new(tests): Add EOF validation tests for stack underflow

[chfast]

🗒️ Description

Port ethereum/tests EOF validation tests for stack underflow.

🔗 Related Issues

✅ Checklist

• All: Set appropriate labels for the changes.
• All: Considered squashing commits to improve commit history.
• All: Added an entry to CHANGELOG.md.
• All: Considered updating the online docs in the ./docs/ directory.
• Tests: All converted JSON/YML tests from ethereum/tests have been added to converted-ethereum-tests.txt.
• Tests: A PR with removal of converted JSON/YML tests from ethereum/tests have been opened. delete converted EOF stack underflow tests tests#1476
• Tests: Included the type and version of evm t8n tool used to locally execute test cases: e.g., ref with commit hash or geth 1.13.1-stable-3f40e65.
• Tests: Ran mkdocs serve locally and verified the auto-generated docs for new tests in the Test Case Reference are correctly formatted.

[marioevz]

A few comments, thanks!

[marioevz]

I would add a duplicate to this case but with max_stack_height=2, just for it to match the code_outputs=2 of the second section.

[marioevz]

Just wanted to confirm that the purpose of this test is to have an underflow in both branches of the RJUMPI, in contrast to the one below where only one of the branches has the underflow.

[marioevz]

Same comment from first test, we could add a max_stack_height=5 variant.

[marioevz]

Might be worth it to add another test case where the underflow happens in the no-jump branch:

code=Op.PUSH0
                    + Op.PUSH1[0]
                    + Op.PUSH1[0]
                    + Op.PUSH1[0]
                    + Op.RJUMPI[1]
                    + Op.POP
                    + Op.CALLF[1]
                    + Op.STOP,

[marioevz]

Suggested change

	code=Op.ADD + Op.STOP,
	code=Op.POP + Op.STOP,

We could also add another test that uses an opcode that barely produces the underflow vs one that underflows by two items (and pushes one).

max_stack_height could also use more variants here, e.g. 1 and 2 in this case, and 0 and 1 in the case of Op.POP for example.

[pdobacz]

can we just iterate the opcodes and use each and every one which requires non-zero stack? Can also boost this to do inputs - 1 PUSH0s preceding the offending opcode and test all

[chfast]

This is exactly what I wanted to do later. But as it was already noted, I will do it up front and then we can remove this test.

The exhaustive test should take all opcodes with static non-zero stack height required?

[pdobacz]

The exhaustive test should take all opcodes with static non-zero stack height required?

ah yes, zero-stack height required cannot overflow. I've meant to say you can either test underflow with 0 items or with < inputs items on stack

[marioevz]

Same comment as in previous file, we could add a variant that underflows in the no-jump case.

[marioevz]

Same comment as previous files, we could add a variant where the underflow happens in the no-jump branch.

[pdobacz]

I second Mario's comments + one suggestion