feat: mpt feature

[greged93]

Adds a "mpt" feature, allowing to feature gate the packing and unpacking of the nibbles, as well as correctly compute the genesis block state root.

Resolves #121 and resolves #117

[frisitano]

Why do we gate the packing here? I thought we didn't need the scroll feature to be activated when we are using the mpt?

[greged93]

The issue is that a lot of the scroll crates need the scroll feature to be active.
Additionally, we need the revm crate to use the scroll feature (for correct execution) as well as the reth-primitive (for the additional field on the receipt). This caused a lot of compilation issue as a lot of the crates ended up needing the scroll feature across the code base. I figured it was easier to keep the scroll feature (technically we are executing scroll mainnet so I think this makes sense) but avoiding incorrect nibble packing or unpacking. Let me know if you think there is an easier way.

[frisitano]

Understood, that makes sense. The alternative would be to put non-standard bmpt logic behind a bmpt feature gate. Let's go with the path of least resistance, whatever makes the solution simpler. I would like to remove support for additional fields in the scroll account and modifications for the key hashing/traversal as quickly as possible (probably completion of milestone 2) and then we can look to refactor this.

[frisitano]

Looks good! Has this been tested? I propose we add some test evidence to this PR before merge. A proposal would be to sync and compare the state root against L2 geth for blocks 1000, 10000 and 50000. What do you think?

[greged93]

@frisitano below is the evidence for correct sync for blocks 1000, 10000 and 50000.

Block 1000

l2geth

Reth

Block 10000

l2geth

Reth

Block 50000

l2geth

Reth

[frisitano]

Thanks. LGTM!

[frisitano]

LGTM