wasmtime / issue #14597 Pulley and s390x don't correctly ... · git-wasmtime

#[wasmtime_test]
fn test_trap_stack_overflow_mutual_recursion(config: &mut Config) -> Result<()> {
    let max_frames = NonZeroUsize::new(32).unwrap();
    let wat = r#"
        (module $rec_mod
            (func $even (export "even") (call $odd))
            (func $odd (call $even))
        )
    "#;

    config.wasm_backtrace_max_frames(Some(max_frames));
    let engine = Engine::new(&config)?;
    let module = Module::new(&engine, wat)?;
    let mut store = Store::new(&engine, ());
    let instance = Instance::new(&mut store, &module, &[])?;
    let even = instance.get_typed_func::<(), ()>(&mut store, "even")?;

    let e = even.call(&mut store, ()).unwrap_err();

    // No frame may be skipped, so the frames must alternate between the two
    // functions.
    let trace = e.downcast_ref::<WasmBacktrace>().unwrap().frames();
    assert_eq!(trace.len(), max_frames.get());
    for pair in trace.windows(2) {
        assert_ne!(pair[0].func_index(), pair[1].func_index(), "{trace:?}");
    }
    assert_eq!(e.downcast::<Trap>()?, Trap::StackOverflow);

    Ok(())
}

passes on all platforms except s390x and Pulley. Neither of them handle stack overflow in such a way where the final frame on the stack is the frame that actually caused the stack overflow. This is a relatively minor issue, but it'd be nice to fix to help for attribution of stack overflows.

Stream: git-wasmtime

Topic: wasmtime / issue #14597 Pulley and s390x don't correctly ...

Wasmtime GitHub notifications bot (Oct 06 2026 at 23:04):

Wasmtime GitHub notifications bot (Oct 06 2026 at 23:04):

Wasmtime GitHub notifications bot (Oct 06 2026 at 23:04):

Wasmtime GitHub notifications bot (Oct 09 2026 at 01:02):