Skip to content
This repository has been archived by the owner on Jul 5, 2024. It is now read-only.

Latest commit

 

History

History
90 lines (71 loc) · 6.77 KB

evm-proof.md

File metadata and controls

90 lines (71 loc) · 6.77 KB

EVM Proof

The EVM proof argues the transition of state trie root is valid by verifying all the included transactions in a block have correct execution results.

EVM circuit re-implements the EVM, but in a perspective of verification, which means the prover can help provide hints as long as it does not contradict the result. For example, the prover can give hints about if this call is going to revert or not, or if this opcode encounters an error or not, then the EVM circuit can verify that execution results is correct.

The included transactions in a block could be simple ether transfers, contract creations, or contract interactions, and because each transaction has variable execution trace, we can't have fixed circuit layout to verify specific logic in specific region, we instead need a chip to be able to verify all possible logics, and this chip repeats itself to fill the whole circuit.

Custom Types

We define the following Python custom types for type hinting and readability:

Name Type Description
GlobalCounter int Order of random access to BusMapping, which should be in sequence
BusMapping List[Tuple[int, ...]] List of random read-write access data in sequence with GlobalCounter as index

Random Accessible Data

In EVM, the interpreter has ability to do any random access to data like account balance, account storage, account transient storage or stack and memory in current scope, but it's hard for the EVM circuit to keep tracking these data to ensure their consistency from time to time. So EVM proof has the state proof to provide a valid list of random read-write access records indexed by the GlobalCounter as a lookup table to do random access at any moment.

We call the list of random read-write access records BusMapping because it acts like the bus in computer which transfers data between components. Similarly, read-only data are loaded as a lookup table into circuit for random access.

Target Index Accessible By Description
Block {enum} Read Block constant decided before executing the block
BlockHash {index} Read Previous 256 block hashes as an encoded word array
AccountNonce {address} Read, Write Account's nonce
AccountBalance {address} Read, Write Account's balance
AccountCodeHash {address} Read, Write Account's code hash
AccountStorage {address}.{key} Read, Write Account's storage as a key-value mapping
AccountTransientStorage {address}.{key} Read, Write Account's transient storage as a key-value mapping
Code {hash}.{index} Read Executed code as a byte array
Call {id}.{enum} Read Call's context decided by caller (includes EOA and internal calls)
CallCalldata {id}.{index} Read Call's calldata as a byte array (only for EOA calls)
CallSignature {id}.{index} Read Call's signature as a byte array (only for EOA calls)
CallState {id}.{enum} Read, Write Call's internal state
CallStateStack {id}.{index} Read, Write Call's stack as an encoded word array
CallStateMemory {id}.{index} Read, Write Call's memory as a byte array

Circuit Constraints

The repeated chip has 2 main states, one is call initialization, another is bytecode execution.

TODO

Account non-existence

The following opcodes contain special cases for non-existing accounts:

  • BALANCE
  • EXTCODEHASH
  • EXTCODECOPY
  • EXTCODESIZE
  • CALL
  • CALLCODE
  • DELEGATECALL
  • STATICCALL

The rest of the opcodes only deal with accounts that are known to exist (by previous conditions).

We encode the state of existence of an account with its value of the code_hash in the rw_table (managed by the State Circuit): code_hash = 0 means that the account doesn't exist, and code_hash != 0 means that the account exists. We can guarantee this fact with the following properties:

  • Every time an account is created, the code_hash is set from zero to a non-zero value.
    • By the read consistency guaranteed by the State Circuit, we know that if a code_hash is non-zero, the account must exist (because it has been created previously)
  • A code_hash read with value 0 is translated to a non-existence account proof in the MPT (via the state circuit).
    • From this we know that if a code_hash is read as zero, the account doesn't exist.
  • There are no other valid transitions of code_hash. In summary, the valid transitions are: 0->0, and 0->H (where H != 0). Note that we're not considering account destruction for now.

Since only code_hash encodes account existence, we must be careful to never read (lookup) another account property unless we have checked that the account exists. This is because the RwTable in the State Circuit has all the entries sorted by [Tag, FieldTag], so CodeHash, Nonce and Balance are treated independently, which means that a lookup to Nonce or Balance could succeed on a non-existing account if the account is created afterwards. For this reason we must guarantee that:

  • A Nonce and Balance lookup to an account must only be performed if we have previously verified that the account exists by reading its CodeHash and checking it to be non-zero. This check can be done in the same step (for opcodes that can deal with non-existing accounts), or in a previous step (as a precondition, for opcodes that work with the caller account).