"Couldn't materialize" error in debugger after indirect call

[catamorphism]

Test Case

function-pointer.wasm.gz

Steps to Reproduce

The .wasm file was created with:

clang --target=wasm32-wasip2 --sysroot=$HOME/wasi-libc/sysroot -o function-pointer.wasm function-pointer.c -g -O0

and wasi-libc was built from source.

function-pointer.c is as follows:

struct myfile {
  int data;
  void (*f)();
} myfile;

void f() {}

int foo(struct myfile* f1) {
    f1->f();
    // If the above line is commented out and the next line is uncommented,
    // it works as expected
    // f()
    if (f1->data == 42)
        return 0;
    return 1;
}

int main() {
    struct myfile f1;
    f1.f = &f;
    f1.data = 42;
    return foo(&f1);
}

Steps:

lldb -- wasmtime run -D debug-info -O opt-level=0 --wasm component-model --dir .::/ function-pointer.wasm
(lldb) b foo
(lldb) run
(lldb) p f1
(WebAssemblyPtrWrapper<myfile>)  (__ptr = 199872)
(lldb) n
(lldb) p f1
error: Couldn't materialize: couldn't get the value of variable f1: variable not available
error: errored out in DoExecute, couldn't PrepareToExecuteJITExpression

Expected Results

I would expect to be able to view f1 in the debugger at any point during the execution of foo().

Actual Results

"Couldn't materialize" error as above.

Versions and Environment

Wasmtime version or commit: 35.0.0

lldb version: 18.1.3

Operating system: Linux (Ubuntu 24.04)

Architecture: x86_64

[cfallin]

Unfortunately I don't have an easy answer/fix, but I'll note that historically our native-debugger support has been quite best-effort, and is more or less unmaintained at the moment. The DWARF translation has had known bugs, and we know that various aspects of our lowering can cause values to become unavailable. All of that to say: I'm not surprised that you are finding holes, and I wouldn't recommend relying on debugging by attaching a native debugger to wasmtime at the moment.

We do have a debugging RFC that captures a lot of thinking about our debugging plans from a year ago; and assuming plans don't change, I should be able to start work on pieces of this later in the year.