WebAssembly is currently always run in the same process synchronously. Thus, once WebAssembly starts executing, you have to wait for the execution to complete to continue running code on the host (your application).
However, there are cases where you may want to interrupt this synchronous execution while the guest Wasm module is calling a host function. This can be useful for saving resources, and not returning back to the guest Wasm for execution, when you already know the Wasm execution will fail, or no longer be needed.
In this example, we will run a Wasm module that calls the imported host function interrupt_execution. This host function will immediately stop executing the WebAssembly module.
First we are going to want to initialize a new project. To do this we can navigate to our project folder, or create one. In this example, we will create a new project. Lets create it and navigate to it:
The final code for this example can be found on GitHub.
There is nothing special or Wasmer specific here but it will be required later in the example.
Defining and importing the host function
To terminate the execution of the Wasm module we'll have to define a function on the host which will then be imported in the guest and called whenever execution is not required to continue. Let's do that:
As we saw in previous examples we defined a Rust function, wrap it in a native function definition and import it in the guest module, in the env namespace, using the ImportObject.
Handling the error
Our module will call the early_exit function once we call its run function (which is an exported function). Let's get the function, call it and see how we can handle the error:
let run_func:NativeFunc<(i32, i32), i32> = instance.exports.get_native_function("run")?;match run_func.call(1, 7) {Ok(result) => {bail!("Expected early termination with `ExitCode`, found: {}", result); } Err(e) =>match e.downcast::<ExitCode>() {// We found the exit code used to terminate execution.Ok(exit_code) => {println!("Exited early with exit code: {}", exit_code);Ok(()) }Err(e) => {bail!("Unknown error `{}` found. expected `ErrorCode`", e); } },}
We expect to get an error when calling the run function so what we do here is look at the result and:
if we get a success, our test will fail;
if we get an error, we try to downcast to our ExitCode error.
If downcasting succeeds it means we actually got the expected error so we make the test pass. If it fails, it means the Wasm module reported an error but it wasn't the one we expected so we make the test fail.
run, err := instance.Exports.GetFunction("run")if err !=nil {panic(fmt.Sprintln("Failed to retrieve the `run` function:", err))}_, err =run(1, 7)if err ==nil {panic(fmt.Sprintln("`run` did not error"))}fmt.Println("Exited early with:", err)
try: instance.exports.run(1, 2)except RuntimeError as err: assert "oops" in str(err)else: assert False
Running
We now have everything we need to run the Wasm module, let's do it!
You should be able to run it using the cargo run command. The output should look like this:
Compiling module...
Instantiating module...
Calling `run` function...
Exited early with exit code: 1
If you want to run the examples from the Wasmer repository codebase directly, you can also do: