[Atomics] use TFJ builtins for atomic add, sub, and, or, and xor

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

Index: src/compiler/ppc/instruction-selector-ppc.cc
diff --git a/src/compiler/ppc/instruction-selector-ppc.cc b/src/compiler/ppc/instruction-selector-ppc.cc
index eb1bd0d9e65f1ae48c4ccb73ff66d5f391ed1365..449e71038999032df537ebc2d332fd8607f3283e 100644
--- a/src/compiler/ppc/instruction-selector-ppc.cc
+++ b/src/compiler/ppc/instruction-selector-ppc.cc
@@ -2123,7 +2123,7 @@ void InstructionSelector::VisitAtomicExchange(Node* node) {
   Node* index = node->InputAt(1);
   Node* value = node->InputAt(2);
   ArchOpcode opcode = kArchNop;
-  MachineType type = AtomicExchangeRepresentationOf(node->op());
+  MachineType type = AtomicOpRepresentationOf(node->op());
   if (type == MachineType::Int8()) {
     opcode = kAtomicExchangeInt8;
   } else if (type == MachineType::Uint8()) {
@@ -2155,6 +2155,16 @@ void InstructionSelector::VisitAtomicCompareExchange(Node* node) {
   UNIMPLEMENTED();
 }
 
+void InstructionSelector::VisitAtomicAdd(Node* node) { UNIMPLEMENTED(); }
+
+void InstructionSelector::VisitAtomicSub(Node* node) { UNIMPLEMENTED(); }
+
+void InstructionSelector::VisitAtomicAnd(Node* node) { UNIMPLEMENTED(); }
+
+void InstructionSelector::VisitAtomicOr(Node* node) { UNIMPLEMENTED(); }
+
+void InstructionSelector::VisitAtomicXor(Node* node) { UNIMPLEMENTED(); }
+
 void InstructionSelector::VisitInt32AbsWithOverflow(Node* node) {
   UNREACHABLE();
 }