PPC: [turbofan] Fix implementation of Float64Min.

The optimized instruction sequences for floating-point min/max do not
have the same behaviour as TurboFan's Float(32|64)(Min|Max) functions
(incorrect handling for NaN operands).

R=dstence@us.ibm.com, michael_dawson@ca.ibm.com
BUG=

Review URL: https://codereview.chromium.org/1193843015

Cr-Commit-Position: refs/heads/master@{#29234}

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

@@ -976,24 +976,16 @@ void InstructionSelector::VisitFloat64Mod(Node* node) {
 }
 
 
-void InstructionSelector::VisitFloat32Max(Node* node) {
-  VisitRRR(this, kPPC_MaxDouble, node);
-}
+void InstructionSelector::VisitFloat32Max(Node* node) { UNREACHABLE(); }
 
 
-void InstructionSelector::VisitFloat64Max(Node* node) {
-  VisitRRR(this, kPPC_MaxDouble, node);
-}
+void InstructionSelector::VisitFloat64Max(Node* node) { UNREACHABLE(); }
 
 
-void InstructionSelector::VisitFloat32Min(Node* node) {
-  VisitRRR(this, kPPC_MinDouble, node);
-}
+void InstructionSelector::VisitFloat32Min(Node* node) { UNREACHABLE(); }
 
 
-void InstructionSelector::VisitFloat64Min(Node* node) {
-  VisitRRR(this, kPPC_MinDouble, node);
-}
+void InstructionSelector::VisitFloat64Min(Node* node) { UNREACHABLE(); }
 
 
 void InstructionSelector::VisitFloat32Abs(Node* node) {
@@ -1628,11 +1620,7 @@ void InstructionSelector::VisitFloat64InsertHighWord32(Node* node) {
 // static
 MachineOperatorBuilder::Flags
 InstructionSelector::SupportedMachineOperatorFlags() {
-  return MachineOperatorBuilder::kFloat32Max |
-         MachineOperatorBuilder::kFloat32Min |
-         MachineOperatorBuilder::kFloat64Max |
-         MachineOperatorBuilder::kFloat64Min |
-         MachineOperatorBuilder::kFloat64RoundDown |
+  return MachineOperatorBuilder::kFloat64RoundDown |
          MachineOperatorBuilder::kFloat64RoundTruncate |
          MachineOperatorBuilder::kFloat64RoundTiesAway;
   // We omit kWord32ShiftIsSafe as s[rl]w use 0x3f as a mask rather than 0x1f.