[llvm-branch-commits] [llvm] [SelectionDAG][X86] Widen <2 x T> vector types for atomic load (PR #120598)

[Matt Arsenault via llvm-branch-commits]

================
@@ -52242,6 +52243,81 @@ static SDValue combineConstantPoolLoads(SDNode *N, const SDLoc &dl,
   return SDValue();
 }
 
+static MVT getScalarTypeFromVectorType(MVT VT) {
+  if (VT == MVT::v2i8)
+    return MVT::i16;
+  if (VT == MVT::v2i16 || VT == MVT::v2f16 || VT == MVT::v2bf16)
+    return MVT::i32;
+  if (VT == MVT::v2i32 || VT == MVT::v2f32)
+    return MVT::i64;
+  LLVM_DEBUG(dbgs() << VT << '\n');
+  llvm_unreachable("Invalid VT for scalar type translation");
+}
+
+static SDValue combineAtomicLoad(SDNode *N, SelectionDAG &DAG,
----------------
arsenm wrote:

I don't think any of this custom handling in the target should be necessary 

https://github.com/llvm/llvm-project/pull/120598