[llvm-branch-commits] [llvm] AMDGPU: Cleanup the handling of flags in getTgtMemIntrinsic (PR #179469)

[Matt Arsenault via llvm-branch-commits]

================
@@ -1360,6 +1359,15 @@ void SITargetLowering::getTgtMemIntrinsic(SmallVectorImpl<IntrinsicInfo> &Infos,
     if (ME.doesNotAccessMemory())
       return;
 
+    bool IsSPrefetch = IntrID == Intrinsic::amdgcn_s_buffer_prefetch_data;
+    if (!IsSPrefetch) {
----------------
arsenm wrote:

Why is this up here? Why isn't this just another case in the switch over intrinsic ID?

https://github.com/llvm/llvm-project/pull/179469