[Core Protocol] Can‘t find the interacting case of step() function in the fault proof distribute game

[joohhnnn]

Did you check the documentation?

• Yes

Did you read the specs?

• Yes
• No

Did you check for duplicate questions?

• Yes

Issue Description

Issue Background:
In the fault proof system, when a dispute reaches the finest granularity, the step() function in the Game contract will be called. This function indirectly invokes the MIPS virtual machine for verification. Before this verification process begins, it is necessary to call the addLocalData() function within the Game contract. This function adds data required by the MIPS virtual machine during the verification process to the PreImageOracle contract.

Issue:
I'm looking to understand the step() function's usage in real-world production or testing environments, but I haven't found any use cases, not even on testnets. Additionally, I've come across several PreImageOracle contracts deployed at different times, and none of them seem to have a direct or indirect loadLocalData() function call.

Here are some PreImageOracle contracts I looked at on various testnets, and none of them seem to have the relevant function calls:

Contract created 92 days ago
Contract created 55 days ago
Contract created 36 days ago

Additional Information

No response

Feedback

No response

[sbvegan]

Hey @joohhnnn, I flagged this internally and hope to have an answer for you soon.

[ajsutton]

It's not necessarily required to call loadLocalData() or to load any data into the preimage oracle when calling step - it's only required if the single MIPS CPU instruction being executed in the step call is actually reading preimage data and then only that one preimage needs to be uploaded. op-challenger will automatically upload the required preimage data as part of its call to step when necessary (https://github.com/ethereum-optimism/optimism/blob/78e1084ec14d3003cb9e546b9eb5a22db7408ac2/op-challenger/game/fault/responder/responder.go#L90-L129 is probably the best place to start for that code).

I believe it costs around 600ETH in total bonds to get a dispute game down to where the step function is called so it is an exceptionally unusual thing to have happen. You're right that it hasn't happened on any of the DisputeGameFactory instances used by the OptimismPortal. There are a number of cases around on public chains but I'm not sure I could find them again now and I'm not sure if they wound up needing to call addLocalData. We do have e2e tests that do this on local devnets to test it - including specifically targeting steps that require specific types of preimage data. We can also use eth_Call to invoke step without having to push a dispute game all the way down.