[llvm] [AMDGPU] Add IR LiveReg type-based optimization (PR #66838)

[Jay Foad via llvm-commits]

================
@@ -371,9 +452,269 @@ bool AMDGPUCodeGenPrepareImpl::run(Function &F) {
       }
     }
   }
+
+  // GlobalISel should directly use the values, and do not need to emit
+  // CopyTo/CopyFrom Regs across blocks
+  if (UsesGlobalISel)
----------------
jayfoad wrote:

Just query `TM->Options.EnableGlobalISel` here, no need for a new `UsesGlobalISel` field?

https://github.com/llvm/llvm-project/pull/66838