feat(tests): add test for worst-case prefetcher performance

[nerolation]

🗒️ Description

Add test_sload_bloated_prefetch_miss benchmark that defeats client prefetchers by making each transaction's SLOAD range depend on state written by its predecessor. A minimal first tx writes an offset to slot 0 via calldata; subsequent max-gas txs read the previous offset, SLOAD from it, and write a new offset for the next tx.

🔗 Related Issues or PRs

N/A.

✅ Checklist

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

  just static

Cute Animal Picture

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 86.25%. Comparing base (a48e0b3) to head (f7d22f3).
⚠️ Report is 7 commits behind head on forks/amsterdam.

Additional details and impacted files

@@                 Coverage Diff                 @@
##           forks/amsterdam    #2657      +/-   ##
===================================================
+ Coverage            86.24%   86.25%   +0.01%     
===================================================
  Files                  599      599              
  Lines                36984    37032      +48     
  Branches              3795     3795              
===================================================
+ Hits                 31895    31943      +48     
  Misses                4525     4525              
  Partials               564      564              

Flag	Coverage Δ
unittests	86.25% <ø> (+0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[jochem-brouwer]

Some comments 😄 👍

[jochem-brouwer]

I somehow did not see that this approach solves the same problem but with less gas (I did not realize this also tricks the prefetcher). One final comment for verification.

[jochem-brouwer]

LGTM! Thanks a lot!!

[jochem-brouwer]

I am having troubles filling these payloads and need to investigate more (can´t seem to retract my previous review 🤔 )

EDIT; this worked

[LouisTsai-Csie]

Leave some style suggestion comment, please take a look!

[jochem-brouwer]

This will likely have the problem that either this code will not fit in initcode, or the predeploy code to execute will use more than 2**24 gas. @LouisTsai-Csie we should likely create a general helper here for the predeploy in execute

[LouisTsai-Csie]

We already have this helper, it is not clear though:

execution-specs/tests/benchmark/stateful/bloatnet/test_single_opcode.py

Line 1262 in 47782b3

def create_sstore_executor(

In the test_sstore_variants benchmark, we delegate to this helper then initialize the storage, later execute the benchmark code. This would not go out of gas, but not sure if it would take too much time or not.

[jochem-brouwer]

Here is fixture output of the sload_bloated_multi_contract. I'll check this soon, wanted to dump this progress. The good point here is that the fixture fills succesfully so it passes any checks made on the output (BAL checks). Will do similar to the other test also.

sload_bloated_multi_contract.json

[LouisTsai-Csie]

Thanks for the update, i did a final check, after these changes let's merge.

[nerolation]

Everything addressed. Ready for one last review!

[LouisTsai-Csie]

Thank you! I will merge it for now, and refactor in PR #2672, as the current storage initialization would break at high gas configuration.