[llvm] [AMDGPU] Move kernarg preload logic to separate pass (PR #130434)

[Christudasan Devadasan via llvm-commits]

================
@@ -1317,6 +1318,10 @@ void AMDGPUPassConfig::addIRPasses() {
 }
 
 void AMDGPUPassConfig::addCodeGenPrepare() {
+  if (TM->getTargetTriple().getArch() == Triple::amdgcn &&
----------------
cdevadas wrote:

I don't know why we need the TargetTriple check here. If the execution reaches here, the compilation is already for amdgcn triple and not for r600. Anyways, it is unrelated to this PR.

https://github.com/llvm/llvm-project/pull/130434