Issue with the `blockhash` instruction at the boundary

[0xOmarA]

It looks like we have an issue with the blockhash instruction around the boundary of it's limits which is 256 blocks away from the tip of the blockchain. We've seen this bug in some of the ML tests where the block hash can be obtained for numbers smaller than 256 but it starts failing around 255 and 256.

Tests failing with this bug are:

• simple/yul_instructions/blockhash.sol::3::E-
• simple/yul_instructions/blockhash.sol::3::Y+
• simple/yul_instructions/blockhash.sol::3::Y-
• simple/yul_instructions/blockhash.sol::3::E+
• simple/yul_instructions/blockhash.sol::4::E+
• simple/yul_instructions/blockhash.sol::4::Y+

[pgherveou]

mmm all these tests seems to pass for me with a blockchain with more than 256 blocks built already

[pgherveou]

this test that I added seems to pass as well

https://github.com/paritytech/evm-test-suite/blob/9959fa1f0fbee04c8f590a5c1cf85595ed229ca4/src/others.test.ts#L264-L313

[0xOmarA]

Hmmm... I think that these tests fail indeterminately. Looking at the CI runs it seems to me that they sometimes work and sometimes fail.

[0xOmarA]

I will try to verify if these were fixed perhaps in one of the PRs since I can't see these failures in this PR paritytech/polkadot-sdk#10213

[pgherveou]

@0xOmarA please do, moving to done for now move it back if it's not fixed

[0xOmarA]

Re opening this after a discussion with PG since it looks like this is still an issue