claim (as this is a valid proposal)
   • dispute-mon includes the new game with status="agree_defender_ahead"

Test countering invalid claims

Post an invalid counter claim to the valid proposal created above:

./op-challenger/bin/op-challenger move \
  --l1-eth-rpc <L1_RPC_ENDPOINT> \
  --game-address <GAME_ADDR> \
  --attack \
  --parent-index 0 \
  --claim 0x0000000000000000000000000000000000000000000000000000000000000000 \
  <SIGNER_OPTIONS>

Verify that op-challenger posts a counter-claim to the invalid claim. You can view claims using:

./op-challenger/bin/op-challenger list-claims \
  --l1-eth-rpc <L1_RPC_ENDPOINT> \
  --game-address <GAME_ADDR>

There should be 3 claims in the game after this test.

Switch to permissionless proofs

After completing all previous steps and verifying their successful operation, you need to update the respectedGameType in the AnchorStateRegistry. This requires execution through the appropriate privileged role (typically the Guardian).

You have two main options for executing this step:

Option 1: Execute using a multisig

If your privileged role (such as the Guardian) is controlled by a multisig or DAO governance system, use the provided JSON payload (input.json):

Show Create an `input.json` file with the transaction to update the `respectedGameType`:

{
  "chainId": "<YOUR_L1_CHAIN_ID>",
  "metadata": {
    "name": "Deputy Guardian - Enable Permissionless Dispute Game",
    "description": "This task updates the `respectedGameType` in the `AnchorStateRegistry` to `CANNON` (game type 0), enabling users to permissionlessly propose outputs as well as for anyone to participate in the dispute of these proposals. This action requires all in-progress withdrawals to be re-proven against a new `FaultDisputeGame` that was created after this update occurs."
  },
  "transactions": [
    {
      "metadata": {
        "name": "Update `respectedGameType` in the `AnchorStateRegistry`",
        "description": "Updates the `respectedGameType` to `CANNON` in the `AnchorStateRegistry`, enabling permissionless proposals and challenging."
      },
      "to": "<ASR_ADDRESS>",
      "value": "0x0",
      "data": "0x7fc485040000000000000000000000000000000000000000000000000000000000000000",
      "contractMethod": {
        "type": "function",
        "name": "setRespectedGameType",
        "inputs": [
          {
            "name": "_gameType",
            "type": "uint32"
          }
        ],
        "outputs": [],
        "stateMutability": "nonpayable"
      },
      "contractInputsValues": {
        "_gameType": "0"
      }
    }
  ]
}

• Submit this JSON payload through your interface (e.g., Safe transaction builder).
• Simulate this transaction using Tenderly before execution to ensure the expected state changes:
  • The respectedGameType in the AnchorStateRegistry should change from 1 (PERMISSIONED) to 0 (CANNON).

---

Option 2: Direct execution via cast (Forge CLI)

Alternatively, if you’re executing directly from a single-privileged wallet or want quicker execution, use the following cast commands:

1. First, encode the transaction calldata using cast calldata:

CALLDATA=$(cast calldata "setRespectedGameType(uint32)" 0)

2. Send the transaction:

# Execute the transaction
cast send --rpc-url <YOUR_RPC_URL> --private-key <YOUR_PRIVATE_KEY> <ASR_ADDRESS> "$CALLDATA"

3. After execution, verify the respected game type:

cast call --rpc-url <YOUR_RPC_URL> <ASR_ADDRESS> "respectedGameType()(uint32)"
# The value should now be 0 (CANNON)

Post-execution configuration:

After updating, configure op-proposer to create proposals using the permissionless CANNON game type:

Via command-line argument:

--game-type 0

Or via environment variable:

OP_PROPOSER_GAME_TYPE=0

This action requires all in-progress withdrawals to be re-proven against a new FaultDisputeGame created after this update occurs.

Next steps

• For more detail on deploying new dispute games with OPCM, see the docs.
• Deploy new dispute games with OPCM via this tutorial.
• Generate an absolute prestate using the absolute prestate guide.
• Understand fault proofs in the Fault proofs explainer.