[turbofan] Properly handle bit->float64 representation changes.

src/compiler/representation-change.cc

 // Copyright 2015 the V8 project authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 #include "src/compiler/representation-change.h"
 
 #include <sstream>
 
 #include "src/base/bits.h"
 #include "src/code-factory.h"
@@ skipping 322 matching lines @@
   const Operator* op = nullptr;
   if (IsWord(output_rep)) {
     if (output_type->Is(Type::Signed32())) {
       op = machine()->ChangeInt32ToFloat64();
     } else if (output_type->Is(Type::Unsigned32()) ||
                use_info.truncation().TruncatesToWord32()) {
       // Either the output is uint32 or the uses only care about the
       // low 32 bits (so we can pick uint32 safely).
       op = machine()->ChangeUint32ToFloat64();
     }
+  } else if (output_rep == MachineRepresentation::kBit) {
+    op = machine()->ChangeUint32ToFloat64();
   } else if (output_rep == MachineRepresentation::kTagged) {
     if (output_type->Is(Type::Undefined())) {
       return jsgraph()->Float64Constant(
           std::numeric_limits<double>::quiet_NaN());
     } else if (output_type->Is(Type::TaggedSigned())) {
       node = InsertChangeTaggedSignedToInt32(node);
       op = machine()->ChangeInt32ToFloat64();
     } else if (output_type->Is(Type::Number())) {
       op = simplified()->ChangeTaggedToFloat64();
     } else if (output_type->Is(Type::NumberOrOddball())) {
@@ skipping 454 matching lines @@
 }
 
 Node* RepresentationChanger::InsertChangeTaggedToFloat64(Node* node) {
   return jsgraph()->graph()->NewNode(simplified()->ChangeTaggedToFloat64(),
                                      node);
 }
 
 }  // namespace compiler
 }  // namespace internal
 }  // namespace v8