Android: makes "RestartAdbdOnDevice" more robust.

build/android/pylib/android_commands.py

Index: build/android/pylib/android_commands.py
diff --git a/build/android/pylib/android_commands.py b/build/android/pylib/android_commands.py
index 6f9b99cc2405f9eab6760fe5d71d4fa5c5a9196c..6d999c867b21238b26ddbcdded92b2cba5866fc6 100644
--- a/build/android/pylib/android_commands.py
+++ b/build/android/pylib/android_commands.py
@@ -465,10 +465,13 @@ class AndroidCommands(object):
 
   def RestartAdbdOnDevice(self):
     logging.info('Killing adbd on the device...')
-    adb_pids = self.KillAll('adbd', signal=signal.SIGTERM, with_su=True)
-    assert adb_pids, 'Unable to obtain adbd pid'
-    logging.info('Waiting for device to settle...')
-    self._adb.SendCommand('wait-for-device')
+    try:
+      if not self.KillAll('adbd', signal=signal.SIGTERM, with_su=True):
+        logging.error('Unable to obtain adbd pid')
+      logging.info('Waiting for device to settle...')
+      self._adb.SendCommand('wait-for-device')

[tonyg, 2013/11/27 16:24:15]

The idea of this wait-for-device is that it must be done after killing adbd.
However, if it needs to be done for some other reason, then callers are relying
on an unintended side effect of this method.

I think we should only do this if we actually kill adbd. That will flush out any
callers or other places that need their own wait-for-device for other reasons.

[bulach, 2013/11/27 18:04:27]

you're right. I was mistaken, KillAll just returns the len of pids it sent the
signal to, no guarantees if the procs were actually restarted...
I made it slightly more robust by checking that the pid has indeed changed, and
printing an error if they haven't. wdyt?

+    except Except as e:

[tonyg, 2013/11/27 16:24:15]

s/Except/Exception/ -- no?

[bulach, 2013/11/27 18:04:27]

facepalm :(

+      logging.error('Exception when trying to kill adbd on the device [%s]', e)
 
   def RestartAdbServer(self):
     """Restart the adb server."""