Create new GLSurface for cross process image transport.
Currently cross process image transport (on OSX and TOUCH_UI) uses the WebGL code path to render to a texture via FBO. This makes a clean break with path and centralizes the GL context specific IPC communication in one location.
By separating from the WebGL code, I was able to avoid wasting a texture attached to the FBO in the decoder (the old code ignored this attachment --- the new code makes no such attachment at the decoder level).
As a test of the new architecture, I have also implemented XComposite/GLX_texture_from_pixmap image transport (i.e. it's now possible to use the --use-gl=desktop backend on touch_ui builds).
BUG=none
TEST=3D CSS and WebGL on TOUCH_UI.
Review URL: http://codereview.chromium.org/7395020
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@93681 0039d316-1c4b-4281-b951-d872f2087c98
diff --git a/content/common/gpu/gpu_command_buffer_stub.cc b/content/common/gpu/gpu_command_buffer_stub.cc
index 3890e81..b86750a 100644
--- a/content/common/gpu/gpu_command_buffer_stub.cc
+++ b/content/common/gpu/gpu_command_buffer_stub.cc
@@ -22,6 +22,8 @@
#if defined(OS_WIN)
#include "base/win/wrapped_window_proc.h"
+#elif defined(TOUCH_UI)
+#include "content/common/gpu/image_transport_surface_linux.h"
#endif
using gpu::Buffer;
@@ -115,14 +117,19 @@
return !scheduler_.get() || scheduler_->IsScheduled();
}
+void GpuCommandBufferStub::OnInitializeFailed(IPC::Message* reply_message) {
+ scheduler_.reset();
+ command_buffer_.reset();
+ GpuCommandBufferMsg_Initialize::WriteReplyParams(reply_message, false);
+ Send(reply_message);
+}
+
void GpuCommandBufferStub::OnInitialize(
base::SharedMemoryHandle ring_buffer,
int32 size,
IPC::Message* reply_message) {
DCHECK(!command_buffer_.get());
- bool result = false;
-
command_buffer_.reset(new gpu::CommandBufferService);
#if defined(OS_WIN)
@@ -142,6 +149,41 @@
scheduler_.reset(gpu::GpuScheduler::Create(command_buffer_.get(),
channel_,
NULL));
+#if defined(TOUCH_UI)
+ if (software_) {
+ OnInitializeFailed(reply_message);
+ return;
+ }
+
+ scoped_refptr<gfx::GLSurface> surface;
+ if (handle_)
+ surface = ImageTransportSurface::CreateSurface(this);
+ else
+ surface = gfx::GLSurface::CreateOffscreenGLSurface(software_,
+ gfx::Size(1, 1));
+ if (!surface.get()) {
+ LOG(ERROR) << "GpuCommandBufferStub: failed to create surface.";
+ OnInitializeFailed(reply_message);
+ return;
+ }
+
+ scoped_refptr<gfx::GLContext> context(
+ gfx::GLContext::CreateGLContext(channel_->share_group(),
+ surface.get()));
+ if (!context.get()) {
+ LOG(ERROR) << "GpuCommandBufferStub: failed to create context.";
+ OnInitializeFailed(reply_message);
+ return;
+ }
+
+ if (scheduler_->InitializeCommon(
+ surface,
+ context,
+ initial_size_,
+ disallowed_extensions_,
+ allowed_extensions_.c_str(),
+ requested_attribs_)) {
+#else
if (scheduler_->Initialize(
handle_,
initial_size_,
@@ -150,6 +192,7 @@
allowed_extensions_.c_str(),
requested_attribs_,
channel_->share_group())) {
+#endif
command_buffer_->SetPutOffsetChangeCallback(
NewCallback(scheduler_.get(),
&gpu::GpuScheduler::PutChanged));
@@ -165,7 +208,7 @@
scheduler_->SetCommandProcessedCallback(
NewCallback(this, &GpuCommandBufferStub::OnCommandProcessed));
-#if defined(OS_MACOSX) || defined(TOUCH_UI)
+#if defined(OS_MACOSX)
if (handle_) {
// This context conceptually puts its output directly on the
// screen, rendered by the accelerated plugin layer in
@@ -175,12 +218,19 @@
NewCallback(this,
&GpuCommandBufferStub::SwapBuffersCallback));
}
-#endif // defined(OS_MACOSX) || defined(TOUCH_UI)
+#endif // defined(OS_MACOSX)
// Set up a pathway for resizing the output window or framebuffer at the
// right time relative to other GL commands.
+#if defined(TOUCH_UI)
+ if (handle_ == gfx::kNullPluginWindow) {
+ scheduler_->SetResizeCallback(
+ NewCallback(this, &GpuCommandBufferStub::ResizeCallback));
+ }
+#else
scheduler_->SetResizeCallback(
NewCallback(this, &GpuCommandBufferStub::ResizeCallback));
+#endif
if (parent_stub_for_initialization_) {
scheduler_->SetParent(parent_stub_for_initialization_->scheduler_.get(),
@@ -189,14 +239,13 @@
parent_texture_for_initialization_ = 0;
}
- result = true;
} else {
- scheduler_.reset();
- command_buffer_.reset();
+ OnInitializeFailed(reply_message);
+ return;
}
}
- GpuCommandBufferMsg_Initialize::WriteReplyParams(reply_message, result);
+ GpuCommandBufferMsg_Initialize::WriteReplyParams(reply_message, true);
Send(reply_message);
}
@@ -438,34 +487,7 @@
scheduler_->SetScheduled(false);
}
-#endif // defined(OS_MACOSX)
-#if defined(TOUCH_UI)
-void GpuCommandBufferStub::SwapBuffersCallback() {
- TRACE_EVENT0("gpu", "GpuCommandBufferStub::SwapBuffersCallback");
- GpuChannelManager* gpu_channel_manager = channel_->gpu_channel_manager();
- GpuHostMsg_AcceleratedSurfaceBuffersSwapped_Params params;
- params.renderer_id = renderer_id_;
- params.render_view_id = render_view_id_;
- params.surface_id = scheduler_->GetFrontSurfaceId();
- params.route_id = route_id();
- params.swap_buffers_count = scheduler_->swap_buffers_count();
- gpu_channel_manager->Send(
- new GpuHostMsg_AcceleratedSurfaceBuffersSwapped(params));
-
- scheduler_->SetScheduled(false);
-}
-
-void GpuCommandBufferStub::AcceleratedSurfaceIOSurfaceSet(uint64 surface_id) {
- scheduler_->SetScheduled(true);
-}
-
-void GpuCommandBufferStub::AcceleratedSurfaceReleased(uint64 surface_id) {
- scheduler_->ReleaseSurface(surface_id);
-}
-#endif // defined(TOUCH_UI)
-
-#if defined(OS_MACOSX) || defined(TOUCH_UI)
void GpuCommandBufferStub::AcceleratedSurfaceBuffersSwapped(
uint64 swap_buffers_count) {
TRACE_EVENT1("gpu",
@@ -485,7 +507,7 @@
scheduler_->SetScheduled(true);
}
}
-#endif // defined(OS_MACOSX) || defined(TOUCH_UI)
+#endif // defined(OS_MACOSX)
void GpuCommandBufferStub::AddSetTokenCallback(
const base::Callback<void(int32)>& callback) {
@@ -511,31 +533,6 @@
size));
scheduler_->SetScheduled(false);
-#elif defined(TOUCH_UI)
- if (scheduler_->GetBackSurfaceId()) {
- GpuHostMsg_AcceleratedSurfaceRelease_Params params;
- params.renderer_id = renderer_id_;
- params.render_view_id = render_view_id_;
- params.identifier = scheduler_->GetBackSurfaceId();
- params.route_id = route_id();
-
- GpuChannelManager* gpu_channel_manager = channel_->gpu_channel_manager();
- gpu_channel_manager->Send(
- new GpuHostMsg_AcceleratedSurfaceRelease(params));
- }
- scheduler_->CreateBackSurface(size);
- GpuHostMsg_AcceleratedSurfaceSetIOSurface_Params params;
- params.renderer_id = renderer_id_;
- params.render_view_id = render_view_id_;
- params.width = size.width();
- params.height = size.height();
- params.identifier = scheduler_->GetBackSurfaceId();
- params.route_id = route_id();
-
- GpuChannelManager* gpu_channel_manager = channel_->gpu_channel_manager();
- gpu_channel_manager->Send(
- new GpuHostMsg_AcceleratedSurfaceSetIOSurface(params));
- scheduler_->SetScheduled(false);
#endif
}
}
