Avoid exiting the renderer process if it has a pending render view.

content/browser/tab_contents/tab_contents.cc

Index: content/browser/tab_contents/tab_contents.cc
diff --git a/content/browser/tab_contents/tab_contents.cc b/content/browser/tab_contents/tab_contents.cc
index 4119f0f22b2fbe09950f2b64a65705a2eda15241..d644abb8b4675c5ffeec5043588c19578feb9208 100644
--- a/content/browser/tab_contents/tab_contents.cc
+++ b/content/browser/tab_contents/tab_contents.cc
@@ -259,6 +259,8 @@ TabContents::TabContents(Profile* profile,
       pref_change_registrar_.Add(kPrefsToObserve[i], this);
   }
 
+  registrar_.Add(this, NotificationType::RENDERER_PROCESS_CLOSING,
+                 NotificationService::AllSources());
   registrar_.Add(this, NotificationType::RENDER_WIDGET_HOST_DESTROYED,
                  NotificationService::AllSources());
 #if defined(OS_LINUX)
@@ -2148,6 +2150,11 @@ void TabContents::Observe(NotificationType type,
       }
       break;
     }
+    case NotificationType::RENDERER_PROCESS_CLOSING:

[jam, 2011/05/06 17:45:11]

why do you listen to this in TabContents and then just have it call
RenderManager?  i.e. why doesn't render_manager_ listen to the notification
itself?

[Charlie Reis, 2011/05/10 00:57:03]

RenderViewHostManager wasn't a NotificationObserver, and adding a registrar
seemed unnecessary if TabContents is already notifying RVHM about events from
the outside world.

I've switched it over as you suggest, though, which lets us avoid exposing the
RendererProcessClosing function.

+      render_manager_.RendererProcessClosing(
+          Source<RenderProcessHost>(source).ptr());
+      break;
+
     case NotificationType::RENDER_WIDGET_HOST_DESTROYED:
       view_->RenderWidgetHostDestroyed(Source<RenderWidgetHost>(source).ptr());
       break;