void RenderWidget::OnRepaint(gfx::Size size_to_paint) void RenderWidgetCompositor::SetNeedsRedrawRect(gfx::Rect damage_rect); void LayerTreeHost::SetNeedsRedrawRect(const gfx::Rect& damage_rect) void ProxyMain::SetNeedsRedraw(const gfx::Rect& damage_rect); void ThreadProxy::SetNeedsRedraw(const gfx::Rect& damage_rect)

PostTask(FROM_HERE, base::Bind(&ThreadProxy::SetNeedsRedrawRectOnImplThread,...);

void ThreadProxy::SetNeedsRedrawRectOnImplThread(const gfx::Rect& damage_rect)

impl().layer_tree_host_impl->SetViewportDamage(damage_rect);

void ThreadProxy::SetNeedsRedrawOnImplThread();

impl().scheduler->SetNeedsRedraw();

void Scheduler::SetNeedsAnimate();

statemachine.SetNeedsAnimate();ProcessScheduledActions(); void Scheduler::ProcessScheduledActions(); call SchedulerClient* serial function according to action type void Scheduler::DrawAndSwapIfPossible()//判断是否可以绘制，如果可以绘制的话，调用client的绘制接口

client_->ScheduledActionDrawAndSwapIfPossible();

statemachine.DidDrawIfPossibleCompleted(result); //绘制完成后更新状态机的状态

DrawResult ThreadProxy::ScheduledActionDrawAndSwapIfPossible()

DCHECK(impl().layer_tree_host_impl->CanDraw());
DrawSwapInternal(forced_draw);

DrawResult ThreadProxy::DrawSwapInternal(bool forced_draw);

layer_tree_host_impl->DrawLayers( const FrameData& frame )
LayerTreeHostImpl::DidDrawAllLayers(const FrameData& frame)

LayerTreeHostImpl::DidDrawAllLayers(const FrameData& frame); 其中，DrawLayers 中会判断这个 framedata中是否有需要更新的区域，如果没有就直接返回了；否则就拿当前系统是使用哪种绘制方式 const DrawMode draw_mode = GetDrawMode();

enum DrawMode {
 DRAW_MODE_NONE,
 DRAW_MODE_HARDWARE,
 DRAW_MODE_SOFTWARE,
 DRAW_MODE_RESOURCELESS_SOFTWARE
};

如果是 DRAW_MODE_RESOURCELESS_SOFTWARE 模式，这里会新创建一个临时的SoftRender来绘制；否则调用layer_tree_host_impl 内部的 SoftRender对象来完成绘制，在DrawFrame中，每完成一部分的绘制，会根据需要copy到bitmap当中；

void DirectRenderer::DrawFrame(RenderPassList* render_passes_in_draw_order,
 float device_scale_factor,
 const gfx::Rect& device_viewport_rect,
 const gfx::Rect& device_clip_rect,
 bool disable_picture_quad_image_filtering) {
 TRACE_EVENT0("cc", "DirectRenderer::DrawFrame");
 UMA_HISTOGRAM_COUNTS("Renderer4.renderPassCount",
 render_passes_in_draw_order->size());
 const RenderPass* root_render_pass = render_passes_in_draw_order->back();
 DCHECK(root_render_pass);
 DrawingFrame frame;
 frame.render_passes_in_draw_order = render_passes_in_draw_order;
 frame.root_render_pass = root_render_pass;
 frame.root_damage_rect = Capabilities().using_partial_swap
 ? root_render_pass->damage_rect
 : root_render_pass->output_rect;
 frame.root_damage_rect.Intersect(gfx::Rect(device_viewport_rect.size()));
 frame.device_viewport_rect = device_viewport_rect;
 frame.device_clip_rect = device_clip_rect;
 frame.disable_picture_quad_image_filtering =
 disable_picture_quad_image_filtering;
 overlay_processor_->ProcessForOverlays(render_passes_in_draw_order,
 &frame.overlay_list);
 EnsureBackbuffer();
 // Only reshape when we know we are going to draw. Otherwise, the reshape
 // can leave the window at the wrong size if we never draw and the proper
 // viewport size is never set.
 output_surface_->Reshape(device_viewport_rect.size(), device_scale_factor);
 BeginDrawingFrame(&frame);
 for (size_t i = 0; i < render_passes_in_draw_order->size(); ++i) {
 RenderPass* pass = render_passes_in_draw_order->at(i);
 DrawRenderPass(&frame, pass);
 for (ScopedPtrVector<CopyOutputRequest>::iterator it =
 pass->copy_requests.begin();
 it != pass->copy_requests.end();
 ++it) {
 if (i > 0) {
 // Doing a readback is destructive of our state on Mac, so make sure
 // we restore the state between readbacks. http://crbug.com/99393.
 UseRenderPass(&frame, pass);
 }
 CopyCurrentRenderPassToBitmap(&frame, pass->copy_requests.take(it));
 }
 }
 FinishDrawingFrame(&frame);
 render_passes_in_draw_order->clear();
}

当绘制完成后调用FinishDrawingFrame(&frame);来完成绘制；绘制完成后调用FinishDrawingFrame函数来通知SoftwareOutputDevice::EndPaint() 来将绘制完成的数据copy到显示后端去完成显示；比如说在我们嵌入式linux平台是，我们对接到dfb的surface上；之后通过dfbwindow的相关接口跟新显示；

很遗憾的是这部分在新的chromium中变动还是很大的，我这部分的分析是基于v39上；在新的版本v50版本上这部分很多函数都有所更新，不过整体的流程还是大致一样额，具体可以到最新的代码上看一下；

LayerTreeHostImpl::DidDrawAllLayers(const FrameData& frame);

ProxyImpl: LayerTreeHostImplClient, SchedulerClient

// LayerTreeHost<-->ProxyMain<-->ChannelMain // | // | // ChannelImpl<-->ProxyImpl<-->LayerTreeHostImpl

RenderWidget::SetNeedsRedrawRect

RenderWidgetCompositor::SetNeedsRedrawRect(const gfx::Recctt&damagerect)

renderwidget 维护一个 RenderWidgetCompositor 【compositor】对象 RenderWidgetCompositor里面有cc::LayerTreeHost 对象； RenderWidget::SetNeedsRedrawRect

RenderWidgetCompositor::SetNeedsRedrawRect(const gfx::Recctt&damagerect)

renderwidget 维护一个 RenderWidgetCompositor 【compositor】对象

RenderWidgetCompositor里面有cc::LayerTreeHost 对象；

std::unique_ptr<RenderWidgetHostIterator> widgets(
      RenderWidgetHostImpl::GetAllRenderWidgetHosts());
  while (RenderWidgetHost* widget = widgets->GetNextHost()) {
    RenderViewHost* rvh = RenderViewHost::From(widget);
    if (!rvh)
      continue;

    // Skip widgets in other processes.
    if (rvh->GetProcess()->GetID() != GetID())
      continue;

    rvh->OnWebkitPreferencesChanged();

HelloMessage PluginProcessHost::OnChannelConnected(int32 peer_pid)

这个调用是从browserChildProcessHostDelegate处继承来的，当BrowserChiledProssHost接收到hellomsg之后就转发到了PluginProcessHost， V47中新增的mediaHost同样的流程；

在每个进程Connect来的时候，同步所有的config到对应的子进程中....

向render进程发送设置可以通过render_process_host_impl.cc 发送 ChildProcessMsg_xxx 个对应的render进程；

有静态函数RenderProcessHost::AllHostIterator();可以拿到所有的render_process_host 来发消息

drafts

results matching ""

No results matching ""