#ifndef CONTENT_COMMON_GPU_GPU_COMMAND_BUFFER_STUB_H_
#define CONTENT_COMMON_GPU_GPU_COMMAND_BUFFER_STUB_H_
#include <deque>
#include <string>
#include <vector>
#include "base/memory/scoped_vector.h"
#include "base/memory/weak_ptr.h"
#include "base/observer_list.h"
#include "content/common/content_export.h"
#include "content/common/gpu/gpu_memory_manager.h"
#include "content/common/gpu/gpu_memory_manager_client.h"
#include "gpu/command_buffer/common/constants.h"
#include "gpu/command_buffer/common/gpu_memory_allocation.h"
#include "gpu/command_buffer/service/command_buffer_service.h"
#include "gpu/command_buffer/service/context_group.h"
#include "gpu/command_buffer/service/gpu_scheduler.h"
#include "ipc/ipc_listener.h"
#include "ipc/ipc_sender.h"
#include "media/base/video_decoder_config.h"
#include "ui/events/latency_info.h"
#include "ui/gfx/gpu_memory_buffer.h"
#include "ui/gfx/native_widget_types.h"
#include "ui/gfx/size.h"
#include "ui/gl/gl_surface.h"
#include "ui/gl/gpu_preference.h"
#include "url/gurl.h"
namespace gpu {
class GpuControlService;
struct Mailbox;
namespace gles2 {
class ImageManager;
class MailboxManager;
}
}
namespace content {
class GpuChannel;
class GpuVideoDecodeAccelerator;
class GpuVideoEncodeAccelerator;
class GpuWatchdog;
class GpuCommandBufferStub
: public GpuMemoryManagerClient,
public IPC::Listener,
public IPC::Sender,
public base::SupportsWeakPtr<GpuCommandBufferStub> {
public:
class DestructionObserver {
public:
virtual void OnWillDestroyStub() = 0;
protected:
virtual ~DestructionObserver() {}
};
typedef base::Callback<void(const std::vector<ui::LatencyInfo>&)>
LatencyInfoCallback;
GpuCommandBufferStub(
GpuChannel* channel,
GpuCommandBufferStub* share_group,
const gfx::GLSurfaceHandle& handle,
gpu::gles2::MailboxManager* mailbox_manager,
gpu::gles2::ImageManager* image_manager,
const gfx::Size& size,
const gpu::gles2::DisallowedFeatures& disallowed_features,
const std::vector<int32>& attribs,
gfx::GpuPreference gpu_preference,
bool use_virtualized_gl_context,
int32 route_id,
int32 surface_id,
GpuWatchdog* watchdog,
bool software,
const GURL& active_url);
virtual ~GpuCommandBufferStub();
virtual bool OnMessageReceived(const IPC::Message& message) OVERRIDE;
virtual bool Send(IPC::Message* msg) OVERRIDE;
virtual gfx::Size GetSurfaceSize() const OVERRIDE;
virtual gpu::gles2::MemoryTracker* GetMemoryTracker() const OVERRIDE;
virtual void SetMemoryAllocation(
const gpu::MemoryAllocation& allocation) OVERRIDE;
virtual void SuggestHaveFrontBuffer(bool suggest_have_frontbuffer) OVERRIDE;
virtual bool GetTotalGpuMemory(uint64* bytes) OVERRIDE;
bool IsScheduled();
bool IsPreempted() const {
return scheduler_.get() && scheduler_->IsPreempted();
}
bool HasUnprocessedCommands();
gpu::gles2::GLES2Decoder* decoder() const { return decoder_.get(); }
gpu::GpuScheduler* scheduler() const { return scheduler_.get(); }
GpuChannel* channel() const { return channel_; }
int32 surface_id() const { return surface_id_; }
int32 route_id() const { return route_id_; }
gfx::GpuPreference gpu_preference() { return gpu_preference_; }
int32 GetRequestedAttribute(int attr) const;
void SendConsoleMessage(int32 id, const std::string& message);
void SendCachedShader(const std::string& key, const std::string& shader);
gfx::GLSurface* surface() const { return surface_.get(); }
void AddDestructionObserver(DestructionObserver* observer);
void RemoveDestructionObserver(DestructionObserver* observer);
void AddSyncPoint(uint32 sync_point);
void SetPreemptByFlag(scoped_refptr<gpu::PreemptionFlag> flag);
void SetLatencyInfoCallback(const LatencyInfoCallback& callback);
void MarkContextLost();
uint64 GetMemoryUsage() const;
private:
GpuMemoryManager* GetMemoryManager() const;
bool MakeCurrent();
void Destroy();
void OnInitializeFailed(IPC::Message* reply_message);
void OnInitialize(base::SharedMemoryHandle shared_state_shm,
IPC::Message* reply_message);
void OnSetGetBuffer(int32 shm_id, IPC::Message* reply_message);
void OnProduceFrontBuffer(const gpu::Mailbox& mailbox);
void OnGetState(IPC::Message* reply_message);
void OnGetStateFast(IPC::Message* reply_message);
void OnAsyncFlush(int32 put_offset, uint32 flush_count);
void OnEcho(const IPC::Message& message);
void OnRescheduled();
void OnRegisterTransferBuffer(int32 id,
base::SharedMemoryHandle transfer_buffer,
uint32 size);
void OnDestroyTransferBuffer(int32 id);
void OnGetTransferBuffer(int32 id, IPC::Message* reply_message);
void OnCreateVideoDecoder(media::VideoCodecProfile profile,
IPC::Message* reply_message);
void OnCreateVideoEncoder(media::VideoFrame::Format input_format,
const gfx::Size& input_visible_size,
media::VideoCodecProfile output_profile,
uint32 initial_bitrate,
IPC::Message* reply_message);
void OnSetSurfaceVisible(bool visible);
void OnEnsureBackbuffer();
void OnRetireSyncPoint(uint32 sync_point);
bool OnWaitSyncPoint(uint32 sync_point);
void OnSyncPointRetired();
void OnSignalSyncPoint(uint32 sync_point, uint32 id);
void OnSignalSyncPointAck(uint32 id);
void OnSignalQuery(uint32 query, uint32 id);
void OnReceivedClientManagedMemoryStats(const gpu::ManagedMemoryStats& stats);
void OnSetClientHasMemoryAllocationChangedCallback(bool has_callback);
void OnRegisterGpuMemoryBuffer(int32 id,
gfx::GpuMemoryBufferHandle gpu_memory_buffer,
uint32 width,
uint32 height,
uint32 internalformat);
void OnDestroyGpuMemoryBuffer(int32 id);
void OnCommandProcessed();
void OnParseError();
void OnSetLatencyInfo(const std::vector<ui::LatencyInfo>& latency_info);
void OnCreateStreamTexture(uint32 texture_id, int32* stream_id);
void ReportState();
void PutChanged();
void PollWork();
bool HasMoreWork();
void ScheduleDelayedWork(int64 delay);
bool CheckContextLost();
GpuChannel* channel_;
scoped_refptr<gpu::gles2::ContextGroup> context_group_;
gfx::GLSurfaceHandle handle_;
gfx::Size initial_size_;
gpu::gles2::DisallowedFeatures disallowed_features_;
std::vector<int32> requested_attribs_;
gfx::GpuPreference gpu_preference_;
bool use_virtualized_gl_context_;
int32 route_id_;
int32 surface_id_;
bool software_;
uint32 last_flush_count_;
scoped_ptr<gpu::CommandBufferService> command_buffer_;
scoped_ptr<gpu::gles2::GLES2Decoder> decoder_;
scoped_ptr<gpu::GpuScheduler> scheduler_;
scoped_refptr<gfx::GLSurface> surface_;
scoped_ptr<gpu::GpuControlService> gpu_control_;
scoped_ptr<GpuMemoryManagerClientState> memory_manager_client_state_;
bool last_memory_allocation_valid_;
gpu::MemoryAllocation last_memory_allocation_;
GpuWatchdog* watchdog_;
ObserverList<DestructionObserver> destruction_observers_;
std::deque<uint32> sync_points_;
int sync_point_wait_count_;
bool delayed_work_scheduled_;
uint64 previous_messages_processed_;
base::TimeTicks last_idle_time_;
scoped_refptr<gpu::PreemptionFlag> preemption_flag_;
LatencyInfoCallback latency_info_callback_;
GURL active_url_;
size_t active_url_hash_;
size_t total_gpu_memory_;
DISALLOW_COPY_AND_ASSIGN(GpuCommandBufferStub);
};
}
#endif