1 # Copyright 2016 The Chromium OS 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. 4 5 # repohooks/pre-upload.py currently does not run pylint. But for developers who 6 # want to check their code manually we disable several harmless pylint warnings 7 # which just distract from more serious remaining issues. 8 # 9 # The instance variable _android_cts is not defined in __init__(). 10 # pylint: disable=attribute-defined-outside-init 11 # 12 # Many short variable names don't follow the naming convention. 13 # pylint: disable=invalid-name 14 15 import logging 16 import os 17 import shutil 18 19 from autotest_lib.client.common_lib import error 20 from autotest_lib.server import utils 21 from autotest_lib.server.cros import tradefed_test 22 23 # likely hang unit the TIMEOUT hits and no RETRY steps will happen. 24 _CTS_MAX_RETRY = {'dev': 5, 'beta': 5, 'stable': 5} 25 # Maximum default time allowed for each individual CTS module. 26 _CTS_TIMEOUT_SECONDS = 3600 27 28 # Public download locations for android cts bundles. 29 _DL_CTS = 'https://dl.google.com/dl/android/cts/' 30 _CTS_URI = { 31 'arm': _DL_CTS + 'android-cts-7.1_r3-linux_x86-arm.zip', 32 'x86': _DL_CTS + 'android-cts-7.1_r3-linux_x86-x86.zip', 33 'media': _DL_CTS + 'android-cts-media-1.2.zip', 34 } 35 36 _SDK_TOOLS_DIR_N = 'gs://chromeos-arc-images/builds/git_nyc-mr1-arc-linux-static_sdk_tools/3544738' 37 _ADB_DIR_N = 'gs://chromeos-arc-images/builds/git_nyc-mr1-arc-linux-cheets_arm-user/3544738' 38 39 40 class cheets_CTS_N(tradefed_test.TradefedTest): 41 """Sets up tradefed to run CTS tests.""" 42 version = 1 43 44 def initialize(self, host=None): 45 super(cheets_CTS_N, self).initialize(host=host, adb_dir=_ADB_DIR_N, 46 sdk_tools_dir=_SDK_TOOLS_DIR_N) 47 48 def setup(self, bundle=None, uri=None): 49 """Download and install a zipfile bundle from Google Storage. 50 51 @param bundle: bundle name, which needs to be key of the _CTS_URI 52 dictionary. Can be 'arm', 'x86' and undefined. 53 @param uri: URI of CTS bundle. Required if |abi| is undefined. 54 """ 55 if bundle in _CTS_URI: 56 self._android_cts = self._install_bundle(_CTS_URI[bundle]) 57 else: 58 self._android_cts = self._install_bundle(uri) 59 60 self._cts_tradefed = os.path.join(self._android_cts, 'android-cts', 61 'tools', 'cts-tradefed') 62 logging.info('CTS-tradefed path: %s', self._cts_tradefed) 63 64 # Load waivers and manual tests so TF doesn't re-run them. 65 self.waivers_and_manual_tests = self._get_expected_failures( 66 'expectations') 67 # Load modules with no tests. 68 self.notest_modules = self._get_expected_failures('notest_modules') 69 70 def _clean_repository(self): 71 """Ensures all old logs, results and plans are deleted. 72 73 This function should be called at the start of each autotest iteration. 74 """ 75 logging.info('Cleaning up repository.') 76 repository = os.path.join(self._android_cts, 'android-cts') 77 for directory in ['logs', 'subplans', 'results']: 78 path = os.path.join(repository, directory) 79 if os.path.exists(path): 80 shutil.rmtree(path) 81 self._safe_makedirs(path) 82 83 def _install_plan(self, plan): 84 logging.info('Install plan: %s', plan) 85 plans_dir = os.path.join(self._android_cts, 'android-cts', 'repository', 86 'plans') 87 src_plan_file = os.path.join(self.bindir, 'plans', '%s.xml' % plan) 88 shutil.copy(src_plan_file, plans_dir) 89 90 def _tradefed_run_command(self, 91 module=None, 92 plan=None, 93 session_id=None, 94 test_class=None, 95 test_method=None): 96 """Builds the CTS tradefed 'run' command line. 97 98 There are five different usages: 99 100 1. Test a module: assign the module name via |module|. 101 2. Test a plan: assign the plan name via |plan|. 102 3. Continue a session: assign the session ID via |session_id|. 103 4. Run all test cases of a class: assign the class name via 104 |test_class|. 105 5. Run a specific test case: assign the class name and method name in 106 |test_class| and |test_method|. 107 108 @param module: the name of test module to be run. 109 @param plan: name of the plan to be run. 110 @param session_id: tradefed session id to continue. 111 @param test_class: the name of the class of which all test cases will 112 be run. 113 @param test_name: the name of the method of |test_class| to be run. 114 Must be used with |test_class|. 115 @return: list of command tokens for the 'run' command. 116 """ 117 if module is not None: 118 # Run a particular module (used to be called package in M). 119 cmd = ['run', 'commandAndExit', 'cts', '--module', module] 120 elif plan is not None and session_id is not None: 121 # In 7.1 R2 we can only retry session_id with the original plan. 122 cmd = ['run', 'commandAndExit', 'cts', '--plan', plan, 123 '--retry', '%d' % session_id] 124 elif plan is not None: 125 # TODO(ihf): This needs testing to support media team. 126 cmd = ['run', 'commandAndExit', 'cts', '--plan', plan] 127 elif test_class is not None: 128 # TODO(ihf): This needs testing to support media team. 129 cmd = ['run', 'commandAndExit', 'cts', '-c', test_class] 130 if test_method is not None: 131 cmd += ['-m', test_method] 132 else: 133 logging.warning('Running all tests. This can take several days.') 134 cmd = ['run', 'commandAndExit', 'cts'] 135 # We handle media download ourselves in the lab, as lazy as possible. 136 cmd.append('--precondition-arg') 137 cmd.append('skip-media-download') 138 # If we are running outside of the lab we can collect more data. 139 if not utils.is_in_container(): 140 logging.info('Running outside of lab, adding extra debug options.') 141 cmd.append('--log-level-display=DEBUG') 142 cmd.append('--screenshot-on-failure') 143 # TODO(ihf): Add log collection once b/28333587 fixed. 144 #cmd.append('--collect-deqp-logs') 145 # TODO(ihf): Add tradefed_test.adb_keepalive() and remove 146 # --disable-reboot. This might be more efficient. 147 # At early stage, cts-tradefed tries to reboot the device by 148 # "adb reboot" command. In a real Android device case, when the 149 # rebooting is completed, adb connection is re-established 150 # automatically, and cts-tradefed expects that behavior. 151 # However, in ARC, it doesn't work, so the whole test process 152 # is just stuck. Here, disable the feature. 153 cmd.append('--disable-reboot') 154 # Create a logcat file for each individual failure. 155 cmd.append('--logcat-on-failure') 156 return cmd 157 158 def _run_cts_tradefed(self, 159 commands, 160 datetime_id=None, 161 collect_results=True): 162 """Runs tradefed, collects logs and returns the result counts. 163 164 Assumes that only last entry of |commands| actually runs tests and has 165 interesting output (results, logs) for collection. Ignores all other 166 commands for this purpose. 167 168 @param commands: List of lists of command tokens. 169 @param datetime_id: For 'continue' datetime of previous run is known. 170 Knowing it makes collecting logs more robust. 171 @param collect: Skip result collection if False. 172 @return: tuple of (tests, pass, fail, notexecuted) counts. 173 """ 174 for command in commands: 175 # Assume only last command actually runs tests and has interesting 176 # output (results, logs) for collection. 177 logging.info('RUN: ./cts-tradefed %s', ' '.join(command)) 178 output = self._run( 179 self._cts_tradefed, 180 args=tuple(command), 181 timeout=self._timeout, 182 verbose=True, 183 ignore_status=False, 184 # Make sure to tee tradefed stdout/stderr to autotest logs 185 # continuously during the test run. 186 stdout_tee=utils.TEE_TO_LOGS, 187 stderr_tee=utils.TEE_TO_LOGS) 188 logging.info('END: ./cts-tradefed %s\n', ' '.join(command)) 189 if not collect_results: 190 return None 191 result_destination = os.path.join(self.resultsdir, 'android-cts') 192 # Gather the global log first. Datetime parsing below can abort the test 193 # if tradefed startup had failed. Even then the global log is useful. 194 self._collect_tradefed_global_log(output, result_destination) 195 if not datetime_id: 196 # Parse stdout to obtain datetime of the session. This is needed to 197 # locate result xml files and logs. 198 datetime_id = self._parse_tradefed_datetime_N(output, self.summary) 199 # Collect tradefed logs for autotest. 200 tradefed = os.path.join(self._android_cts, 'android-cts') 201 self._collect_logs(tradefed, datetime_id, result_destination) 202 return self._parse_result_v2(output, 203 waivers=self.waivers_and_manual_tests) 204 205 def _tradefed_retry(self, test_name, session_id): 206 """Retries failing tests in session. 207 208 It is assumed that there are no notexecuted tests of session_id, 209 otherwise some tests will be missed and never run. 210 211 @param test_name: the name of test to be retried. 212 @param session_id: tradefed session id to retry. 213 @param result_type: [N only] either 'failed' or 'not_executed' 214 @return: tuple of (new session_id, tests, pass, fail, notexecuted). 215 """ 216 # Creating new test plan for retry. 217 derivedplan = 'retry.%s.%s' % (test_name, session_id) 218 logging.info('Retrying failures using derived plan %s.', derivedplan) 219 # The list commands are not required. It allows the reader to inspect 220 # the tradefed state when examining the autotest logs. 221 commands = [ 222 ['add', 'subplan', '--name', derivedplan, 223 '--session', '%d' % session_id, 224 '--result-type', 'failed', '--result-type', 'not_executed'], 225 ['list', 'subplans'], 226 ['list', 'results'], 227 self._tradefed_run_command(plan=derivedplan, 228 session_id=session_id)] 229 # TODO(ihf): Consider if diffing/parsing output of "list results" for 230 # new session_id might be more reliable. For now just assume simple 231 # increment. This works if only one tradefed instance is active and 232 # only a single run command is executing at any moment. 233 return session_id + 1, self._run_cts_tradefed(commands) 234 235 def _get_release_channel(self): 236 """Returns the DUT channel of the image ('dev', 'beta', 'stable').""" 237 # TODO(ihf): check CHROMEOS_RELEASE_DESCRIPTION and return channel. 238 return 'dev' 239 240 def _get_channel_retry(self): 241 """Returns the maximum number of retries for DUT image channel.""" 242 channel = self._get_release_channel() 243 if channel in _CTS_MAX_RETRY: 244 return _CTS_MAX_RETRY[channel] 245 retry = _CTS_MAX_RETRY['dev'] 246 logging.warning('Could not establish channel. Using retry=%d.', retry) 247 return retry 248 249 def _consistent(self, tests, passed, failed, notexecuted): 250 """Verifies that the given counts are plausible. 251 252 Used for finding bad logfile parsing using accounting identities. 253 254 TODO(ihf): change to tests != passed + failed + notexecuted 255 only once b/35530394 fixed.""" 256 return ((tests == passed + failed) or 257 (tests == passed + failed + notexecuted)) 258 259 def run_once(self, 260 target_module=None, 261 target_plan=None, 262 target_class=None, 263 target_method=None, 264 needs_push_media=False, 265 max_retry=None, 266 timeout=_CTS_TIMEOUT_SECONDS): 267 """Runs the specified CTS once, but with several retries. 268 269 There are four usages: 270 1. Test the whole module named |target_module|. 271 2. Test with a plan named |target_plan|. 272 3. Run all the test cases of class named |target_class|. 273 4. Run a specific test method named |target_method| of class 274 |target_class|. 275 276 @param target_module: the name of test module to run. 277 @param target_plan: the name of the test plan to run. 278 @param target_class: the name of the class to be tested. 279 @param target_method: the name of the method to be tested. 280 @param needs_push_media: need to push test media streams. 281 @param max_retry: number of retry steps before reporting results. 282 @param timeout: time after which tradefed can be interrupted. 283 """ 284 # On dev and beta channels timeouts are sharp, lenient on stable. 285 self._timeout = timeout 286 if self._get_release_channel == 'stable': 287 self._timeout += 3600 288 # Retries depend on channel. 289 self._max_retry = max_retry 290 if not self._max_retry: 291 self._max_retry = self._get_channel_retry() 292 logging.info('Maximum number of retry steps %d.', self._max_retry) 293 session_id = 0 294 295 self.result_history = {} 296 steps = -1 # For historic reasons the first iteration is not counted. 297 total_tests = 0 298 total_passed = 0 299 self.summary = '' 300 if target_module is not None: 301 test_name = 'module.%s' % target_module 302 test_command = self._tradefed_run_command( 303 module=target_module, session_id=session_id) 304 elif target_plan is not None: 305 test_name = 'plan.%s' % target_plan 306 test_command = self._tradefed_run_command( 307 plan=target_plan, session_id=session_id) 308 elif target_class is not None: 309 test_name = 'testcase.%s' % target_class 310 if target_method is not None: 311 test_name += '.' + target_method 312 test_command = self._tradefed_run_command( 313 test_class=target_class, 314 test_method=target_method, 315 session_id=session_id) 316 else: 317 test_command = self._tradefed_run_command() 318 test_name = 'all_CTS' 319 320 # Unconditionally run CTS module until we see some tests executed. 321 while total_tests == 0 and steps < self._max_retry: 322 with self._login_chrome(): 323 self._ready_arc() 324 325 # Only push media for tests that need it. b/29371037 326 if needs_push_media: 327 self._push_media(_CTS_URI) 328 # copy_media.sh is not lazy, but we try to be. 329 needs_push_media = False 330 331 # Start each valid iteration with a clean repository. This 332 # allows us to track session_id blindly. 333 self._clean_repository() 334 if target_plan is not None: 335 self._install_plan(target_plan) 336 logging.info('Running %s:', test_name) 337 338 # The list command is not required. It allows the reader to 339 # inspect the tradefed state when examining the autotest logs. 340 commands = [['list', 'results'], test_command] 341 counts = self._run_cts_tradefed(commands) 342 tests, passed, failed, notexecuted, waived = counts 343 self.result_history[steps] = counts 344 msg = 'run(t=%d, p=%d, f=%d, ne=%d, w=%d)' % counts 345 logging.info('RESULT: %s', msg) 346 self.summary += msg 347 if tests == 0 and target_module in self.notest_modules: 348 logging.info('Package has no tests as expected.') 349 return 350 if tests > 0 and target_module in self.notest_modules: 351 # We expected no tests, but the new bundle drop must have 352 # added some for us. Alert us to the situation. 353 raise error.TestFail('Failed: Remove module %s from ' 354 'notest_modules directory!' % 355 target_module) 356 if tests == 0 and target_module not in self.notest_modules: 357 logging.error('Did not find any tests in module. Hoping ' 358 'this is transient. Retry after reboot.') 359 if not self._consistent(tests, passed, failed, notexecuted): 360 # Try to figure out what happened. Example: b/35605415. 361 self._run_cts_tradefed([['list', 'results']], 362 collect_results=False) 363 logging.warning('Test count inconsistent. %s', 364 self.summary) 365 # Keep track of global count, we can't trust continue/retry. 366 if total_tests == 0: 367 total_tests = tests 368 total_passed += passed 369 steps += 1 370 # The DUT has rebooted at this point and is in a clean state. 371 if total_tests == 0: 372 raise error.TestFail('Error: Could not find any tests in module.') 373 374 retry_inconsistency_error = None 375 # If the results were not completed or were failing then continue or 376 # retry them iteratively MAX_RETRY times. 377 while steps < self._max_retry and failed > 0: 378 # TODO(ihf): Use result_history to heuristically stop retries early. 379 if failed > waived: 380 with self._login_chrome(): 381 steps += 1 382 self._ready_arc() 383 logging.info('Retrying failures of %s with session_id %d:', 384 test_name, session_id) 385 expected_tests = failed + notexecuted 386 session_id, counts = self._tradefed_retry(test_name, 387 session_id) 388 tests, passed, failed, notexecuted, waived = counts 389 self.result_history[steps] = counts 390 # Consistency check, did we really run as many as we 391 # thought initially? 392 if expected_tests != tests: 393 msg = ('Retry inconsistency - ' 394 'initially saw %d failed+notexecuted, ran %d tests. ' 395 'passed=%d, failed=%d, notexecuted=%d, waived=%d.' % 396 (expected_tests, tests, passed, failed, notexecuted, 397 waived)) 398 logging.warning(msg) 399 if expected_tests > tests: 400 # See b/36523200#comment8. Due to the existence of 401 # the multiple tests having the same ID, more cases 402 # may be run than previous fail count. As a 403 # workaround, making it an error only when the tests 404 # run were less than expected. 405 # TODO(kinaba): Find a way to handle this dup. 406 retry_inconsistency_error = msg 407 if not self._consistent(tests, passed, failed, notexecuted): 408 logging.warning('Tradefed inconsistency - retrying.') 409 session_id, counts = self._tradefed_retry(test_name, 410 session_id) 411 tests, passed, failed, notexecuted, waived = counts 412 self.result_history[steps] = counts 413 msg = 'retry(t=%d, p=%d, f=%d, ne=%d, w=%d)' % counts 414 logging.info('RESULT: %s', msg) 415 self.summary += ' ' + msg 416 if not self._consistent(tests, passed, failed, notexecuted): 417 logging.warning('Test count inconsistent. %s', 418 self.summary) 419 total_passed += passed 420 # The DUT has rebooted at this point and is in a clean state. 421 422 # Final classification of test results. 423 if total_passed == 0 or failed > waived: 424 raise error.TestFail( 425 'Failed: after %d retries giving up. ' 426 'passed=%d, failed=%d, notexecuted=%d, waived=%d. %s' % 427 (steps, total_passed, failed, notexecuted, waived, 428 self.summary)) 429 if not self._consistent(total_tests, total_passed, failed, notexecuted): 430 raise error.TestFail('Error: Test count inconsistent. %s' % 431 self.summary) 432 if retry_inconsistency_error: 433 raise error.TestFail('Error: %s %s' % (retry_inconsistency_error, 434 self.summary)) 435 if steps > 0: 436 # TODO(ihf): Make this error.TestPass('...') once available. 437 raise error.TestWarn( 438 'Passed: after %d retries passing %d tests, waived=%d. %s' % 439 (steps, total_passed, waived, self.summary)) 440
