[turbofan] Introduce new operators Float32SubPreserveNan and Float64SubPreserveNan for wasm.

src/compiler/s390/instruction-selector-s390.cc

Index: src/compiler/s390/instruction-selector-s390.cc
diff --git a/src/compiler/s390/instruction-selector-s390.cc b/src/compiler/s390/instruction-selector-s390.cc
index 6fdd7ec87fd56695c7c7eefd3c5cfd4f13a4514a..00782d15a0b4230a7096dbb1baa44acf12854fab 100644
--- a/src/compiler/s390/instruction-selector-s390.cc
+++ b/src/compiler/s390/instruction-selector-s390.cc
@@ -1117,6 +1117,11 @@ void InstructionSelector::VisitFloat32Sub(Node* node) {
   VisitRRR(this, kS390_SubFloat, node);
 }
 
+void InstructionSelector::VisitFloat32SubPreserveNan(Node* node) {
+  S390OperandGenerator g(this);
+  VisitRRR(this, kS390_SubFloat, node);
+}
+
 void InstructionSelector::VisitFloat64Sub(Node* node) {
   // TODO(mbrandy): detect multiply-subtract
   S390OperandGenerator g(this);
@@ -1142,6 +1147,10 @@ void InstructionSelector::VisitFloat64Sub(Node* node) {
   VisitRRR(this, kS390_SubDouble, node);
 }
 
+void InstructionSelector::VisitFloat64SubPreserveNan(Node* node) {
+  VisitRRR(this, kS390_SubDouble, node);
+}
+
 void InstructionSelector::VisitFloat32Mul(Node* node) {
   VisitRRR(this, kS390_MulFloat, node);
 }