Added tombstones in instrumentation tests results.

build/android/pylib/local/device/local_device_instrumentation_test_run.py

Index: build/android/pylib/local/device/local_device_instrumentation_test_run.py
diff --git a/build/android/pylib/local/device/local_device_instrumentation_test_run.py b/build/android/pylib/local/device/local_device_instrumentation_test_run.py
index 26d10f4d1d7ed797a4682a0acae6ff13b382281e..eb8c9ab87f0f5eeb1d7e09abcdb16c09e5f72a43 100644
--- a/build/android/pylib/local/device/local_device_instrumentation_test_run.py
+++ b/build/android/pylib/local/device/local_device_instrumentation_test_run.py
@@ -6,6 +6,7 @@ import logging
 import os
 import posixpath
 import re
+import sys
 import time
 
 from devil.android import device_errors
@@ -16,6 +17,10 @@ from pylib.base import base_test_result
 from pylib.local.device import local_device_environment
 from pylib.local.device import local_device_test_run
 
+from pylib.constants import host_paths
+sys.path.append(os.path.join(host_paths.DIR_SOURCE_ROOT, 'build', 'android'))
+import tombstones
+
 
 TIMEOUT_ANNOTATIONS = [
   ('Manual', 10 * 60 * 60),
@@ -143,6 +148,8 @@ class LocalDeviceInstrumentationTestRun(
       else:
         for step in steps:
           step()
+      if self._test_instance.store_tombstones:
+        tombstones.ClearAllTombstones(dev)

[mikecase (-- gone --), 2016/08/04 01:12:05]

There is a recipe step on our infrastructure that collects all the tombstones at
the end of the test runs and lists them out. Will this break that script? I
suppose as long as store_tombstones defaults to False to begin that shouldnt be
a problem.

[BigBossZhiling, 2016/08/04 23:20:28]

Acknowledged.

 
     self._env.parallel_devices.pMap(
         individual_device_set_up,
@@ -329,6 +336,15 @@ class LocalDeviceInstrumentationTestRun(
           self._test_instance.coverage_directory)
       device.RunShellCommand('rm -f %s' % os.path.join(coverage_directory,
           '*'))
+    if self._test_instance.store_tombstones:
+      for result in results:
+        if result.GetType() == base_test_result.ResultType.FAIL:
+          tombstones.ResolveTombstones(
+            device,

[mikecase (-- gone --), 2016/08/04 01:12:05]

nit: wrong indent. Need 4 spaces.

[BigBossZhiling, 2016/08/04 23:20:28]

Done.

+            resolve_all_tombstones=True,
+            include_stack_symbols=False,
+            wipe_tombstones=True)
+          result.SetTombstones(tombstones)

[mikecase (-- gone --), 2016/08/04 01:12:05]

So, Im not too sure what tombstones really are to be honest. Do you know if they
are generated by the death tests? 

[BigBossZhiling, 2016/08/04 23:20:28]

What are death tests? I think if a test crashes, then a tombstone will be
created. Passing and Failing testcases do not usually create tombstones. I also
sent out an email containing a sample tombstone.

     return results
 
   #override