Optimism execution

[Wollac]

When I tried to use the Optimism execution behind the optimism feature flag of version v3.5.0, I ran into the following problems:

• Handler::optimism() is never used. So even if env.cfg.optimism is true, the Optimism specific fee handling is not applied.
• The Base Fee Vault calculation in the Optimism handler uses the L1 base, but according to the L2 Execution Engine spec, I was under the impression that env.block.basefee * (gas.spend() - gas_refund) should be used instead.
• When the optimism feature is enabled while env.cfg.optimism is false for Ethereum transactions, the tx_l1_cost is still computed, which can lead to a panic as obviously the enveloped_tx is not provided in this case.

If the above observations are indeed correct and it would be helpful, I can provide a PR for these issues.

[refcell]

Thank you for reporting this! Will take a look, reproduce, and get back to you!

As a side note, if you make a PR, it would be preferable to split it up to make reviewing and adding the relevant unit tests easier.

[rakita]

@Wollac thank you!

Handler::optimism() should have been used if optimism flag is true here:

revm/crates/revm/src/evm_impl.rs

Line 418 in 02d85ac

handler: Handler::mainnet::<GSPEC>(),

For the future, the idea is to create the RevmBuilder that would allow setting either of the handler (Still need to flush some thing for it), but for now, just changing the code to ::optimism() should do the work.

The Second point I will leave for @refcell, for the third point we definitely should not panic.

[Wollac]

I have created a PR that addresses these issues. It has been validated in zeth(https://github.com/risc0/zeth/tree/feat/revm-optimism) against a handful of Mainnet Ethereum and Optimism blocks. After these changes, blocks can now be constructed that lead to the correct state.

It does not yet include specific unit tests, as @refcell mentioned. None of the affected functions had tests before, so designing relevant tests would require more thought.

[refcell]

Left a couple of nits, but besides those, you're pr looks great! Really exciting that you were able to test this against Optimism Blocks!!

On the topic of unit tests, I think pushing those into other prs is totally reasonable.

[clabby]

Very exciting news that you guys are successfully executing Optimism blocks! Thanks a ton for fixing these up :)

[Wollac]

Fixed with #789