Re: [PATCH vkd3d 1/5] vkd3d: Return S_OK from ID3D12CommandQueue::GetClockCalibration().

[Henri Verbeet]

On Mon, 10 Jan 2022 at 15:02, Conor McCarthy  wrote:
> @@ -6225,10 +6225,14 @@ static HRESULT STDMETHODCALLTYPE d3d12_command_queue_GetTimestampFrequency(ID3D1
>  static HRESULT STDMETHODCALLTYPE d3d12_command_queue_GetClockCalibration(ID3D12CommandQueue *iface,
>          UINT64 *gpu_timestamp, UINT64 *cpu_timestamp)
>  {
> - FIXME("iface %p, gpu_timestamp %p, cpu_timestamp %p stub!\n",
> - TRACE("iface %p, gpu_timestamp %p, cpu_timestamp %p.\n",
>         iface, gpu_timestamp, cpu_timestamp);
> - return E_NOTIMPL;
> - WARN("Setting timestamps to zero.\n");
> - *gpu_timestamp = 0;
> - *cpu_timestamp = 0;
> -
> - return S_OK;
>
> }

Always returning 0 here (without FIXME) seems like something that may
end up being problematic in hard to diagnose ways at some point.
Couldn't we use VK_EXT_calibrated_timestamps here, or failing that,
build something on top of vkCmdWriteTimestamp()?