[llvm] [NVPTX] Improve device function byval parameter lowering (PR #129188)

[Akshay Deodhar via llvm-commits]

================
@@ -15,9 +15,9 @@ target triple = "nvptx64-nvidia-cuda"
 ; COMMON-LABEL: load_alignment
 define void @load_alignment(ptr nocapture readonly byval(%class.outer) align 8 %arg) {
 entry:
----------------
akshayrdeodhar wrote:

Can this be converted to autogenerated checks? The deleted checks, for example, are not clear- the can't figure out what the memcpy was doing.

https://github.com/llvm/llvm-project/pull/129188