Open goodlyrottenapple opened 2 years ago
I'd try to minimize the spec again, which means doing a "run-lemma/done-lemma" style proof:
claim run-lemma(relevant-part-of-config) => done-lemma(what-you-expect-to-get)
You can then use https://github.com/runtimeverification/haskell-backend/wiki/Debugging#function-evaluation to see if that simplification rule is actually getting applied or not.
We need to get this minimized before we can figure out if it's a Haskell backend bug or not.
@ana-pantilie I managed to reproduce this error. backend version:
Kore version 0.60.0.0
Git:
revision: 74bc7a963a825d0f0f034b5d3412dc59448e51bb (dirty)
branch: master
last commit: Tue Feb 7 05:02:49 2023 -0700
In the following version of K and
kore-exec
:The following proof from the evm-semantics repo fails to discharge this final equality:
specifically, it gets stuck trying to prove
Given that the claim contains the following clause
and
wordpack.k
contains the following simplificationthe backend should be able to simplify the above goal. Even attempting to rewrite the above to use the
#Equals
(using a version of K with https://github.com/runtimeverification/haskell-backend/pull/3042) doesn't work.However, tweaking the above
#WordPackAddrUInt48UInt48
function symbol fromto
where the simplification rule now becomes
we now get
#Top
when running the modified proof.Attached below is the generated bug report. I tried minimizing the proof, but could not re-create the conditions of the proof, such that the prover would get stuck on this claim.
kevm-bug-flipper-bids-pass-rough.tar.gz