Temporary fix for installer/visitedlink crash (crbug.com/171475)

chrome/browser/visitedlink/visitedlink_event_listener.cc

Index: chrome/browser/visitedlink/visitedlink_event_listener.cc
===================================================================
--- chrome/browser/visitedlink/visitedlink_event_listener.cc	(revision 181040)
+++ chrome/browser/visitedlink/visitedlink_event_listener.cc	(working copy)
@@ -116,6 +116,17 @@
                  content::NotificationService::AllBrowserContextsAndSources());
   registrar_.Add(this, content::NOTIFICATION_RENDER_WIDGET_VISIBILITY_CHANGED,
                  content::NotificationService::AllBrowserContextsAndSources());
+  for (content::RenderProcessHost::iterator i(
+         content::RenderProcessHost::AllHostsIterator());
+       !i.IsAtEnd(); i.Advance()) {
+    content::RenderProcessHost* process = i.GetCurrentValue();
+    Profile* rph_profile =
+        Profile::FromBrowserContext(process->GetBrowserContext());
+    if (!profile_->IsSameProfile(rph_profile))
+      continue;
+    updaters_[process->GetID()] =

[brettw, 2013/02/13 21:57:24]

Below in the NOTIFICATION_RENDERER_PROCESS_CREATED we also send the table. Here
we don't. Why don't we need to send the RPH's the table in this case?

[Cait (Slow), 2013/02/14 16:58:13]

We do need to send the table here. This was a mistake on my part. I looked into
sending the table and it looks like this will require changing the init process
quite a bit (the table is not ready yet when the Listener is created).

As it stands, this code isn't doing much, so I've removed it. 

+          make_linked_ptr(new VisitedLinkUpdater(process->GetID()));
+  }
 }
 
 VisitedLinkEventListener::~VisitedLinkEventListener() {