[turbofan] Enable concurrent (re)compilation.

src/compiler/pipeline.h

 // Copyright 2014 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.
 
 #ifndef V8_COMPILER_PIPELINE_H_
 #define V8_COMPILER_PIPELINE_H_
 
 // Clients of this interface shouldn't depend on lots of compiler internals.
 // Do not include anything from src/compiler here!
 #include "src/compiler.h"
 
 namespace v8 {
 namespace internal {
 namespace compiler {
 
 class CallDescriptor;
 class Graph;
 class InstructionSequence;
 class Linkage;
 class PipelineData;
 class RegisterConfiguration;
 class Schedule;
 
-class Pipeline {
+class Pipeline final {
  public:
-  explicit Pipeline(CompilationInfo* info) : info_(info) {}
+  explicit Pipeline(CompilationInfo* info);
+  ~Pipeline();
+
+  // Separate pipeline steps.
+  bool CreateGraph();
+  bool OptimizeGraph();
+  Handle<Code> CreateCode() { return DoGenerateCode(); }
 
   // Run the entire pipeline and generate a handle to a code object.
   Handle<Code> GenerateCode();
 
   // Run the pipeline on a machine graph and generate code. If {schedule} is
   // {nullptr}, then compute a new schedule for code generation.
   static Handle<Code> GenerateCodeForTesting(CompilationInfo* info,
                                              Graph* graph,
                                              Schedule* schedule = nullptr);
 
   // Run the pipeline on a machine graph and generate code. If {schedule} is
   // {nullptr}, then compute a new schedule for code generation.
   static Handle<Code> GenerateCodeForTesting(Isolate* isolate,
                                              CallDescriptor* call_descriptor,
                                              Graph* graph,
                                              Schedule* schedule = nullptr);
 
   // Run just the register allocator phases.
   static bool AllocateRegistersForTesting(const RegisterConfiguration* config,
                                           InstructionSequence* sequence,
                                           bool run_verifier);
 
   static inline bool SupportedBackend() { return V8_TURBOFAN_BACKEND != 0; }
   static inline bool SupportedTarget() { return V8_TURBOFAN_TARGET != 0; }
 
  private:
+  explicit Pipeline(PipelineData* data) : data_(data) {
+    DCHECK_NOT_NULL(data_);
+  }
+
   static Handle<Code> GenerateCodeForTesting(CompilationInfo* info,
                                              CallDescriptor* call_descriptor,
                                              Graph* graph, Schedule* schedule);
 
-  CompilationInfo* info_;
-  PipelineData* data_;
+  PipelineData* const data_;
 
   // Helpers for executing pipeline phases.
   template <typename Phase>
   void Run();
   template <typename Phase, typename Arg0>
   void Run(Arg0 arg_0);
 
-  CompilationInfo* info() const { return info_; }
-  Isolate* isolate() { return info_->isolate(); }
+  PipelineData* data() const { return data_; }
+  CompilationInfo* info() const;
+  Isolate* isolate() const { return info()->isolate(); }
 
   void BeginPhaseKind(const char* phase_kind);
   void RunPrintAndVerify(const char* phase, bool untyped = false);
+  bool ScheduleAndAllocateRegisters(CallDescriptor* call_descriptor);
+  Handle<Code> DoGenerateCode();
   Handle<Code> ScheduleAndGenerateCode(CallDescriptor* call_descriptor);
   void AllocateRegisters(const RegisterConfiguration* config,
                          bool run_verifier);
 };
 
 }  // namespace compiler
 }  // namespace internal
 }  // namespace v8
 
 #endif  // V8_COMPILER_PIPELINE_H_