Fixed RLZTracker::GetAccessPointRlz to route IO calls to the FILE thread.

chrome/browser/search_engines/search_terms_data.cc

Index: chrome/browser/search_engines/search_terms_data.cc
diff --git a/chrome/browser/search_engines/search_terms_data.cc b/chrome/browser/search_engines/search_terms_data.cc
index 7e5b9ffb7dc14b218c8848458fe6982873312441..a0e5fc01802c98b5c7865f4965fa0462866750a5 100644
--- a/chrome/browser/search_engines/search_terms_data.cc
+++ b/chrome/browser/search_engines/search_terms_data.cc
@@ -1,4 +1,4 @@
-// Copyright (c) 2010 The Chromium Authors. All rights reserved.
+// Copyright (c) 2011 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.
 
@@ -80,11 +80,13 @@ std::wstring UIThreadSearchTermsData::GetRlzParameterValue() const {
   // For organic brandcodes do not use rlz at all. Empty brandcode usually
   // means a chromium install. This is ok.
   std::wstring brand;
-  // See http://crbug.com/62337.
-  base::ThreadRestrictions::ScopedAllowIO allow_io;
   if (GoogleUpdateSettings::GetBrand(&brand) && !brand.empty() &&
-      !GoogleUpdateSettings::IsOrganic(brand))
+      !GoogleUpdateSettings::IsOrganic(brand)) {
+    // This call will return false the first time(s) it is called until the
+    // value has been cached. This normally would mean that at most one omnibox
+    // search might not send the RLZ data but this is not really a problem.

[Roger Tawa OOO till Jul 10th, 2011/01/12 15:46:52]

There will be a new "first time" each time chrome is restarted.  But since
DelayedInitTask calls GetAccessPointRlz(), that helps reduce the chance of
getting an empty string here.

Do we have any metrics that tell us how many searches are performed for each
invocation of chrome, or time to first search after launch?  It would be
interesting to know what percentage of searches will not have an rlz parameter
due to this change.  I would also discuss this with the PSO folks to determine
if this is not really a problem.

     RLZTracker::GetAccessPointRlz(rlz_lib::CHROME_OMNIBOX, &rlz_string);
+  }
   return rlz_string;
 }
 #endif