1 # Copyright 2013 The Chromium Authors. All rights reserved.
2 # Use of this source code is governed by a BSD-style license that can be
3 # found in the LICENSE file.
5 """Dispatches tests, either sharding or replicating them.
7 Performs the following steps:
8 * Create a test collection factory, using the given tests
9 - If sharding: test collection factory returns the same shared test collection
11 - If replciating: test collection factory returns a unique test collection to
12 each test runner, with the same set of tests in each.
13 * Create a test runner for each device.
14 * Run each test runner in its own thread, grabbing tests from the test
15 collection until there are no tests left.
18 # TODO(jbudorick) Deprecate and remove this class after any relevant parts have
19 # been ported to the new environment / test instance model.
24 from devil
.android
import device_errors
25 from devil
.utils
import reraiser_thread
26 from devil
.utils
import watchdog_timer
27 from pylib
import constants
28 from pylib
.base
import base_test_result
29 from pylib
.base
import test_collection
32 DEFAULT_TIMEOUT
= 7 * 60 # seven minutes
35 class _ThreadSafeCounter(object):
36 """A threadsafe counter."""
39 self
._lock
= threading
.Lock()
42 def GetAndIncrement(self
):
43 """Get the current value and increment it atomically.
46 The value before incrementing.
49 pre_increment
= self
._value
55 """Holds a test with additional metadata."""
57 def __init__(self
, test
, tries
=0):
58 """Initializes the _Test object.
62 tries: Number of tries so far.
68 def _RunTestsFromQueue(runner
, collection
, out_results
, watcher
,
69 num_retries
, tag_results_with_device
=False):
70 """Runs tests from the collection until empty using the given runner.
72 Adds TestRunResults objects to the out_results list and may add tests to the
76 runner: A TestRunner object used to run the tests.
77 collection: A TestCollection from which to get _Test objects to run.
78 out_results: A list to add TestRunResults to.
79 watcher: A watchdog_timer.WatchdogTimer object, used as a shared timeout.
80 num_retries: Number of retries for a test.
81 tag_results_with_device: If True, appends the name of the device on which
82 the test was run to the test name. Used when replicating to identify
83 which device ran each copy of the test, and to ensure each copy of the
84 test is recorded separately.
87 def TagTestRunResults(test_run_results
):
88 """Tags all results with the last 4 digits of the device id.
90 Used when replicating tests to distinguish the same tests run on different
91 devices. We use a set to store test results, so the hash (generated from
92 name and tag) must be unique to be considered different results.
94 new_test_run_results
= base_test_result
.TestRunResults()
95 for test_result
in test_run_results
.GetAll():
96 test_result
.SetName('%s_%s' % (runner
.device_serial
[-4:],
97 test_result
.GetName()))
98 new_test_run_results
.AddResult(test_result
)
99 return new_test_run_results
101 for test
in collection
:
104 if not runner
.device
.IsOnline():
105 # Device is unresponsive, stop handling tests on this device.
106 msg
= 'Device %s is unresponsive.' % runner
.device_serial
108 raise device_errors
.DeviceUnreachableError(msg
)
109 result
, retry
= runner
.RunTest(test
.test
)
110 if tag_results_with_device
:
111 result
= TagTestRunResults(result
)
113 if retry
and test
.tries
<= num_retries
:
114 # Retry non-passing results, only record passing results.
115 pass_results
= base_test_result
.TestRunResults()
116 pass_results
.AddResults(result
.GetPass())
117 out_results
.append(pass_results
)
118 logging
.warning('Will retry test %s, try #%s.', retry
, test
.tries
)
119 collection
.add(_Test(test
=retry
, tries
=test
.tries
))
121 # All tests passed or retry limit reached. Either way, record results.
122 out_results
.append(result
)
124 # An unhandleable exception, ensure tests get run by another device and
125 # reraise this exception on the main thread.
129 # Retries count as separate tasks so always mark the popped test as done.
130 collection
.test_completed()
133 def _SetUp(runner_factory
, device
, out_runners
, threadsafe_counter
):
134 """Creates a test runner for each device and calls SetUp() in parallel.
136 Note: if a device is unresponsive the corresponding TestRunner will not be
137 added to out_runners.
140 runner_factory: Callable that takes a device and index and returns a
142 device: The device serial number to set up.
143 out_runners: List to add the successfully set up TestRunner object.
144 threadsafe_counter: A _ThreadSafeCounter object used to get shard indices.
147 index
= threadsafe_counter
.GetAndIncrement()
148 logging
.warning('Creating shard %s for device %s.', index
, device
)
149 runner
= runner_factory(device
, index
)
151 out_runners
.append(runner
)
152 except device_errors
.DeviceUnreachableError
as e
:
153 logging
.warning('Failed to create shard for %s: [%s]', device
, e
)
156 def _RunAllTests(runners
, test_collection_factory
, num_retries
, timeout
=None,
157 tag_results_with_device
=False):
158 """Run all tests using the given TestRunners.
161 runners: A list of TestRunner objects.
162 test_collection_factory: A callable to generate a TestCollection object for
164 num_retries: Number of retries for a test.
165 timeout: Watchdog timeout in seconds.
166 tag_results_with_device: If True, appends the name of the device on which
167 the test was run to the test name. Used when replicating to identify
168 which device ran each copy of the test, and to ensure each copy of the
169 test is recorded separately.
172 A tuple of (TestRunResults object, exit code)
174 logging
.warning('Running tests with %s test runners.', len(runners
))
177 run_results
= base_test_result
.TestRunResults()
178 watcher
= watchdog_timer
.WatchdogTimer(timeout
)
179 test_collections
= [test_collection_factory() for _
in runners
]
182 reraiser_thread
.ReraiserThread(
184 [r
, tc
, results
, watcher
, num_retries
, tag_results_with_device
],
185 name
=r
.device_serial
[-4:])
186 for r
, tc
in zip(runners
, test_collections
)]
188 workers
= reraiser_thread
.ReraiserThreadGroup(threads
)
192 workers
.JoinAll(watcher
)
193 except device_errors
.CommandFailedError
:
194 logging
.exception('Command failed on device.')
195 except device_errors
.CommandFailedError
:
196 logging
.exception('Command timed out on device.')
197 except device_errors
.DeviceUnreachableError
:
198 logging
.exception('Device became unreachable.')
200 if not all((len(tc
) == 0 for tc
in test_collections
)):
201 logging
.error('Only ran %d tests (all devices are likely offline).',
203 for tc
in test_collections
:
204 run_results
.AddResults(base_test_result
.BaseTestResult(
205 t
, base_test_result
.ResultType
.UNKNOWN
) for t
in tc
.test_names())
208 run_results
.AddTestRunResults(r
)
209 if not run_results
.DidRunPass():
210 exit_code
= constants
.ERROR_EXIT_CODE
211 return (run_results
, exit_code
)
214 def _CreateRunners(runner_factory
, devices
, timeout
=None):
215 """Creates a test runner for each device and calls SetUp() in parallel.
217 Note: if a device is unresponsive the corresponding TestRunner will not be
218 included in the returned list.
221 runner_factory: Callable that takes a device and index and returns a
223 devices: List of device serial numbers as strings.
224 timeout: Watchdog timeout in seconds, defaults to the default timeout.
227 A list of TestRunner objects.
229 logging
.warning('Creating %s test runners.', len(devices
))
231 counter
= _ThreadSafeCounter()
232 threads
= reraiser_thread
.ReraiserThreadGroup(
233 [reraiser_thread
.ReraiserThread(_SetUp
,
234 [runner_factory
, d
, runners
, counter
],
238 threads
.JoinAll(watchdog_timer
.WatchdogTimer(timeout
))
242 def _TearDownRunners(runners
, timeout
=None):
243 """Calls TearDown() for each test runner in parallel.
246 runners: A list of TestRunner objects.
247 timeout: Watchdog timeout in seconds, defaults to the default timeout.
249 threads
= reraiser_thread
.ReraiserThreadGroup(
250 [reraiser_thread
.ReraiserThread(r
.TearDown
, name
=r
.device_serial
[-4:])
253 threads
.JoinAll(watchdog_timer
.WatchdogTimer(timeout
))
256 def ApplyMaxPerRun(tests
, max_per_run
):
257 """Rearrange the tests so that no group contains more than max_per_run tests.
264 A list of tests with no more than max_per_run per run.
267 for test_group
in tests
:
268 if type(test_group
) != str:
269 # Do not split test objects which are not strings.
270 tests_expanded
.append(test_group
)
272 test_split
= test_group
.split(':')
273 for i
in range(0, len(test_split
), max_per_run
):
274 tests_expanded
.append(':'.join(test_split
[i
:i
+max_per_run
]))
275 return tests_expanded
278 def RunTests(tests
, runner_factory
, devices
, shard
=True,
279 test_timeout
=DEFAULT_TIMEOUT
, setup_timeout
=DEFAULT_TIMEOUT
,
280 num_retries
=2, max_per_run
=256):
281 """Run all tests on attached devices, retrying tests that don't pass.
284 tests: List of tests to run.
285 runner_factory: Callable that takes a device and index and returns a
287 devices: List of attached devices.
288 shard: True if we should shard, False if we should replicate tests.
289 - Sharding tests will distribute tests across all test runners through a
290 shared test collection.
291 - Replicating tests will copy all tests to each test runner through a
292 unique test collection for each test runner.
293 test_timeout: Watchdog timeout in seconds for running tests.
294 setup_timeout: Watchdog timeout in seconds for creating and cleaning up
296 num_retries: Number of retries for a test.
297 max_per_run: Maximum number of tests to run in any group.
300 A tuple of (base_test_result.TestRunResults object, exit code).
303 logging
.critical('No tests to run.')
304 return (base_test_result
.TestRunResults(), constants
.ERROR_EXIT_CODE
)
306 tests_expanded
= ApplyMaxPerRun(tests
, max_per_run
)
308 # Generate a shared TestCollection object for all test runners, so they
309 # draw from a common pool of tests.
310 shared_test_collection
= test_collection
.TestCollection(
311 [_Test(t
) for t
in tests_expanded
])
312 test_collection_factory
= lambda: shared_test_collection
313 tag_results_with_device
= False
314 log_string
= 'sharded across devices'
316 # Generate a unique TestCollection object for each test runner, but use
317 # the same set of tests.
318 test_collection_factory
= lambda: test_collection
.TestCollection(
319 [_Test(t
) for t
in tests_expanded
])
320 tag_results_with_device
= True
321 log_string
= 'replicated on each device'
323 logging
.info('Will run %d tests (%s): %s',
324 len(tests_expanded
), log_string
, str(tests_expanded
))
325 runners
= _CreateRunners(runner_factory
, devices
, setup_timeout
)
327 return _RunAllTests(runners
, test_collection_factory
,
328 num_retries
, test_timeout
, tag_results_with_device
)
331 _TearDownRunners(runners
, setup_timeout
)
332 except device_errors
.DeviceUnreachableError
as e
:
333 logging
.warning('Device unresponsive during TearDown: [%s]', e
)
334 except Exception: # pylint: disable=broad-except
335 logging
.exception('Unexpected exception caught during TearDown')