fix(tests): fix storage slot of actual value for eip7516_blobgasfee tests in execute remote

[acuarica]

🗒️ Description

This PR fixes the storage slot of the actual value in the test.

In execute remote mode, it used NUMBER as the storage slot, but the expected value is stored in slot 1. This causes problems when running tests with execute remote where NUMBER cannot be ensured to be 1.

🔗 Related Issues or PRs

Fixes #2255.

✅ Checklist

• All: Ran fast tox checks to avoid unnecessary CI fails, see also Code Standards and Enabling Pre-commit Checks:

  uvx --with=tox-uv tox -e lint,typecheck,spellcheck,markdownlint

• All: PR title adheres to the repo standard - it will be used as the squash commit message and should start type(scope):.
• All: Considered adding an entry to CHANGELOG.md.
• All: Considered updating the online docs in the ./docs/ directory.
• All: Set appropriate labels for the changes (only maintainers can apply labels).
• Tests: Ran mkdocs serve locally and verified the auto-generated docs for new tests in the Test Case Reference are correctly formatted.
• Tests: For PRs implementing a missed test case, update the post-mortem document to add an entry the list.
• Ported Tests: All converted JSON/YML tests from ethereum/tests or tests/static have been assigned @ported_from marker.

[LouisTsai-Csie]

This fix resolves the issue in execute mode, but it will fail under the fill mode in our framework. Please take a look at the failing case test_blobbasefee_during_fork, where the block number is used as the storage key.

 pre-set storage just to make sure we detect the change
        code_caller_post_storage[block_number] = 0 if timestamp < 15_000 else 1

That is the reason for the failing CI, you could also run it locally via

uv run fill -v tests/cancun/eip7516_blobgasfee/test_blobgasfee_opcode.py::test_blobbasefee_during_fork --fork Osaka --clean

[acuarica]

Hi @LouisTsai-Csie, thanks for taking a look. I've managed to solve the issue essentially by parameterizing the expected value slot. Happy to explore other solutions if required.

[acuarica]

However, I had to skip this gas validation, otherwise the test will fail in execute mode

execution-spec-tests/src/ethereum_test_execution/transaction_post.py

Lines 69 to 71 in ea972f4

	# Perform gas validation if required for benchmarking
	# Ensures benchmark tests consume exactly the expected gas
	if not self.skip_gas_used_validation and self.expected_benchmark_gas_used is not None:

Am I missing something?

[LouisTsai-Csie]

Gas usage validation is intended for benchmark mode (see docs).
You can temporarily use skip_gas_used_validation to bypass it, we’ll later limit this check to benchmark mode after this wip issue is done

[LouisTsai-Csie]

For the failing evmone CI issue, I checked the report and found an LBC. The root cause is the use of a different storage key. In the previous setup, Op.NUMBER was used to manage the storage slot, whereas the latest approach instead uses slot 1. As a result, evmone detects the discrepancy and reports it as an LBC. This is fine so nothing else needs to be changed.

Report data

LBC Found

[LouisTsai-Csie]

A recent merged PR might help you with the gas validation issue. For non benchmark mode, it would not apply the gas validation

[acuarica]

Thanks for clarifying #2266 (comment), and in general for taking the time to review this.

[marioevz]

Thanks for this! Just one suggestion that might simplify and fix the issues found in the previous review.

[LouisTsai-Csie]

The CI is failing but i think it's fine, it is due to the changed testing method.

[marioevz]

LGTM, thanks!

Coverage CI issue is because we swapped opcodes, so it's ok to merge.