[llvm] [AMDGPU][SDAG] Support source modifiers on select integer operands (PR #147325)

[Chris Jackson via llvm-commits]

================
@@ -12175,6 +12176,71 @@ SDValue DAGCombiner::foldSelectToABD(SDValue LHS, SDValue RHS, SDValue True,
   return SDValue();
 }
 
+static SDValue getBitwiseToSrcModifierOp(SDValue N, SelectionDAG &DAG) {
+
+  unsigned Opc = N.getNode()->getOpcode();
+  if (Opc != ISD::AND && Opc != ISD::XOR && Opc != ISD::OR)
+    return SDValue();
----------------
chrisjbris wrote:

I believe this should be checked in both locations as one is AMDGPU backend specific and the other is TI.

https://github.com/llvm/llvm-project/pull/147325