1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
|
# Copyright 2013 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.
"""Runs perf tests.
Our buildbot infrastructure requires each slave to run steps serially.
This is sub-optimal for android, where these steps can run independently on
multiple connected devices.
The buildbots will run this script multiple times per cycle:
- First: all steps listed in --steps in will be executed in parallel using all
connected devices. Step results will be pickled to disk. Each step has a unique
name. The result code will be ignored if the step name is listed in
--flaky-steps.
The buildbot will treat this step as a regular step, and will not process any
graph data.
- Then, with -print-step STEP_NAME: at this stage, we'll simply print the file
with the step results previously saved. The buildbot will then process the graph
data accordingly.
The JSON steps file contains a dictionary in the format:
{ "version": int,
"steps": {
"foo": {
"device_affinity": int,
"cmd": "script_to_execute foo"
},
"bar": {
"device_affinity": int,
"cmd": "script_to_execute bar"
}
}
}
The JSON flaky steps file contains a list with step names which results should
be ignored:
[
"step_name_foo",
"step_name_bar"
]
Note that script_to_execute necessarily have to take at least the following
option:
--device: the serial number to be passed to all adb commands.
"""
import collections
import io
import json
import logging
import os
import pickle
import re
import shutil
import sys
import tempfile
import threading
import time
import zipfile
from devil.android import battery_utils
from devil.android import device_errors
from devil.android import forwarder
from devil.constants import exit_codes
from devil.utils import cmd_helper
from pylib import constants
from pylib.base import base_test_result
from pylib.base import base_test_runner
# Regex for the master branch commit position.
_GIT_CR_POS_RE = re.compile(r'^Cr-Commit-Position: refs/heads/master@{#(\d+)}$')
def _GetChromiumRevision():
# pylint: disable=line-too-long
"""Get the git hash and commit position of the chromium master branch.
See: https://chromium.googlesource.com/chromium/tools/build/+/master/scripts/slave/runtest.py#212
Returns:
A dictionary with 'revision' and 'commit_pos' keys.
"""
# pylint: enable=line-too-long
status, output = cmd_helper.GetCmdStatusAndOutput(
['git', 'log', '-n', '1', '--pretty=format:%H%n%B', 'HEAD'],
constants.DIR_SOURCE_ROOT)
revision = None
commit_pos = None
if not status:
lines = output.splitlines()
revision = lines[0]
for line in reversed(lines):
m = _GIT_CR_POS_RE.match(line.strip())
if m:
commit_pos = int(m.group(1))
break
return {'revision': revision, 'commit_pos': commit_pos}
def GetPersistedResult(test_name):
file_name = os.path.join(constants.PERF_OUTPUT_DIR, test_name)
if not os.path.exists(file_name):
logging.error('File not found %s', file_name)
return None
with file(file_name, 'r') as f:
return pickle.loads(f.read())
def OutputJsonList(json_input, json_output):
with file(json_input, 'r') as i:
all_steps = json.load(i)
step_values = []
for k, v in all_steps['steps'].iteritems():
data = {'test': k, 'device_affinity': v['device_affinity']}
persisted_result = GetPersistedResult(k)
if persisted_result:
data['start_time'] = persisted_result['start_time']
data['end_time'] = persisted_result['end_time']
data['total_time'] = persisted_result['total_time']
data['has_archive'] = persisted_result['archive_bytes'] is not None
step_values.append(data)
with file(json_output, 'w') as o:
o.write(json.dumps(step_values))
return 0
def PrintTestOutput(test_name, json_file_name=None, archive_file_name=None):
"""Helper method to print the output of previously executed test_name.
Args:
test_name: name of the test that has been previously executed.
json_file_name: name of the file to output chartjson data to.
archive_file_name: name of the file to write the compressed ZIP archive.
Returns:
exit code generated by the test step.
"""
persisted_result = GetPersistedResult(test_name)
if not persisted_result:
return exit_codes.INFRA
logging.info('*' * 80)
logging.info('Output from:')
logging.info(persisted_result['cmd'])
logging.info('*' * 80)
output_formatted = ''
persisted_outputs = persisted_result['output']
for i in xrange(len(persisted_outputs)):
output_formatted += '\n\nOutput from run #%d:\n\n%s' % (
i, persisted_outputs[i])
print output_formatted
if json_file_name:
with file(json_file_name, 'w') as f:
f.write(persisted_result['chartjson'])
if archive_file_name:
if persisted_result['archive_bytes'] is not None:
with file(archive_file_name, 'wb') as f:
f.write(persisted_result['archive_bytes'])
else:
logging.error('The output dir was not archived.')
return persisted_result['exit_code']
def PrintSummary(test_names):
logging.info('*' * 80)
logging.info('Sharding summary')
device_total_time = collections.defaultdict(int)
for test_name in test_names:
file_name = os.path.join(constants.PERF_OUTPUT_DIR, test_name)
if not os.path.exists(file_name):
logging.info('%s : No status file found', test_name)
continue
with file(file_name, 'r') as f:
result = pickle.loads(f.read())
logging.info('%s : exit_code=%d in %d secs at %s',
result['name'], result['exit_code'], result['total_time'],
result['device'])
device_total_time[result['device']] += result['total_time']
for device, device_time in device_total_time.iteritems():
logging.info('Total for device %s : %d secs', device, device_time)
logging.info('Total steps time: %d secs', sum(device_total_time.values()))
class _HeartBeatLogger(object):
# How often to print the heartbeat on flush().
_PRINT_INTERVAL = 30.0
def __init__(self):
"""A file-like class for keeping the buildbot alive."""
self._len = 0
self._tick = time.time()
self._stopped = threading.Event()
self._timer = threading.Thread(target=self._runner)
self._timer.start()
def _runner(self):
while not self._stopped.is_set():
self.flush()
self._stopped.wait(_HeartBeatLogger._PRINT_INTERVAL)
def write(self, data):
self._len += len(data)
def flush(self):
now = time.time()
if now - self._tick >= _HeartBeatLogger._PRINT_INTERVAL:
self._tick = now
print '--single-step output length %d' % self._len
sys.stdout.flush()
def stop(self):
self._stopped.set()
class TestRunner(base_test_runner.BaseTestRunner):
def __init__(self, test_options, device, shard_index, max_shard, tests,
flaky_tests):
"""A TestRunner instance runs a perf test on a single device.
Args:
test_options: A PerfOptions object.
device: Device to run the tests.
shard_index: the index of this device.
max_shards: the maximum shard index.
tests: a dict mapping test_name to command.
flaky_tests: a list of flaky test_name.
"""
super(TestRunner, self).__init__(device, None)
self._options = test_options
self._shard_index = shard_index
self._max_shard = max_shard
self._tests = tests
self._flaky_tests = flaky_tests
self._output_dir = None
self._device_battery = battery_utils.BatteryUtils(self.device)
@staticmethod
def _SaveResult(result):
pickled = os.path.join(constants.PERF_OUTPUT_DIR, result['name'])
if os.path.exists(pickled):
with file(pickled, 'r') as f:
previous = pickle.loads(f.read())
result['output'] = previous['output'] + result['output']
with file(pickled, 'w') as f:
f.write(pickle.dumps(result))
def _CheckDeviceAffinity(self, test_name):
"""Returns True if test_name has affinity for this shard."""
affinity = (self._tests['steps'][test_name]['device_affinity'] %
self._max_shard)
if self._shard_index == affinity:
return True
logging.info('Skipping %s on %s (affinity is %s, device is %s)',
test_name, self.device_serial, affinity, self._shard_index)
return False
def _CleanupOutputDirectory(self):
if self._output_dir:
shutil.rmtree(self._output_dir, ignore_errors=True)
self._output_dir = None
def _ReadChartjsonOutput(self):
if not self._output_dir:
return ''
json_output_path = os.path.join(self._output_dir, 'results-chart.json')
try:
with open(json_output_path) as f:
return f.read()
except IOError:
logging.exception('Exception when reading chartjson.')
logging.error('This usually means that telemetry did not run, so it could'
' not generate the file. Please check the device running'
' the test.')
return ''
def _WriteBuildBotJson(self):
"""Write metadata about the buildbot environment to the output dir."""
data = {
'chromium': _GetChromiumRevision(),
'environment': dict(os.environ)}
logging.info('BuildBot environment: %s', data)
with open(os.path.join(self._output_dir, 'buildbot.json'), 'w') as f:
json.dump(data, f, sort_keys=True, indent=2, separators=(',', ': '))
def _ArchiveOutputDir(self):
"""Archive all files in the output dir, and return as compressed bytes."""
with io.BytesIO() as archive:
with zipfile.ZipFile(archive, 'w', zipfile.ZIP_DEFLATED) as contents:
num_files = 0
for absdir, _, files in os.walk(self._output_dir):
reldir = os.path.relpath(absdir, self._output_dir)
for filename in files:
src_path = os.path.join(absdir, filename)
# We use normpath to turn './file.txt' into just 'file.txt'.
dst_path = os.path.normpath(os.path.join(reldir, filename))
contents.write(src_path, dst_path)
num_files += 1
if num_files:
logging.info('%d files in the output dir were archived.', num_files)
else:
logging.warning('No files in the output dir. Archive is empty.')
return archive.getvalue()
def _LaunchPerfTest(self, test_name):
"""Runs a perf test.
Args:
test_name: the name of the test to be executed.
Returns:
A tuple containing (Output, base_test_result.ResultType)
"""
if not self._CheckDeviceAffinity(test_name):
return '', base_test_result.ResultType.PASS
try:
logging.warning('Unmapping device ports')
forwarder.Forwarder.UnmapAllDevicePorts(self.device)
self.device.RestartAdbd()
except Exception as e: # pylint: disable=broad-except
logging.error('Exception when tearing down device %s', e)
test_config = self._tests['steps'][test_name]
cmd = ('%s --device %s' % (test_config['cmd'], self.device_serial))
if (self._options.collect_chartjson_data
or test_config.get('archive_output_dir')):
self._output_dir = tempfile.mkdtemp()
self._WriteBuildBotJson()
cmd = cmd + ' --output-dir=%s' % self._output_dir
logging.info(
'temperature: %s (0.1 C)',
str(self._device_battery.GetBatteryInfo().get('temperature')))
if self._options.max_battery_temp:
self._device_battery.LetBatteryCoolToTemperature(
self._options.max_battery_temp)
logging.info('Charge level: %s%%',
str(self._device_battery.GetBatteryInfo().get('level')))
if self._options.min_battery_level:
self._device_battery.ChargeDeviceToLevel(
self._options.min_battery_level)
self.device.SetScreen(True)
logging.info('%s : %s', test_name, cmd)
start_time = time.time()
timeout = test_config.get('timeout', 3600)
if self._options.no_timeout:
timeout = None
logging.info('Timeout for %s test: %s', test_name, timeout)
full_cmd = cmd
if self._options.dry_run:
full_cmd = 'echo %s' % cmd
logfile = sys.stdout
archive_bytes = None
if self._options.single_step:
# Just print a heart-beat so that the outer buildbot scripts won't timeout
# without response.
logfile = _HeartBeatLogger()
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))
try:
exit_code, output = cmd_helper.GetCmdStatusAndOutputWithTimeout(
full_cmd, timeout, cwd=cwd, shell=True, logfile=logfile)
json_output = self._ReadChartjsonOutput()
if test_config.get('archive_output_dir'):
archive_bytes = self._ArchiveOutputDir()
except cmd_helper.TimeoutError as e:
exit_code = -1
output = e.output
json_output = ''
finally:
self._CleanupOutputDirectory()
if self._options.single_step:
logfile.stop()
end_time = time.time()
if exit_code is None:
exit_code = -1
logging.info('%s : exit_code=%d in %d secs at %s',
test_name, exit_code, end_time - start_time,
self.device_serial)
try:
# Some perf configs run the same benchmark with different options on
# different step names. Here we disambiguate those, so that data is
# uploaded to the perf dashoards based on their step name instead.
chart_data = json.loads(json_output)
if chart_data['benchmark_name'] != test_name:
logging.info('Benchmark %r will be reported as %r in chartjson.',
chart_data['benchmark_name'], test_name)
chart_data['telemetry_benchmark_name'] = chart_data['benchmark_name']
chart_data['benchmark_name'] = test_name
json_output = json.dumps(chart_data, sort_keys=True, indent=2,
separators=(',', ': '))
except StandardError:
logging.exception('Could not read data from chartjson.')
if exit_code == 0:
result_type = base_test_result.ResultType.PASS
else:
result_type = base_test_result.ResultType.FAIL
# Since perf tests use device affinity, give the device a chance to
# recover if it is offline after a failure. Otherwise, the master sharder
# will remove it from the pool and future tests on this device will fail.
try:
self.device.WaitUntilFullyBooted(timeout=120)
except device_errors.CommandTimeoutError as e:
logging.error('Device failed to return after %s: %s', test_name, e)
actual_exit_code = exit_code
if test_name in self._flaky_tests:
# The exit_code is used at the second stage when printing the
# test output. If the test is flaky, force to "0" to get that step green
# whilst still gathering data to the perf dashboards.
# The result_type is used by the test_dispatcher to retry the test.
exit_code = 0
persisted_result = {
'name': test_name,
'output': [output],
'chartjson': json_output,
'archive_bytes': archive_bytes,
'exit_code': exit_code,
'actual_exit_code': actual_exit_code,
'result_type': result_type,
'start_time': start_time,
'end_time': end_time,
'total_time': end_time - start_time,
'device': self.device_serial,
'cmd': cmd,
}
self._SaveResult(persisted_result)
return (output, result_type)
def RunTest(self, test_name):
"""Run a perf test on the device.
Args:
test_name: String to use for logging the test result.
Returns:
A tuple of (TestRunResults, retry).
"""
_, result_type = self._LaunchPerfTest(test_name)
results = base_test_result.TestRunResults()
results.AddResult(base_test_result.BaseTestResult(test_name, result_type))
retry = None
if not results.DidRunPass():
retry = test_name
return results, retry
|