Regression: zero-root output when running statetest

[holiman]

There seems to be a regression, the bug holiman/goevmlab#126 (comment) which was fixed in #1169.

When running a statetest, the stateroot is output twice, but the first is all zeroes:

root@b0da57b9f571:/# /revme statetest --json /fuzztmp/00000000-mixed-0.json 1>/dev/null
{"pc":0,"op":96,"gas":"0x79bcd8","gasCost":"0x3","stack":[],"depth":1,"returnData":"0x","refund":"0x0","memSize":"0","opName":"PUSH1"}
{"pc":2,"op":96,"gas":"0x79bcd5","gasCost":"0x3","stack":["0xa6"],"depth":1,"returnData":"0x","refund":"0x0","memSize":"0","opName":"PUSH1"}
{"pc":4,"op":96,"gas":"0x79bcd2","gasCost":"0x3","stack":["0xa6","0x19"],"depth":1,"returnData":"0x","refund":"0x0","memSize":"0","opName":"PUSH1"}
{"pc":6,"op":96,"gas":"0x79bccf","gasCost":"0x3","stack":["0xa6","0x19","0x7e"],"depth":1,"returnData":"0x","refund":"0x0","memSize":"0","opName":"PUSH1"}
{"pc":8,"op":96,"gas":"0x79bccc","gasCost":"0x3","stack":["0xa6","0x19","0x7e","0x78"],"depth":1,"returnData":"0x","refund":"0x0","memSize":"0","opName":"PUSH1"}
{"pc":10,"op":96,"gas":"0x79bcc9","gasCost":"0x3","stack":["0xa6","0x19","0x7e","0x78","0xa3"],"depth":1,"returnData":"0x","refund":"0x0","memSize":"0","opName":"PUSH1"}
{"pc":12,"op":96,"gas":"0x79bcc6","gasCost":"0x3","stack":["0xa6","0x19","0x7e","0x78","0xa3","0x24"],"depth":1,"returnData":"0x","refund":"0x0","memSize":"0","opName":"PUSH1"}
{"pc":14,"op":162,"gas":"0x79bcc3","gasCost":"0x58e","stack":["0xa6","0x19","0x7e","0x78","0xa3","0x24","0x2f"],"depth":1,"returnData":"0x","refund":"0x0","memSize":"0","opName":"LOG2"}
{"pc":15,"op":143,"gas":"0x79b735","gasCost":"0x3","stack":["0xa6","0x19","0x7e"],"depth":1,"returnData":"0x","refund":"0x0","memSize":"96","opName":"DUP16","error":"StackUnderflow"}
{"stateRoot":"0x0000000000000000000000000000000000000000000000000000000000000000","output":"0x","gasUsed":"0x7a1200","pass":false,"fork":"Cancun"}
{"stateRoot":"0x867aa4fc1b93afe97a14837ab60a460bc512983a07dc77dcce44ff6c69ce263f","logsRoot":"0x1dcc4de8dec75d7aab85b567b6ccd41ad312451b948a7413f0a142fd40d49347","output":"0x","gasUsed":8000000,"pass":false,"errorMsg":"logs root mismatch: expected 0x0000000000000000000000000000000000000000000000000000000000000000, got 0x1dcc4de8dec75d7aab85b567b6ccd41ad312451b948a7413f0a142fd40d49347","evmResult":"Ok","postLogsHash":"0x1dcc4de8dec75d7aab85b567b6ccd41ad312451b948a7413f0a142fd40d49347","fork":"CANCUN","test":"00000000-mixed-0","d":0,"g":0,"v":0}

I don't know the exact git-commit it was built on, but it was built around 8 hours ago, so presumably af5665176271c87c9ff0600f6d8432d117012345

[holiman]

Tested just now, seems to be fixed?

[holiman]

Tested just now, seems to be fixed?

Nah, my bad, it's still broken. I did, however, git bisect it for you, and a201a2d is the culprit:

user@debian-work:~/workspace/revm$ git bisect good
a201a2d92c4a41f6530fe2b84ec23467255c0636 is the first bad commit
commit a201a2d92c4a41f6530fe2b84ec23467255c0636
Author: A_A <21040751+Otto-AA@users.noreply.github.com>
Date:   Fri Mar 29 20:32:35 2024 +0100

    feat: TracerEip3155 optionally traces memory (#1234)
    
    * feat: TracerEip3155 optionally traces memory
    
    BREAKING CHANGE: the TracerEIP3155 constructor takes an additional
    "include_memory" flag
    
    * perf: use hex::encode_prefixed for TracerEip3155
    
    * refactor: use builder pattern for TracerEip3155
    
    BREAKING CHANGE: update constructor syntax for TracerEip3155
    
    * refactor: use String for TracerEip3155 memory

 bins/revme/src/cmd/statetest/runner.rs |  7 ++-----
 crates/revm/src/inspector/eip3155.rs   | 29 +++++++++++++++++++++++++----
 examples/db_by_ref.rs                  |  2 +-
 examples/generate_block_traces.rs      |  2 +-

Please fix it, it's annoying not being able to fuzz with revm

[rakita]

@holiman It is now fixed, thank you for finding the bug commit!