Current model supports RTL where core is stalled as soon as first access is denied.
RTL has been improved so that the core stall only after second access while first one is still denied.