Wasmtime GitHub notifications bot (Jun 05 2023 at 13:51): Wasmtime GitHub notifications bot (Jun 05 2023 at 13:51):afonso360 opened PR #6520 from afonso360:fuzzgen-fix-timeout to bytecodealliance:main:
:wave: Hey,
This PR fixes #6415. When running in interpreter vs interpreter mode, we just run the mid end passes. Some of these can actually increase the number of instructions that the interpreter has to run (i.e.
stack_load -> load+stack_addr), this can cause a timeout in the second run if we were barely at the fuel limit on the first run.The solution here is to ignore timeouts in the second run. They can only happen if the second run is an interpreter run since we don't have timeouts in host runs.
Wasmtime GitHub notifications bot (Jun 05 2023 at 13:51):afonso360 requested jameysharp for a review on PR #6520.
Wasmtime GitHub notifications bot (Jun 05 2023 at 13:51):afonso360 requested wasmtime-fuzz-reviewers for a review on PR #6520.
Wasmtime GitHub notifications bot (Jun 06 2023 at 17:17):jameysharp submitted PR review:
How do the timeouts work? Does the interpreter count instructions executed, and the native execution doesn't, and that's why you don't expect timeouts from native execution? Assuming that's true, this seems great.
Wasmtime GitHub notifications bot (Jun 06 2023 at 18:22):afonso360 merged PR #6520.
Last updated: Oct 23 2024 at 20:03 UTC