Evaluate 4337 Module Changes For EntryPoint 0.7.0

[nlordell]

Evaluate the required Safe 4337 Module contract changes required for the EntryPoint version 0.7. This should use the “candidate” EntryPoint version that was submitted for audit.

The expected outcome of this issue is:

1. A detailed task summary of required code changes
2. List of additional supporting tasks required for the new module version related to a new release
3. Optional: Draft PR containing code and test changes

[mmv08]

There are no migration docs except the release notes, and the quality feels subpar, to be honest. It feels strange to be pushed to update, given that the documentation is insufficient. But anyway, I went through the release notes' changes one by one.

1. Simulation functions removed from deployed EntryPoint. - Irrelevant for accounts
2. Added delegateAndRevert() helper in EntryPoint to accommodate nodes without state overrides in eth_call and eth_estimateGas - Irrelevant for accounts
3. Added ERC-165 "supportsInterface" to the EntryPoint. - We could add an erc-165 check during module deployment, but I don't think it's necessary and the contract should be tested anyway.
4. Added a sample TokenPaymaster. - This should be relevant if we decide to go for a paymaster contract.
5. Added a 10% penalty charge for unused execution gas limit. - Should be relevant for our SDKs. It's interesting that they charge the penalty on all unused gas; I'd expect a certain threshold after which the penalty applies because gas calculations can sometimes be complicated.
6. Removed paymaster’s second call to postOp. - Irrelevant for us atm
7. Added validatePaymasterUserOp and postOp gas limits. - Basically, paymaster and account gas limits are separate now. So far the biggest change for us.
   
8. Added “gasPrice” parameter to paymaster’s postOp - Irrelevant for us atm
9. Separated offchain UserOperation and packed UserOperation onchain struct. - Not sure I fully understand this change, this seems to be related to 7)
10. Added a new optional account contract api method executeUserOp(). - This matches our method name but with different parameters (this one gets the hash and full user operation struct). I don't see much value in adding it because we do not access any of the struct values.
11. There are many new storage access rules during the validation step; so far, none of them seem to affect us, but they still have to be tested with the production bundler.

So, that makes the action items:

• Mandatory - use the new PackedUserOperation struct with packed gas fields Use the new PackedUserOperation struct with packed gas fields from entrypoint 0.7.0 #225
• Unrelated to 0.7.0 entrypoint changes - use the production entrypoint in integration tests Use the production entrypoint contract for testing #226
• Unrelated to 0.7.0 entrypoint changes - Refactor end-to-end tests to use the 0.7.0 entrypoint and the new bundler Update end-to-end tests with 0.7.0 entrypoint & corresponding bundler #227
• Question - add an erc-165 interface check during module deployment for the entrypoint address?
• Question - support the new executeUserOp method. We could include it for better developer tooling compatibility.

[nlordell]

add an erc-165 interface check during module deployment for the entrypoint address?

Is this required by the IAccount interface?

support the new executeUserOp method. We could include it for better developer tooling compatibility.

What does this entail? Just changing the signature of the existing executeUserOp function?

[mmv08]

Is this required by the IAccount interface?

No, it's not about the account, it's about the entrypoint contract.

What does this entail? Just changing the signature of the existing executeUserOp function?

I would not change it but add a new method, I have a gut feeling the new method would use more gas due to more complicated calldata encoding.