[llvm] [AMDGPU] Handle invariant marks in `AMDGPUPromoteAllocaPass` (PR #124607)

[Shilei Tian via llvm-commits]

================
@@ -1556,12 +1556,34 @@ bool AMDGPUPromoteAllocaImpl::tryPromoteAllocaToLDS(AllocaInst &I,
     case Intrinsic::invariant_start:
     case Intrinsic::invariant_end:
     case Intrinsic::launder_invariant_group:
-    case Intrinsic::strip_invariant_group:
+    case Intrinsic::strip_invariant_group: {
+      SmallVector<Type *> ArgTy;
+      SmallVector<Value *> Args;
+      if (Intr->getIntrinsicID() == Intrinsic::invariant_start) {
+        Value *Size = Intr->getArgOperand(0);
+        ArgTy.emplace_back(Offset->getType());
+        Args.emplace_back(Size);
+        Args.emplace_back(Offset);
----------------
shiltian wrote:

Sure, will do a follow up.

https://github.com/llvm/llvm-project/pull/124607