Added support for EXTDisjointTimerQuery on the Blink side.

Source/modules/webgl/WebGLTimerQueryEXT.cpp

+// 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.
+
+#include "config.h"
+
+#include "modules/webgl/WebGLTimerQueryEXT.h"
+
+#include "modules/webgl/WebGLRenderingContextBase.h"
+
+namespace blink {
+
+WebGLTimerQueryEXT* WebGLTimerQueryEXT::create(WebGLRenderingContextBase* ctx)
+{
+    return new WebGLTimerQueryEXT(ctx);
+}
+
+WebGLTimerQueryEXT::~WebGLTimerQueryEXT()
+{
+    // See the comment in WebGLObject::detachAndDeleteObject().
+    detachAndDeleteObject();
+}
+
+WebGLTimerQueryEXT::WebGLTimerQueryEXT(WebGLRenderingContextBase* ctx)
+    : WebGLContextObject(ctx)
+    , m_target(0)
+    , m_queryId(0)
+{
+    m_queryId = context()->webContext()->createQueryEXT();
+}
+
+void WebGLTimerQueryEXT::deleteObjectImpl(WebGraphicsContext3D*)
+{
+    context()->webContext()->deleteQueryEXT(m_queryId);

[haraken, 2015/09/03 23:32:58]

This looks unsafe. deleteObjectImpl() is called in a destructor of
WebGLTimerQueryEXT and touches context(), which is an on-heap object. A
destructor is not allowed to touch other on-heap objects.

Can we change the code to context3D->deleteQueryEXT(m_queryId)?

[Ken Russell (switch to Gerrit), 2015/09/04 01:17:09]

This is a good point and should be done, but it looks like there's a further,
similar bug in WebGLContextObject::getAWebGraphicsContext3D. haraken@: should
that class be changed to cache the WebGraphicsContext3D* pointer? If David fixes
this case here, can I fix that in a follow-on patch?

[haraken, 2015/09/04 01:30:13]

In conclusion, I realized that this code is safe for a complicated reason below.
Either way, it would be better to use the passed-in context3D for consistency
with other WebGLObject::deleteObjectImpls.

a) If the WebGL object is finalized before the context object, the
deleteObjectImpl gets called. Since the context object is still alive, it is
safe to touch the context object in the deleteObjectImpl.

b) If the context object is finalized before the WebGL object, the
deleteObjectImpl isn't called (See
https://code.google.com/p/chromium/codesearch#chromium/src/third_party/WebKit...).
So we won't reach deleteObjectImpl in this case.

Sigbjorn: We should have a good comment about the lifetime relationship between
WebGLObjects and the context object.

[David Yen, 2015/09/04 16:04:41]

Done.

+    m_queryId = 0;
+}
+
+} // namespace blink