Initial checkin of the new Mozart compositor.

services/gfx/compositor/backend/gpu_output.h

Index: services/gfx/compositor/backend/gpu_output.h
diff --git a/services/gfx/compositor/backend/gpu_output.h b/services/gfx/compositor/backend/gpu_output.h
new file mode 100644
index 0000000000000000000000000000000000000000..89ef8aca3e6fc3b138c1c44f6d6adb3e66f9a4db
--- /dev/null
+++ b/services/gfx/compositor/backend/gpu_output.h
@@ -0,0 +1,92 @@
+// Copyright 2015 The Chromium 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 SERVICES_GFX_COMPOSITOR_BACKEND_GPU_OUTPUT_H_
+#define SERVICES_GFX_COMPOSITOR_BACKEND_GPU_OUTPUT_H_
+
+#include <memory>
+#include <mutex>
+
+#include "base/callback.h"
+#include "base/macros.h"
+#include "base/memory/ref_counted.h"
+#include "base/memory/scoped_ptr.h"
+#include "base/task_runner.h"
+#include "base/threading/thread.h"
+#include "mojo/services/gpu/interfaces/context_provider.mojom.h"
+#include "services/gfx/compositor/backend/output.h"
+#include "services/gfx/compositor/backend/vsync_scheduler.h"
+
+namespace compositor {
+
+class GpuRasterizer;
+
+// Renderer backed by a ContextProvider.
+class GpuOutput : public Output {
+ public:
+  GpuOutput(mojo::ContextProviderPtr context_provider,
+            const SchedulerCallbacks& scheduler_callbacks,
+            const base::Closure& error_callback);
+  ~GpuOutput() override;
+
+  Scheduler* scheduler() override;
+  void SubmitFrame(const std::shared_ptr<RenderFrame>& frame) override;
+
+ private:
+  // Frame queue, held by a std::shared_ptr.
+  // This object acts as a shared fifo between both threads.
+  class FrameQueue {
+   public:
+    FrameQueue();
+    ~FrameQueue();
+
+    // Puts a pending frame into the queue, drops existing frames if needed.
+    // Returns true if the queue was previously empty.
+    bool PutFrame(const std::shared_ptr<RenderFrame>& frame);
+
+    // Takes a pending frame from the queue.
+    std::shared_ptr<RenderFrame> TakeFrame();

[abarth, 2016/01/10 04:22:09]

Do we want to use shared_ptr in production code?

[jeffbrown, 2016/01/16 03:28:31]

I don't have strong opinions on this.  shared_ptr does not have any additional
heap overhead as long as every instance is allocated using make_shared.  Works
fine for me.

+
+   private:
+    std::mutex mutex_;
+    std::shared_ptr<RenderFrame> next_frame_;  // guarded by |mutex_|
+
+    DISALLOW_COPY_AND_ASSIGN(FrameQueue);
+  };
+
+  // Wrapper around state which is only accessible by the rasterizer thread.
+  class RasterizerDelegate {
+   public:
+    explicit RasterizerDelegate(const std::shared_ptr<FrameQueue>& frame_queue);
+    ~RasterizerDelegate();
+
+    void CreateRasterizer(
+        mojo::InterfacePtrInfo<mojo::ContextProvider> context_provider_info,
+        const std::shared_ptr<VsyncScheduler>& scheduler,
+        const scoped_refptr<base::TaskRunner>& task_runner,
+        const base::Closure& error_callback);
+
+    void SubmitNextFrame();
+
+   private:
+    std::shared_ptr<FrameQueue> frame_queue_;
+    std::unique_ptr<GpuRasterizer> rasterizer_;
+
+    DISALLOW_COPY_AND_ASSIGN(RasterizerDelegate);
+  };
+
+  std::shared_ptr<FrameQueue> frame_queue_;
+  std::shared_ptr<VsyncScheduler> scheduler_;
+  scoped_ptr<RasterizerDelegate> rasterizer_delegate_;  // can't use unique_ptr
+                                                        // here due to
+                                                        // base::Bind (sadness)
+  std::unique_ptr<base::Thread> rasterizer_thread_;
+  scoped_refptr<base::SingleThreadTaskRunner> rasterizer_task_runner_;
+
+  DISALLOW_COPY_AND_ASSIGN(GpuOutput);
+};
+
+}  // namespace compositor
+
+#endif  // SERVICES_GFX_COMPOSITOR_BACKEND_GPU_OUTPUT_H_