EIP-2935: Save historical block hashes in state

[OlivierBBB]

EIP-2935: Save historical block hashes in state

Progress

• spec
• constraints
• tracer

Impact

• if dealt with, deal with it the same way as with Beacon root hash EIP-4788 (Cancun)

Impact on BLOCKHASH opcode

We could add

• a miscellaneous-row when unexceptional i.e. when
  • XAHOY ≡ false
• a storage-row when unexceptional and in range i.e. when
  • XAHOY ≡ false
  • CHECK_BLOCKHASH_AGAINST_STATE ≡ true

to the instruction processing of BLOCKHASH opcodes. This would have the following benefit: the output of BLOCKHASH would provably be the same from one conflation to another

We would add a OOB_INST_BLOCKHASH instruction along the misc row which does the following:

// OOB_INST_BLOCKHASH description

// HUB inputs
trusted:
    REL_BLOCK
    BLOCK_HASH_ARGUMENT_HI
    BLOCK_HASH_ARGUMENT_LO // together: ARG

prediction:
    NUMBER
    ARGUMENT_IN_RANGE
    CHECK_BLOCKHASH_AGAINST_STATE
    STORAGE_SLOT

// lookup to BLOCK_DATA module

[oob/REL_BLOCK, oob/NUMBER] --> [blc/REL_BLOCK, blc/NUMBER] // confirms NUMBER

// computations

tiny_block_number ≡ [256 > NUMBER]
argument_in_range ≡
    if tiny_block_number ≡ true  : [ARG < NUMBER]
    if tiny_block_number ≡ false : [ARG < NUMBER] ∧ [NUMBER - 256 ≤ ARG]
    ∴ ARGUMENT_IN_RANGE  ←  argument_in_range
storage_slot ≡
    if argument_in_range ≡ true  : ARG % 8191
    if argument_in_range ≡ false : <undefined>
    ∴ STORAGE_SLOT  ←  storage_slot
requested_block_is_prague_or_later ≡ [ARG ≥ constants/LINEA_PRAGUE_HARDFORK_BLOCK_NUMBER]
check_blockhash_against_state ≡ argument_in_range ∧ requested_block_is_prague_or_later
    ∴ CHECK_BLOCKHASH_AGAINST_STATE  ←  check_blockhash_against_state

We would still use the BLOCKHASH module as we do today to retrieve the output of the BLOCKHASH opcode. We may (optionally) make the lookup conditional to ARGUMENT_IN_RANGE ≡ true and simplify the BLOCKHASH module. But tbh there is no reason to do so.

Test vectors