Fix silent failures in test runner that lead to long timeouts

build/android/pylib/perf/test_runner.py

Index: build/android/pylib/perf/test_runner.py
diff --git a/build/android/pylib/perf/test_runner.py b/build/android/pylib/perf/test_runner.py
index cedf5734d8315bd6947f12c10b13140246a8ea57..71a613e93c0e1c5903c9fc7a87ed4d6b1bd513ae 100644
--- a/build/android/pylib/perf/test_runner.py
+++ b/build/android/pylib/perf/test_runner.py
@@ -49,9 +49,9 @@ import sys
 import threading
 import time
 
+from pylib import cmd_helper
 from pylib import constants
 from pylib import forwarder
-from pylib import pexpect
 from pylib.base import base_test_result
 from pylib.base import base_test_runner
 
@@ -200,13 +200,12 @@ class TestRunner(base_test_runner.BaseTestRunner):
     cwd = os.path.abspath(constants.DIR_SOURCE_ROOT)
     if full_cmd.startswith('src/'):
       cwd = os.path.abspath(os.path.join(constants.DIR_SOURCE_ROOT, os.pardir))
-    output, exit_code = pexpect.run(
-        full_cmd, cwd=cwd,
-        withexitstatus=True, logfile=logfile, timeout=timeout,
-        env=os.environ)
-    if self._options.single_step:
-      # Stop the logger.
-      logfile.stop()
+    try:
+      exit_code, output = cmd_helper.GetCmdStatusAndOutputWithTimeout(

[Andrew Hayden (chromium.org), 2014/05/28 21:43:37]

Minor thing but for the sake of conformity maybe we should reverse the order of
the returned tuple to match the existing code?

+          full_cmd, timeout, cwd=cwd, shell=True, logfile=logfile)
+    finally:
+      if self._options.single_step:
+        logfile.stop()
     end_time = datetime.datetime.now()
     if exit_code is None:
       exit_code = -1