For target environment amdnas and amdnascl, AMDGPU will use new address space mapping where generic address space is 0 and private address space is 5. The data layout is also changed correspondingly.
This is the beginning of an upstreaming effort for changing address space mapping of AMDGPU target. Sema/CodeGen changes to make OpenCL/C++ working with the new address space mapping will follow.
The depending LLVM change: https://reviews.llvm.org/D31211