conrad-watt opened PR #3797 from patch-1
to main
:
A while ago, it was observed that the official reference interpreter has performance issues which limit its usefulness as a fuzzing oracle. I currently maintain a fork of the reference interpreter which implements some optimisations to avoid the quadratic behaviour above. The core of this interpreter is a verified definition generated from my mechanisation of Wasm's semantics.
I've been speaking to @cfallin, @alexcrichton, @abrown, and @fitzgen about the feasibility of using this interpreter as a fuzzing oracle, and they suggested I bring this idea up for wider discussion in a Wasmtime meeting.
conrad-watt edited PR #3797 from patch-1
to main
:
A while ago, it was observed that the official reference interpreter has performance issues which limit its usefulness as a fuzzing oracle. I currently maintain a fork of the reference interpreter which uses a more efficient representation of Wasm's runtime state, avoiding the quadratic behaviour above. The core of this interpreter is a generated from and verified against my mechanisation of Wasm's semantics.
I've been speaking to @cfallin, @alexcrichton, @abrown, and @fitzgen about the feasibility of using this interpreter as a fuzzing oracle, and they suggested I bring this idea up for wider discussion in a Wasmtime meeting.
tschneidereit submitted PR review.
tschneidereit merged PR #3797.
Last updated: Nov 22 2024 at 16:03 UTC