[llvm-branch-commits] [llvm] [AMDGPU][SDAG] Add ISD::PTRADD DAG combines (PR #142739)

[Fabian Ritter via llvm-branch-commits]

================
@@ -2627,6 +2629,93 @@ SDValue DAGCombiner::foldSubToAvg(SDNode *N, const SDLoc &DL) {
   return SDValue();
 }
 
+/// Try to fold a pointer arithmetic node.
+/// This needs to be done separately from normal addition, because pointer
+/// addition is not commutative.
+SDValue DAGCombiner::visitPTRADD(SDNode *N) {
+  SDValue N0 = N->getOperand(0);
+  SDValue N1 = N->getOperand(1);
+  EVT PtrVT = N0.getValueType();
+  EVT IntVT = N1.getValueType();
+  SDLoc DL(N);
+
+  // This is already ensured by an assert in SelectionDAG::getNode(). Several
+  // combines here depend on this assumption.
+  assert(PtrVT == IntVT &&
+         "PTRADD with different operand types is not supported");
+
+  // fold (ptradd undef, y) -> undef
+  if (N0.isUndef())
----------------
ritter-x2a wrote:

I've added some in the `ptradd-sdag-undef-poison.ll` test. Can't have them diff nicely like the other tests because the folds are already enabled on trunk, but I did verify that additions are generated if the folds are disabled again.
We obviously need to ignore the undef deprecator github action for this test (or should I limit it to poison?).

https://github.com/llvm/llvm-project/pull/142739