Make bisect to abort early when the return codes for known good and known bad revisions are same.

tools/auto_bisect/bisect_perf_regression.py

Index: tools/auto_bisect/bisect_perf_regression.py
diff --git a/tools/auto_bisect/bisect_perf_regression.py b/tools/auto_bisect/bisect_perf_regression.py
index ca97d67e422649087a3561debd89bdb8f94d169c..a4150a3e94dbc03aea0de33dd747af3e082dcedf 100755
--- a/tools/auto_bisect/bisect_perf_regression.py
+++ b/tools/auto_bisect/bisect_perf_regression.py
@@ -2307,9 +2307,18 @@ class BisectPerformanceMetrics(object):
 
       # We need these reference values to determine if later runs should be
       # classified as pass or fail.
+
       known_bad_value = bad_results[0]
       known_good_value = good_results[0]
 
+      # Early abort bisection when the return_codes for known good

[qyearsley, 2015/06/23 20:47:45]

Phrasing nits:
 "Early abort bisection" -> "Abort bisect early"
 "return_codes" -> "return codes"

[prasadv, 2015/06/23 21:37:48]

Done.

+      # and known bad revisions are same.
+      if (self._IsBisectModeReturnCode() and
+          known_bad_value['mean'] == known_good_value['mean']):
+        return BisectResults(abort_reason=('known good and known bad revisions '
+            'returned same return code (return code=%s). '
+            'Continuing bisect might not yield any results.' %
+            known_bad_value['mean']))
       # Check the direction of improvement only if the improvement_direction
       # option is set to a specific direction (1 for higher is better or -1 for
       # lower is better).