Context: My react app is accessing a large dataset from a worker using a SharedArrayBuffer. I'd like to process the SharedArrayBuffer in C# with minimal overhead. I have a hard time converting the SharedArrayBuffer to a Span in C#.
Here's what I tried:

public static int TestSum(SharedArrayBuffer sharedArrayBuffer)
{
    Int32Array array = new Int32Array(sharedArrayBuffer);
    Span<int> nativeArray = array; // error                        
    
    int sum = 0;
    for (int i = 0; i < nativeArray.Length; i++)
    {
        sum += nativeArray[i];
    }

    return sum;
}

This unfortunately fails with "Object '...' is not a typed array" in js_typed_array_to_array.

Here's the function from dotnet.js:

js_typed_array_to_array : function (js_obj) {
	if (!!(js_obj.buffer instanceof ArrayBuffer && js_obj.BYTES_PER_ELEMENT)) 
	{
		var arrayType = js_obj[Symbol.for("wasm type")];
		var heapBytes = this.js_typedarray_to_heap(js_obj);
		var bufferArray = this.mono_typed_array_new(heapBytes.byteOffset, js_obj.length, js_obj.BYTES_PER_ELEMENT, arrayType);
		Module._free(heapBytes.byteOffset);
		return bufferArray;
	}
	else {
		throw new Error("Object '" + js_obj + "' is not a typed array");
	} 

I was able to workaround this by changing

if (!!(js_obj.buffer instanceof ArrayBuffer && js_obj.BYTES_PER_ELEMENT)) 

to

if (!!(js_obj.buffer instanceof (ArrayBuffer, SharedArrayBuffer) && js_obj.BYTES_PER_ELEMENT)) 

Now the code executes with no errors, however there's a copying operation being done in mono_typed_array_new. Doesn't that make the whole idea behind using a SharedArrayBuffer in order to avoid copying useless?

Is this the proper way to process a SharedArrayBuffer? Is there a way to process it without copying the data?