EQT: Record the max queueing time from navigation start to navigation away

content/renderer/render_frame_impl.cc

 // Copyright 2013 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 "content/renderer/render_frame_impl.h"
 
 #include <map>
 #include <string>
 #include <utility>
 #include <vector>
@@ skipping 3642 matching lines @@
     if (render_thread_impl) {  // Can be NULL in tests.
       render_thread_impl->histogram_customizer()->
           RenderViewNavigatedToHost(GURL(GetLoadingUrl()).host(),
                                     RenderView::GetRenderViewCount());
       // The scheduler isn't interested in history inert commits unless they
       // are reloads.
       if (commit_type != blink::kWebHistoryInertCommit ||
           PageTransitionCoreTypeIs(navigation_state->GetTransitionType(),
                                    ui::PAGE_TRANSITION_RELOAD)) {
         render_thread_impl->GetRendererScheduler()->OnNavigationStarted();
+        render_thread_impl->GetRendererScheduler()->OnCommitProvisionalLoad();

[Charlie Reis, 2017/06/02 22:38:36]

Why do we need both?

[Liquan (Max) Gu, 2017/06/02 23:32:06]

There are two OnNavigationStarted(). One is this; another is in
RenderFrameImpl::OnNavigate(). By experiment, it shows that OnNavigate() is not
triggered from the second navigation of the same renderer, but this line is
triggered on every navigation start.

That's why we need a different signal from OnNavigationStarted().

I guess OnCommitProvisionalLoadAndNavigationStarted() will be more precise here,
based on the position of the code. Do you think so?

[Charlie Reis, 2017/06/05 05:02:34]

No, I think OnCommitProvisionalLoad was a better name.  (See my other comment
for more detail.)

My confusion here was why we still had OnNavigationStarted at all, since I
thought we were talking about removing it.  But looking closer, it looks like
you're hopefully planning to remove OnNavigationStarted in a followup CL as part
of https://crbug.com/726341.  If that's the case, I'm ok with it, but please
include a TODO to remove OnNavigationStarted (citing bug 726341).

[Liquan (Max) Gu, 2017/06/05 16:15:56]

Thanks for the detailed explanation. I have added a todo and will address
onNavigationStart in a followup CL.

       }
     }
   }
 
   // Remember that we've already processed this request, so we don't update
   // the session history again.  We do this regardless of whether this is
   // a session history navigation, because if we attempted a session history
   // navigation without valid HistoryItem state, WebCore will think it is a
   // new navigation.
   navigation_state->set_request_committed(true);
@@ skipping 3109 matching lines @@
       policy(info.default_policy),
       replaces_current_history_item(info.replaces_current_history_item),
       history_navigation_in_new_child_frame(
           info.is_history_navigation_in_new_child_frame),
       client_redirect(info.is_client_redirect),
       cache_disabled(info.is_cache_disabled),
       form(info.form),
       source_location(info.source_location) {}
 
 }  // namespace content