[llvm] expandFMINIMUMNUM_FMAXIMUMNUM: Improve compare between zeros (PR #140193)

[Matt Arsenault via llvm-commits]

================
@@ -8866,13 +8867,19 @@ SDValue TargetLowering::expandFMINIMUMNUM_FMAXIMUMNUM(SDNode *Node,
       DAG.getTargetConstant(IsMax ? fcPosZero : fcNegZero, DL, MVT::i32);
   SDValue IsZero = DAG.getSetCC(DL, CCVT, MinMax,
                                 DAG.getConstantFP(0.0, DL, VT), ISD::SETEQ);
-  SDValue LCmp = DAG.getSelect(
-      DL, VT, DAG.getNode(ISD::IS_FPCLASS, DL, CCVT, LHS, TestZero), LHS,
+  EVT IntVT = VT.changeTypeToInteger();
+  EVT FloatVT = VT.changeElementType(MVT::f32);
+  SDValue LHSTrunc = LHS;
+  if (!isTypeLegal(IntVT) && !isOperationLegal(ISD::IS_FPCLASS, VT)) {
+    LHSTrunc = DAG.getNode(ISD::FP_ROUND, DL, FloatVT, LHS,
+                           DAG.getIntPtrConstant(0, DL, /*isTarget=*/true));
+  }
+  // It's OK to select from LHS and MinMax, with only one ISD::IS_FPCLASS, as
+  // we preferred RHS when generate MinMax, if the operands are equal.
+  SDValue RetZero = DAG.getSelect(
----------------
arsenm wrote:

Maybe it would be better to introduce a new utility function in TargetLowering to emit a sign bit check for an FP value and use it here 

https://github.com/llvm/llvm-project/pull/140193