Fang Deng | 5d518f4 | 2013-08-02 14:04:32 -0700 | [diff] [blame] | 1 | # Copyright (c) 2013 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 | # Expects to be run in an environment with sudo and no interactive password |
| 6 | # prompt, such as within the Chromium OS development chroot. |
| 7 | |
| 8 | |
| 9 | """This file provides core logic for servo verify/repair process.""" |
| 10 | |
| 11 | |
Fang Deng | 5d518f4 | 2013-08-02 14:04:32 -0700 | [diff] [blame] | 12 | import logging |
Raul E Rangel | 52ca2e8 | 2018-07-03 14:10:14 -0600 | [diff] [blame] | 13 | import os |
Ruben Rodriguez Buchillon | 93084d0 | 2020-01-21 15:17:36 -0800 | [diff] [blame] | 14 | import re |
| 15 | import tarfile |
Garry Wang | c1288cf | 2019-12-17 14:58:00 -0800 | [diff] [blame] | 16 | import time |
Gregory Nisbet | 265a52c | 2019-12-10 20:38:42 -0800 | [diff] [blame] | 17 | import traceback |
Dana Goyette | 4dc0adc | 2019-05-06 14:51:53 -0700 | [diff] [blame] | 18 | import xmlrpclib |
Fang Deng | 5d518f4 | 2013-08-02 14:04:32 -0700 | [diff] [blame] | 19 | |
| 20 | from autotest_lib.client.bin import utils |
Garry Wang | 79e9af6 | 2019-06-12 15:19:19 -0700 | [diff] [blame] | 21 | from autotest_lib.client.common_lib import error |
Richard Barnette | 9a26ad6 | 2016-06-10 12:03:08 -0700 | [diff] [blame] | 22 | from autotest_lib.client.common_lib import hosts |
Fang Deng | 5d518f4 | 2013-08-02 14:04:32 -0700 | [diff] [blame] | 23 | from autotest_lib.client.common_lib.cros import retry |
Christopher Wiley | cef1f90 | 2014-06-19 11:11:23 -0700 | [diff] [blame] | 24 | from autotest_lib.client.common_lib.cros.network import ping_runner |
Richard Barnette | 9a26ad6 | 2016-06-10 12:03:08 -0700 | [diff] [blame] | 25 | from autotest_lib.server.cros.servo import servo |
Richard Barnette | d31580e | 2018-05-14 19:58:00 +0000 | [diff] [blame] | 26 | from autotest_lib.server.hosts import servo_repair |
Garry Wang | ebc015b | 2019-06-06 17:45:06 -0700 | [diff] [blame] | 27 | from autotest_lib.server.hosts import base_servohost |
Garry Wang | 11b5e87 | 2020-03-11 15:14:08 -0700 | [diff] [blame^] | 28 | from autotest_lib.server.hosts import servo_constants |
| 29 | from autotest_lib.client.common_lib import global_config |
Garry Wang | d736748 | 2020-02-27 13:52:40 -0800 | [diff] [blame] | 30 | |
Dan Shi | 3b2adf6 | 2015-09-02 17:46:54 -0700 | [diff] [blame] | 31 | _CONFIG = global_config.global_config |
Fang Deng | 5d518f4 | 2013-08-02 14:04:32 -0700 | [diff] [blame] | 32 | |
Otabek Kasimov | a7ba91a | 2020-03-09 08:31:01 -0700 | [diff] [blame] | 33 | |
Garry Wang | ebc015b | 2019-06-06 17:45:06 -0700 | [diff] [blame] | 34 | class ServoHost(base_servohost.BaseServoHost): |
| 35 | """Host class for a servo host(e.g. beaglebone, labstation) |
Dana Goyette | 0b6e640 | 2019-10-04 11:09:24 -0700 | [diff] [blame] | 36 | that with a servo instance for a specific port. |
| 37 | |
| 38 | @type _servo: servo.Servo | None |
| 39 | """ |
Fang Deng | 5d518f4 | 2013-08-02 14:04:32 -0700 | [diff] [blame] | 40 | |
Raul E Rangel | 52ca2e8 | 2018-07-03 14:10:14 -0600 | [diff] [blame] | 41 | DEFAULT_PORT = int(os.getenv('SERVOD_PORT', '9999')) |
Richard Barnette | 9a26ad6 | 2016-06-10 12:03:08 -0700 | [diff] [blame] | 42 | |
Dan Shi | e5b3c51 | 2014-08-21 12:12:09 -0700 | [diff] [blame] | 43 | # Timeout for initializing servo signals. |
Wai-Hong Tam | 37b6ed3 | 2017-09-19 15:52:39 -0700 | [diff] [blame] | 44 | INITIALIZE_SERVO_TIMEOUT_SECS = 60 |
Richard Barnette | 9a26ad6 | 2016-06-10 12:03:08 -0700 | [diff] [blame] | 45 | |
xixuan | 6cf6d2f | 2016-01-29 15:29:00 -0800 | [diff] [blame] | 46 | # Ready test function |
| 47 | SERVO_READY_METHOD = 'get_version' |
Fang Deng | 5d518f4 | 2013-08-02 14:04:32 -0700 | [diff] [blame] | 48 | |
Ruben Rodriguez Buchillon | 93084d0 | 2020-01-21 15:17:36 -0800 | [diff] [blame] | 49 | # Directory prefix on the servo host where the servod logs are stored. |
| 50 | SERVOD_LOG_PREFIX = '/var/log/servod' |
| 51 | |
| 52 | # Exit code to use when symlinks for servod logs are not found. |
| 53 | NO_SYMLINKS_CODE = 9 |
| 54 | |
| 55 | # Directory in the job's results directory to dump the logs into. |
| 56 | LOG_DIR = 'servod' |
| 57 | |
| 58 | # Prefix for joint loglevel files in the logs. |
| 59 | JOINT_LOG_PREFIX = 'log' |
| 60 | |
| 61 | # Regex group to extract timestamp from logfile name. |
| 62 | TS_GROUP = 'ts' |
| 63 | |
| 64 | # This regex is used to extract the timestamp from servod logs. |
| 65 | # files always start with log. |
| 66 | TS_RE = (r'log.' |
| 67 | # The timestamp is of format %Y-%m-%d--%H-%M-%S.MS |
| 68 | r'(?P<%s>\d{4}(\-\d{2}){2}\-(-\d{2}){3}.\d{3})' |
| 69 | # The loglevel is optional depending on labstation version. |
| 70 | r'(.(INFO|DEBUG|WARNING))?' % TS_GROUP) |
| 71 | TS_EXTRACTOR = re.compile(TS_RE) |
| 72 | |
| 73 | # Regex group to extract MCU name from logline in servod logs. |
| 74 | MCU_GROUP = 'mcu' |
| 75 | |
| 76 | # Regex group to extract logline from MCU logline in servod logs. |
| 77 | LINE_GROUP = 'line' |
| 78 | |
| 79 | # This regex is used to extract the mcu and the line content from an |
| 80 | # MCU logline in servod logs. e.g. EC or servo_v4 console logs. |
| 81 | # Here is an example log-line: |
| 82 | # |
| 83 | # 2020-01-23 13:15:12,223 - servo_v4 - EC3PO.Console - DEBUG - |
| 84 | # console.py:219:LogConsoleOutput - /dev/pts/9 - cc polarity: cc1 |
| 85 | # |
| 86 | # Here is conceptually how they are formatted: |
| 87 | # |
| 88 | # <time> - <MCU> - EC3PO.Console - <LVL> - <file:line:func> - <pts> - |
| 89 | # <output> |
| 90 | # |
| 91 | # The log format starts with a timestamp |
| 92 | MCU_RE = (r'[\d\-]+ [\d:,]+ ' |
| 93 | # The mcu that is logging this is next. |
| 94 | r'- (?P<%s>\w+) - ' |
| 95 | # Next, we have more log outputs before the actual line. |
| 96 | # Information about the file line, logging function etc. |
| 97 | # Anchor on EC3PO Console, LogConsoleOutput and dev/pts. |
| 98 | # NOTE: if the log format changes, this regex needs to be |
| 99 | # adjusted. |
| 100 | r'EC3PO\.Console[\s\-\w\d:.]+LogConsoleOutput - /dev/pts/\d+ - ' |
| 101 | # Lastly, we get the MCU's console line. |
| 102 | r'(?P<%s>.+$)' % (MCU_GROUP, LINE_GROUP)) |
| 103 | MCU_EXTRACTOR = re.compile(MCU_RE) |
| 104 | |
| 105 | # Suffix to identify compressed logfiles. |
| 106 | COMPRESSION_SUFFIX = '.tbz2' |
| 107 | |
Otabek Kasimov | cc9738e | 2020-02-14 16:17:15 -0800 | [diff] [blame] | 108 | def _init_attributes(self): |
| 109 | self._servo_state = None |
| 110 | self.servo_port = None |
| 111 | self.servo_board = None |
| 112 | self.servo_model = None |
| 113 | self.servo_serial = None |
| 114 | self._servo = None |
| 115 | self._servod_server_proxy = None |
Ruben Rodriguez Buchillon | 93084d0 | 2020-01-21 15:17:36 -0800 | [diff] [blame] | 116 | # Flag to make sure that multiple calls to close do not result in the |
| 117 | # logic executing multiple times. |
| 118 | self._closed = False |
Fang Deng | 5d518f4 | 2013-08-02 14:04:32 -0700 | [diff] [blame] | 119 | |
Richard Barnette | 17bfc6c | 2016-08-04 18:41:43 -0700 | [diff] [blame] | 120 | def _initialize(self, servo_host='localhost', |
Richard Barnette | e519dcd | 2016-08-15 17:37:17 -0700 | [diff] [blame] | 121 | servo_port=DEFAULT_PORT, servo_board=None, |
Nick Sanders | 2f3c985 | 2018-10-24 12:10:24 -0700 | [diff] [blame] | 122 | servo_model=None, servo_serial=None, is_in_lab=None, |
| 123 | *args, **dargs): |
Fang Deng | 5d518f4 | 2013-08-02 14:04:32 -0700 | [diff] [blame] | 124 | """Initialize a ServoHost instance. |
| 125 | |
| 126 | A ServoHost instance represents a host that controls a servo. |
| 127 | |
| 128 | @param servo_host: Name of the host where the servod process |
| 129 | is running. |
Raul E Rangel | 52ca2e8 | 2018-07-03 14:10:14 -0600 | [diff] [blame] | 130 | @param servo_port: Port the servod process is listening on. Defaults |
| 131 | to the SERVOD_PORT environment variable if set, |
| 132 | otherwise 9999. |
Kevin Cheng | 5f2ba6c | 2016-09-28 10:20:05 -0700 | [diff] [blame] | 133 | @param servo_board: Board that the servo is connected to. |
Nick Sanders | 2f3c985 | 2018-10-24 12:10:24 -0700 | [diff] [blame] | 134 | @param servo_model: Model that the servo is connected to. |
Dan Shi | 4d47852 | 2014-02-14 13:46:32 -0800 | [diff] [blame] | 135 | @param is_in_lab: True if the servo host is in Cros Lab. Default is set |
| 136 | to None, for which utils.host_is_in_lab_zone will be |
| 137 | called to check if the servo host is in Cros lab. |
Fang Deng | 5d518f4 | 2013-08-02 14:04:32 -0700 | [diff] [blame] | 138 | |
| 139 | """ |
| 140 | super(ServoHost, self)._initialize(hostname=servo_host, |
Garry Wang | ebc015b | 2019-06-06 17:45:06 -0700 | [diff] [blame] | 141 | is_in_lab=is_in_lab, *args, **dargs) |
Otabek Kasimov | cc9738e | 2020-02-14 16:17:15 -0800 | [diff] [blame] | 142 | self._init_attributes() |
Richard Barnette | 42f4db9 | 2018-08-23 15:05:15 -0700 | [diff] [blame] | 143 | self.servo_port = int(servo_port) |
Richard Barnette | e519dcd | 2016-08-15 17:37:17 -0700 | [diff] [blame] | 144 | self.servo_board = servo_board |
Nick Sanders | 2f3c985 | 2018-10-24 12:10:24 -0700 | [diff] [blame] | 145 | self.servo_model = servo_model |
Kevin Cheng | 643ce8a | 2016-09-15 15:42:12 -0700 | [diff] [blame] | 146 | self.servo_serial = servo_serial |
Wai-Hong Tam | 3a8a255 | 2019-11-19 14:28:04 +0800 | [diff] [blame] | 147 | |
Ruben Rodriguez Buchillon | 93084d0 | 2020-01-21 15:17:36 -0800 | [diff] [blame] | 148 | # The location of the log files on the servo host for this instance. |
| 149 | self.remote_log_dir = '%s_%s' % (self.SERVOD_LOG_PREFIX, |
| 150 | self.servo_port) |
Garry Wang | 79e9af6 | 2019-06-12 15:19:19 -0700 | [diff] [blame] | 151 | # Path of the servo host lock file. |
| 152 | self._lock_file = (self.TEMP_FILE_DIR + str(self.servo_port) |
| 153 | + self.LOCK_FILE_POSTFIX) |
| 154 | # File path to declare a reboot request. |
| 155 | self._reboot_file = (self.TEMP_FILE_DIR + str(self.servo_port) |
| 156 | + self.REBOOT_FILE_POSTFIX) |
| 157 | |
| 158 | # Lock the servo host if it's an in-lab labstation to prevent other |
| 159 | # task to reboot it until current task completes. We also wait and |
| 160 | # make sure the labstation is up here, in the case of the labstation is |
| 161 | # in the middle of reboot. |
Garry Wang | 7c00b0f | 2019-06-25 17:28:17 -0700 | [diff] [blame] | 162 | self._is_locked = False |
Garry Wang | 42b4d86 | 2019-06-25 15:50:49 -0700 | [diff] [blame] | 163 | if (self.wait_up(self.REBOOT_TIMEOUT) and self.is_in_lab() |
| 164 | and self.is_labstation()): |
Garry Wang | 79e9af6 | 2019-06-12 15:19:19 -0700 | [diff] [blame] | 165 | self._lock() |
Garry Wang | ebc015b | 2019-06-06 17:45:06 -0700 | [diff] [blame] | 166 | |
Richard Barnette | 9a26ad6 | 2016-06-10 12:03:08 -0700 | [diff] [blame] | 167 | self._repair_strategy = ( |
| 168 | servo_repair.create_servo_repair_strategy()) |
Richard Barnette | e519dcd | 2016-08-15 17:37:17 -0700 | [diff] [blame] | 169 | |
Richard Barnette | 9a26ad6 | 2016-06-10 12:03:08 -0700 | [diff] [blame] | 170 | def connect_servo(self): |
Kevin Cheng | 5f2ba6c | 2016-09-28 10:20:05 -0700 | [diff] [blame] | 171 | """Establish a connection to the servod server on this host. |
Richard Barnette | 9a26ad6 | 2016-06-10 12:03:08 -0700 | [diff] [blame] | 172 | |
| 173 | Initializes `self._servo` and then verifies that all network |
| 174 | connections are working. This will create an ssh tunnel if |
| 175 | it's required. |
| 176 | |
| 177 | As a side effect of testing the connection, all signals on the |
| 178 | target servo are reset to default values, and the USB stick is |
| 179 | set to the neutral (off) position. |
| 180 | """ |
Kevin Cheng | 643ce8a | 2016-09-15 15:42:12 -0700 | [diff] [blame] | 181 | servo_obj = servo.Servo(servo_host=self, servo_serial=self.servo_serial) |
Kuang-che Wu | 05763f5 | 2019-08-30 16:48:21 +0800 | [diff] [blame] | 182 | self._servo = servo_obj |
Richard Barnette | 9a26ad6 | 2016-06-10 12:03:08 -0700 | [diff] [blame] | 183 | timeout, _ = retry.timeout( |
| 184 | servo_obj.initialize_dut, |
| 185 | timeout_sec=self.INITIALIZE_SERVO_TIMEOUT_SECS) |
| 186 | if timeout: |
| 187 | raise hosts.AutoservVerifyError( |
| 188 | 'Servo initialize timed out.') |
Richard Barnette | 9a26ad6 | 2016-06-10 12:03:08 -0700 | [diff] [blame] | 189 | |
| 190 | |
| 191 | def disconnect_servo(self): |
Kevin Cheng | 5f2ba6c | 2016-09-28 10:20:05 -0700 | [diff] [blame] | 192 | """Disconnect our servo if it exists. |
Richard Barnette | 9a26ad6 | 2016-06-10 12:03:08 -0700 | [diff] [blame] | 193 | |
| 194 | If we've previously successfully connected to our servo, |
| 195 | disconnect any established ssh tunnel, and set `self._servo` |
| 196 | back to `None`. |
| 197 | """ |
| 198 | if self._servo: |
| 199 | # N.B. This call is safe even without a tunnel: |
| 200 | # rpc_server_tracker.disconnect() silently ignores |
| 201 | # unknown ports. |
| 202 | self.rpc_server_tracker.disconnect(self.servo_port) |
| 203 | self._servo = None |
Fang Deng | 5d518f4 | 2013-08-02 14:04:32 -0700 | [diff] [blame] | 204 | |
Garry Wang | c1288cf | 2019-12-17 14:58:00 -0800 | [diff] [blame] | 205 | |
Wai-Hong Tam | 3a8a255 | 2019-11-19 14:28:04 +0800 | [diff] [blame] | 206 | def _create_servod_server_proxy(self): |
| 207 | """Create a proxy that can be used to communicate with servod server. |
Fang Deng | 5d518f4 | 2013-08-02 14:04:32 -0700 | [diff] [blame] | 208 | |
| 209 | @returns: An xmlrpclib.ServerProxy that is connected to the servod |
| 210 | server on the host. |
Fang Deng | 5d518f4 | 2013-08-02 14:04:32 -0700 | [diff] [blame] | 211 | """ |
Garry Wang | 11b5e87 | 2020-03-11 15:14:08 -0700 | [diff] [blame^] | 212 | if (servo_constants.ENABLE_SSH_TUNNEL_FOR_SERVO |
| 213 | and not self.is_localhost()): |
Richard Barnette | 9a26ad6 | 2016-06-10 12:03:08 -0700 | [diff] [blame] | 214 | return self.rpc_server_tracker.xmlrpc_connect( |
| 215 | None, self.servo_port, |
| 216 | ready_test_name=self.SERVO_READY_METHOD, |
Allen Li | 2b1a899 | 2018-11-27 14:17:18 -0800 | [diff] [blame] | 217 | timeout_seconds=60, |
Allen Li | 556f453 | 2018-12-03 18:11:23 -0800 | [diff] [blame] | 218 | request_timeout_seconds=3600) |
Richard Barnette | 9a26ad6 | 2016-06-10 12:03:08 -0700 | [diff] [blame] | 219 | else: |
| 220 | remote = 'http://%s:%s' % (self.hostname, self.servo_port) |
| 221 | return xmlrpclib.ServerProxy(remote) |
Fang Deng | 5d518f4 | 2013-08-02 14:04:32 -0700 | [diff] [blame] | 222 | |
| 223 | |
Wai-Hong Tam | 3a8a255 | 2019-11-19 14:28:04 +0800 | [diff] [blame] | 224 | def get_servod_server_proxy(self): |
| 225 | """Return a cached proxy if exists; otherwise, create a new one. |
| 226 | |
| 227 | @returns: An xmlrpclib.ServerProxy that is connected to the servod |
| 228 | server on the host. |
| 229 | """ |
| 230 | # Single-threaded execution, no race |
| 231 | if self._servod_server_proxy is None: |
| 232 | self._servod_server_proxy = self._create_servod_server_proxy() |
| 233 | return self._servod_server_proxy |
| 234 | |
| 235 | |
Richard Barnette | 1edbb16 | 2016-11-01 11:47:50 -0700 | [diff] [blame] | 236 | def verify(self, silent=False): |
| 237 | """Update the servo host and verify it's in a good state. |
| 238 | |
| 239 | @param silent If true, suppress logging in `status.log`. |
| 240 | """ |
Richard Barnette | abbdc25 | 2018-07-26 16:57:42 -0700 | [diff] [blame] | 241 | message = 'Beginning verify for servo host %s port %s serial %s' |
| 242 | message %= (self.hostname, self.servo_port, self.servo_serial) |
| 243 | self.record('INFO', None, None, message) |
Richard Barnette | 9a26ad6 | 2016-06-10 12:03:08 -0700 | [diff] [blame] | 244 | try: |
Richard Barnette | 1edbb16 | 2016-11-01 11:47:50 -0700 | [diff] [blame] | 245 | self._repair_strategy.verify(self, silent) |
Garry Wang | 11b5e87 | 2020-03-11 15:14:08 -0700 | [diff] [blame^] | 246 | self._servo_state = servo_constants.SERVO_STATE_WORKING |
| 247 | self.record('INFO', None, None, |
| 248 | 'ServoHost verify set servo_state as WORKING') |
Richard Barnette | 9a26ad6 | 2016-06-10 12:03:08 -0700 | [diff] [blame] | 249 | except: |
Garry Wang | 11b5e87 | 2020-03-11 15:14:08 -0700 | [diff] [blame^] | 250 | self._servo_state = servo_constants.SERVO_STATE_BROKEN |
| 251 | self.record('INFO', None, None, |
| 252 | 'ServoHost verify set servo_state as BROKEN') |
Richard Barnette | 9a26ad6 | 2016-06-10 12:03:08 -0700 | [diff] [blame] | 253 | self.disconnect_servo() |
Garry Wang | c1288cf | 2019-12-17 14:58:00 -0800 | [diff] [blame] | 254 | self.stop_servod() |
Richard Barnette | 9a26ad6 | 2016-06-10 12:03:08 -0700 | [diff] [blame] | 255 | raise |
Fang Deng | 5d518f4 | 2013-08-02 14:04:32 -0700 | [diff] [blame] | 256 | |
| 257 | |
Richard Barnette | 1edbb16 | 2016-11-01 11:47:50 -0700 | [diff] [blame] | 258 | def repair(self, silent=False): |
| 259 | """Attempt to repair servo host. |
| 260 | |
| 261 | @param silent If true, suppress logging in `status.log`. |
| 262 | """ |
Richard Barnette | abbdc25 | 2018-07-26 16:57:42 -0700 | [diff] [blame] | 263 | message = 'Beginning repair for servo host %s port %s serial %s' |
| 264 | message %= (self.hostname, self.servo_port, self.servo_serial) |
| 265 | self.record('INFO', None, None, message) |
Richard Barnette | 9a26ad6 | 2016-06-10 12:03:08 -0700 | [diff] [blame] | 266 | try: |
Richard Barnette | 1edbb16 | 2016-11-01 11:47:50 -0700 | [diff] [blame] | 267 | self._repair_strategy.repair(self, silent) |
Garry Wang | 11b5e87 | 2020-03-11 15:14:08 -0700 | [diff] [blame^] | 268 | self._servo_state = servo_constants.SERVO_STATE_WORKING |
| 269 | self.record('INFO', None, None, |
| 270 | 'ServoHost repair set servo_state as WORKING') |
Garry Wang | 464ff1e | 2019-07-18 17:20:34 -0700 | [diff] [blame] | 271 | # If target is a labstation then try to withdraw any existing |
| 272 | # reboot request created by this servo because it passed repair. |
| 273 | if self.is_labstation(): |
| 274 | self.withdraw_reboot_request() |
Richard Barnette | 9a26ad6 | 2016-06-10 12:03:08 -0700 | [diff] [blame] | 275 | except: |
Garry Wang | 11b5e87 | 2020-03-11 15:14:08 -0700 | [diff] [blame^] | 276 | self._servo_state = servo_constants.SERVO_STATE_BROKEN |
| 277 | self.record('INFO', None, None, |
| 278 | 'ServoHost repair set servo_state as BROKEN') |
Richard Barnette | 9a26ad6 | 2016-06-10 12:03:08 -0700 | [diff] [blame] | 279 | self.disconnect_servo() |
Garry Wang | c1288cf | 2019-12-17 14:58:00 -0800 | [diff] [blame] | 280 | self.stop_servod() |
Richard Barnette | 9a26ad6 | 2016-06-10 12:03:08 -0700 | [diff] [blame] | 281 | raise |
Fang Deng | 5d518f4 | 2013-08-02 14:04:32 -0700 | [diff] [blame] | 282 | |
| 283 | |
Dan Shi | 4d47852 | 2014-02-14 13:46:32 -0800 | [diff] [blame] | 284 | def get_servo(self): |
| 285 | """Get the cached servo.Servo object. |
Fang Deng | 5d518f4 | 2013-08-02 14:04:32 -0700 | [diff] [blame] | 286 | |
Dan Shi | 4d47852 | 2014-02-14 13:46:32 -0800 | [diff] [blame] | 287 | @return: a servo.Servo object. |
Dana Goyette | 353d1d9 | 2019-06-27 10:43:59 -0700 | [diff] [blame] | 288 | @rtype: autotest_lib.server.cros.servo.servo.Servo |
Fang Deng | 5d518f4 | 2013-08-02 14:04:32 -0700 | [diff] [blame] | 289 | """ |
Dan Shi | 4d47852 | 2014-02-14 13:46:32 -0800 | [diff] [blame] | 290 | return self._servo |
| 291 | |
| 292 | |
Garry Wang | 79e9af6 | 2019-06-12 15:19:19 -0700 | [diff] [blame] | 293 | def request_reboot(self): |
| 294 | """Request servohost to be rebooted when it's safe to by touch a file. |
| 295 | """ |
| 296 | logging.debug('Request to reboot servohost %s has been created by ' |
Garry Wang | 464ff1e | 2019-07-18 17:20:34 -0700 | [diff] [blame] | 297 | 'servo with port # %s', self.hostname, self.servo_port) |
Garry Wang | 79e9af6 | 2019-06-12 15:19:19 -0700 | [diff] [blame] | 298 | self.run('touch %s' % self._reboot_file, ignore_status=True) |
| 299 | |
| 300 | |
Garry Wang | 464ff1e | 2019-07-18 17:20:34 -0700 | [diff] [blame] | 301 | def withdraw_reboot_request(self): |
| 302 | """Withdraw a servohost reboot request if exists by remove the flag |
| 303 | file. |
| 304 | """ |
| 305 | logging.debug('Withdrawing request to reboot servohost %s that created' |
| 306 | ' by servo with port # %s if exists.', |
| 307 | self.hostname, self.servo_port) |
| 308 | self.run('rm -f %s' % self._reboot_file, ignore_status=True) |
| 309 | |
| 310 | |
Garry Wang | c1288cf | 2019-12-17 14:58:00 -0800 | [diff] [blame] | 311 | def start_servod(self, quick_startup=False): |
| 312 | """Start the servod process on servohost. |
| 313 | """ |
Garry Wang | 2ac15ee | 2019-12-30 19:03:02 -0800 | [diff] [blame] | 314 | # Skip if running on the localhost.(crbug.com/1038168) |
| 315 | if self.is_localhost(): |
| 316 | logging.debug("Servohost is a localhost, skipping start servod.") |
| 317 | return |
| 318 | |
| 319 | cmd = 'start servod' |
Garry Wang | c1288cf | 2019-12-17 14:58:00 -0800 | [diff] [blame] | 320 | if self.servo_board: |
Garry Wang | 2ac15ee | 2019-12-30 19:03:02 -0800 | [diff] [blame] | 321 | cmd += ' BOARD=%s' % self.servo_board |
Garry Wang | c1288cf | 2019-12-17 14:58:00 -0800 | [diff] [blame] | 322 | if self.servo_model: |
| 323 | cmd += ' MODEL=%s' % self.servo_model |
Garry Wang | c1288cf | 2019-12-17 14:58:00 -0800 | [diff] [blame] | 324 | else: |
Garry Wang | 2ac15ee | 2019-12-30 19:03:02 -0800 | [diff] [blame] | 325 | logging.warning('Board for DUT is unknown; starting servod' |
| 326 | ' assuming a pre-configured board.') |
| 327 | |
| 328 | cmd += ' PORT=%d' % self.servo_port |
| 329 | if self.servo_serial: |
| 330 | cmd += ' SERIAL=%s' % self.servo_serial |
Garry Wang | d736748 | 2020-02-27 13:52:40 -0800 | [diff] [blame] | 331 | |
| 332 | # Start servod with dual_v4 if the DUT/servo from designated pools. |
| 333 | dut_host_info = self.get_dut_host_info() |
| 334 | if dut_host_info: |
Garry Wang | 11b5e87 | 2020-03-11 15:14:08 -0700 | [diff] [blame^] | 335 | if bool(dut_host_info.pools & |
| 336 | servo_constants.POOLS_SUPPORT_DUAL_V4): |
Garry Wang | d736748 | 2020-02-27 13:52:40 -0800 | [diff] [blame] | 337 | logging.debug('The DUT is detected in following designated' |
| 338 | ' pools %s,starting servod with DUAL_V4 option.', |
Garry Wang | 11b5e87 | 2020-03-11 15:14:08 -0700 | [diff] [blame^] | 339 | servo_constants.POOLS_SUPPORT_DUAL_V4) |
Garry Wang | d736748 | 2020-02-27 13:52:40 -0800 | [diff] [blame] | 340 | cmd += ' DUAL_V4=1' |
| 341 | |
Ruben Rodriguez Buchillon | 93084d0 | 2020-01-21 15:17:36 -0800 | [diff] [blame] | 342 | # Remove the symbolic links from the logs. This helps ensure that |
| 343 | # a failed servod instantiation does not cause us to grab old logs |
| 344 | # by mistake. |
| 345 | self.remove_latest_log_symlinks() |
Garry Wang | cdd27b2 | 2020-01-13 14:59:11 -0800 | [diff] [blame] | 346 | self.run(cmd, timeout=60) |
Garry Wang | c1288cf | 2019-12-17 14:58:00 -0800 | [diff] [blame] | 347 | |
| 348 | # There's a lag between when `start servod` completes and when |
| 349 | # the _ServodConnectionVerifier trigger can actually succeed. |
| 350 | # The call to time.sleep() below gives time to make sure that |
| 351 | # the trigger won't fail after we return. |
| 352 | |
| 353 | # Normally servod on servo_v3 and labstation take ~10 seconds to ready, |
| 354 | # But in the rare case all servo on a labstation are in heavy use they |
| 355 | # may take ~30 seconds. So the timeout value will double these value, |
| 356 | # and we'll try quick start up when first time initialize servohost, |
| 357 | # and use standard start up timeout in repair. |
| 358 | if quick_startup: |
Garry Wang | 11b5e87 | 2020-03-11 15:14:08 -0700 | [diff] [blame^] | 359 | timeout = servo_constants.SERVOD_QUICK_STARTUP_TIMEOUT |
Garry Wang | c1288cf | 2019-12-17 14:58:00 -0800 | [diff] [blame] | 360 | else: |
Garry Wang | 11b5e87 | 2020-03-11 15:14:08 -0700 | [diff] [blame^] | 361 | timeout = servo_constants.SERVOD_STARTUP_TIMEOUT |
Garry Wang | c1288cf | 2019-12-17 14:58:00 -0800 | [diff] [blame] | 362 | logging.debug('Wait %s seconds for servod process fully up.', timeout) |
| 363 | time.sleep(timeout) |
| 364 | |
| 365 | |
| 366 | def stop_servod(self): |
| 367 | """Stop the servod process on servohost. |
| 368 | """ |
Garry Wang | 2ac15ee | 2019-12-30 19:03:02 -0800 | [diff] [blame] | 369 | # Skip if running on the localhost.(crbug.com/1038168) |
| 370 | if self.is_localhost(): |
| 371 | logging.debug("Servohost is a localhost, skipping stop servod.") |
| 372 | return |
| 373 | |
Garry Wang | c1288cf | 2019-12-17 14:58:00 -0800 | [diff] [blame] | 374 | logging.debug('Stopping servod on port %s', self.servo_port) |
Garry Wang | cdd27b2 | 2020-01-13 14:59:11 -0800 | [diff] [blame] | 375 | self.run('stop servod PORT=%d' % self.servo_port, |
| 376 | timeout=60, ignore_status=True) |
Garry Wang | c1288cf | 2019-12-17 14:58:00 -0800 | [diff] [blame] | 377 | logging.debug('Wait %s seconds for servod process fully teardown.', |
Garry Wang | 11b5e87 | 2020-03-11 15:14:08 -0700 | [diff] [blame^] | 378 | servo_constants.SERVOD_TEARDOWN_TIMEOUT) |
| 379 | time.sleep(servo_constants.SERVOD_TEARDOWN_TIMEOUT) |
Garry Wang | c1288cf | 2019-12-17 14:58:00 -0800 | [diff] [blame] | 380 | |
| 381 | |
| 382 | def restart_servod(self, quick_startup=False): |
| 383 | """Restart the servod process on servohost. |
| 384 | """ |
| 385 | self.stop_servod() |
| 386 | self.start_servod(quick_startup) |
| 387 | |
Garry Wang | 11b5e87 | 2020-03-11 15:14:08 -0700 | [diff] [blame^] | 388 | |
Ruben Rodriguez Buchillon | 93084d0 | 2020-01-21 15:17:36 -0800 | [diff] [blame] | 389 | def _extract_compressed_logs(self, logdir, relevant_files): |
| 390 | """Decompress servod logs in |logdir|. |
| 391 | |
| 392 | @param logdir: directory containing compressed servod logs. |
| 393 | @param relevant_files: list of files in |logdir| to consider. |
| 394 | |
| 395 | @returns: tuple, (tarfiles, files) where |
| 396 | tarfiles: list of the compressed filenames that have been |
| 397 | extracted and deleted |
| 398 | files: list of the uncompressed files that were generated |
| 399 | """ |
| 400 | # For all tar-files, first extract them to the directory, and |
| 401 | # then let the common flow handle them. |
| 402 | tarfiles = [cf for cf in relevant_files if |
| 403 | cf.endswith(self.COMPRESSION_SUFFIX)] |
| 404 | files = [] |
| 405 | for f in tarfiles: |
| 406 | norm_name = os.path.basename(f)[:-len(self.COMPRESSION_SUFFIX)] |
| 407 | with tarfile.open(f) as tf: |
| 408 | # Each tarfile has only one member, as |
| 409 | # that's the compressed log. |
| 410 | member = tf.members[0] |
| 411 | # Manipulate so that it only extracts the basename, and not |
| 412 | # the directories etc. |
| 413 | member.name = norm_name |
| 414 | files.append(os.path.join(logdir, member.name)) |
| 415 | tf.extract(member, logdir) |
| 416 | # File has been extracted: remove the compressed file. |
| 417 | os.remove(f) |
| 418 | return tarfiles, files |
| 419 | |
| 420 | def _extract_mcu_logs(self, log_subdir): |
| 421 | """Extract MCU (EC, Cr50, etc) console output from servod debug logs. |
| 422 | |
| 423 | Using the MCU_EXTRACTOR regex (above) extract and split out MCU console |
| 424 | lines from the logs to generate invidiual console logs e.g. after |
| 425 | this method, you can find an ec.txt and servo_v4.txt in |log_dir| if |
| 426 | those MCUs had any console input/output. |
| 427 | |
| 428 | @param log_subdir: directory with log.DEBUG.txt main servod debug logs. |
| 429 | """ |
| 430 | # Extract the MCU for each one. The MCU logs are only in the .DEBUG |
| 431 | # files |
| 432 | mcu_lines_file = os.path.join(log_subdir, 'log.DEBUG.txt') |
| 433 | if not os.path.exists(mcu_lines_file): |
| 434 | logging.info('No DEBUG logs found to extract MCU logs from.') |
| 435 | return |
| 436 | mcu_files = {} |
| 437 | mcu_file_template = '%s.txt' |
| 438 | with open(mcu_lines_file, 'r') as f: |
| 439 | for line in f: |
| 440 | match = self.MCU_EXTRACTOR.match(line) |
| 441 | if match: |
| 442 | mcu = match.group(self.MCU_GROUP).lower() |
| 443 | line = match.group(self.LINE_GROUP) |
| 444 | if mcu not in mcu_files: |
| 445 | mcu_file = os.path.join(log_subdir, |
| 446 | mcu_file_template % mcu) |
| 447 | mcu_files[mcu] = open(mcu_file, 'a') |
| 448 | fd = mcu_files[mcu] |
| 449 | fd.write(line + '\n') |
| 450 | for f in mcu_files: |
| 451 | mcu_files[f].close() |
| 452 | |
| 453 | |
| 454 | def remove_latest_log_symlinks(self): |
| 455 | """Remove the conveninence symlinks 'latest' servod logs.""" |
| 456 | symlink_wildcard = '%s/latest*' % self.remote_log_dir |
| 457 | cmd = 'rm ' + symlink_wildcard |
| 458 | self.run(cmd, stderr_tee=None, ignore_status=True) |
| 459 | |
| 460 | def grab_logs(self, outdir): |
| 461 | """Retrieve logs from servo_host to |outdir|/servod_{port}.{ts}/. |
| 462 | |
| 463 | This method first collects all logs on the servo_host side pertaining |
| 464 | to this servod instance (port, instatiation). It glues them together |
| 465 | into combined log.[level].txt files and extracts all available MCU |
| 466 | console I/O from the logs into individual files e.g. servo_v4.txt |
| 467 | |
| 468 | All the output can be found in a directory inside |outdir| that |
| 469 | this generates based on |LOG_DIR|, the servod port, and the instance |
| 470 | timestamp on the servo_host side. |
| 471 | |
| 472 | @param outdir: directory to create a subdirectory into to place the |
| 473 | servod logs into. |
| 474 | """ |
| 475 | # First, extract the timestamp. This cmd gives the real filename of |
| 476 | # the latest aka current log file. |
| 477 | cmd = ('if [ -f %(dir)s/latest.DEBUG ];' |
| 478 | 'then realpath %(dir)s/latest.DEBUG;' |
| 479 | 'elif [ -f %(dir)s/latest ];' |
| 480 | 'then realpath %(dir)s/latest;' |
| 481 | 'else exit %(code)d;' |
| 482 | 'fi' % {'dir': self.remote_log_dir, |
| 483 | 'code': self.NO_SYMLINKS_CODE}) |
| 484 | res = self.run(cmd, stderr_tee=None, ignore_status=True) |
| 485 | if res.exit_status != 0: |
| 486 | if res.exit_status == self.NO_SYMLINKS_CODE: |
| 487 | logging.warning('servod log latest symlinks not found. ' |
| 488 | 'This is likely due to an error starting up ' |
| 489 | 'servod. Ignoring..') |
| 490 | else: |
| 491 | logging.warning('Failed to find servod logs on servo host.') |
| 492 | logging.warning(res.stderr.strip()) |
| 493 | return |
| 494 | fname = os.path.basename(res.stdout.strip()) |
| 495 | # From the fname, ought to extract the timestamp using the TS_EXTRACTOR |
Ruben Rodriguez Buchillon | e9aa2b0 | 2020-03-04 12:14:28 -0800 | [diff] [blame] | 496 | ts_match = self.TS_EXTRACTOR.match(fname) |
| 497 | if not ts_match: |
| 498 | logging.warning('Failed to extract timestamp from servod log file ' |
| 499 | '%r. Skipping. The servo host is using outdated ' |
| 500 | 'servod logging and needs to be updated.', fname) |
| 501 | return |
| 502 | instance_ts = ts_match.group(self.TS_GROUP) |
Ruben Rodriguez Buchillon | 93084d0 | 2020-01-21 15:17:36 -0800 | [diff] [blame] | 503 | # Create the local results log dir. |
| 504 | log_dir = os.path.join(outdir, '%s_%s.%s' % (self.LOG_DIR, |
| 505 | str(self.servo_port), |
| 506 | instance_ts)) |
| 507 | logging.info('Saving servod logs to %s.', log_dir) |
| 508 | os.mkdir(log_dir) |
| 509 | # Now, get all files with that timestamp. |
| 510 | cmd = 'find %s -maxdepth 1 -name "log.%s*"' % (self.remote_log_dir, |
| 511 | instance_ts) |
| 512 | res = self.run(cmd, stderr_tee=None, ignore_status=True) |
| 513 | files = res.stdout.strip().split() |
| 514 | try: |
| 515 | self.get_file(files, log_dir, try_rsync=False) |
| 516 | |
| 517 | except error.AutoservRunError as e: |
| 518 | result = e.result_obj |
| 519 | if result.exit_status != 0: |
| 520 | stderr = result.stderr.strip() |
| 521 | logging.warning("Couldn't retrieve servod logs. Ignoring: %s", |
| 522 | stderr or '\n%s' % result) |
| 523 | return |
| 524 | local_files = [os.path.join(log_dir, f) for f in os.listdir(log_dir)] |
| 525 | # TODO(crrev.com/c/1793030): remove no-level case once CL is pushed |
| 526 | for level_name in ('DEBUG', 'INFO', 'WARNING', ''): |
| 527 | # Create the joint files for each loglevel. i.e log.DEBUG |
| 528 | joint_file = self.JOINT_LOG_PREFIX |
| 529 | if level_name: |
| 530 | joint_file = '%s.%s' % (self.JOINT_LOG_PREFIX, level_name) |
| 531 | # This helps with some online tools to avoid complaints about an |
| 532 | # unknown filetype. |
| 533 | joint_file = joint_file + '.txt' |
| 534 | joint_path = os.path.join(log_dir, joint_file) |
| 535 | files = [f for f in local_files if level_name in f] |
| 536 | if not files: |
| 537 | # TODO(crrev.com/c/1793030): remove no-level case once CL |
| 538 | # is pushed |
| 539 | continue |
| 540 | # Extract compressed logs if any. |
| 541 | compressed, extracted = self._extract_compressed_logs(log_dir, |
| 542 | files) |
| 543 | files = list(set(files) - set(compressed)) |
| 544 | files.extend(extracted) |
| 545 | # Need to sort. As they all share the same timestamp, and |
| 546 | # loglevel, the index itself is sufficient. The highest index |
| 547 | # is the oldest file, therefore we need a descending sort. |
| 548 | def sortkey(f, level=level_name): |
| 549 | """Custom sortkey to sort based on rotation number int.""" |
| 550 | if f.endswith(level_name): return 0 |
| 551 | return int(f.split('.')[-1]) |
| 552 | |
| 553 | files.sort(reverse=True, key=sortkey) |
| 554 | # Just rename the first file rather than building from scratch. |
| 555 | os.rename(files[0], joint_path) |
| 556 | with open(joint_path, 'a') as joint_f: |
| 557 | for logfile in files[1:]: |
| 558 | # Transfer the file to the joint file line by line. |
| 559 | with open(logfile, 'r') as log_f: |
| 560 | for line in log_f: |
| 561 | joint_f.write(line) |
| 562 | # File has been written over. Delete safely. |
| 563 | os.remove(logfile) |
| 564 | # Need to remove all files form |local_files| so we don't |
| 565 | # analyze them again. |
| 566 | local_files = list(set(local_files) - set(files) - set(compressed)) |
| 567 | # Lastly, extract MCU logs from the joint logs. |
| 568 | self._extract_mcu_logs(log_dir) |
| 569 | |
Garry Wang | c1288cf | 2019-12-17 14:58:00 -0800 | [diff] [blame] | 570 | |
Garry Wang | 79e9af6 | 2019-06-12 15:19:19 -0700 | [diff] [blame] | 571 | def _lock(self): |
| 572 | """lock servohost by touching a file. |
| 573 | """ |
| 574 | logging.debug('Locking servohost %s by touching %s file', |
| 575 | self.hostname, self._lock_file) |
| 576 | self.run('touch %s' % self._lock_file, ignore_status=True) |
Garry Wang | 7c00b0f | 2019-06-25 17:28:17 -0700 | [diff] [blame] | 577 | self._is_locked = True |
Garry Wang | 79e9af6 | 2019-06-12 15:19:19 -0700 | [diff] [blame] | 578 | |
| 579 | |
| 580 | def _unlock(self): |
| 581 | """Unlock servohost by removing the lock file. |
| 582 | """ |
| 583 | logging.debug('Unlocking servohost by removing %s file', |
| 584 | self._lock_file) |
| 585 | self.run('rm %s' % self._lock_file, ignore_status=True) |
Garry Wang | 7c00b0f | 2019-06-25 17:28:17 -0700 | [diff] [blame] | 586 | self._is_locked = False |
Garry Wang | 79e9af6 | 2019-06-12 15:19:19 -0700 | [diff] [blame] | 587 | |
| 588 | |
Congbin Guo | a1f9cba | 2018-07-03 11:36:59 -0700 | [diff] [blame] | 589 | def close(self): |
Congbin Guo | fc3b896 | 2019-03-22 17:38:46 -0700 | [diff] [blame] | 590 | """Close the associated servo and the host object.""" |
Ruben Rodriguez Buchillon | 93084d0 | 2020-01-21 15:17:36 -0800 | [diff] [blame] | 591 | if self._closed: |
| 592 | logging.debug('ServoHost is already closed.') |
| 593 | return |
Congbin Guo | a1f9cba | 2018-07-03 11:36:59 -0700 | [diff] [blame] | 594 | if self._servo: |
Ruben Rodriguez Buchillon | 93084d0 | 2020-01-21 15:17:36 -0800 | [diff] [blame] | 595 | outdir = None if not self.job else self.job.resultdir |
Congbin Guo | 2e5e2a2 | 2018-07-27 10:32:48 -0700 | [diff] [blame] | 596 | # In some cases when we run as lab-tools, the job object is None. |
Ruben Rodriguez Buchillon | 93084d0 | 2020-01-21 15:17:36 -0800 | [diff] [blame] | 597 | self._servo.close(outdir) |
| 598 | |
| 599 | if self.job and not self.is_localhost(): |
| 600 | # Grab all logs from this servod instance before stopping servod. |
| 601 | # TODO(crbug.com/1011516): once enabled, remove the check against |
| 602 | # localhost and instead check against log-rotiation enablement. |
| 603 | try: |
| 604 | self.grab_logs(self.job.resultdir) |
| 605 | except error.AutoservRunError as e: |
| 606 | logging.info('Failed to grab servo logs due to: %s. ' |
| 607 | 'This error is forgiven.', str(e)) |
Congbin Guo | a1f9cba | 2018-07-03 11:36:59 -0700 | [diff] [blame] | 608 | |
Garry Wang | 7c00b0f | 2019-06-25 17:28:17 -0700 | [diff] [blame] | 609 | if self._is_locked: |
| 610 | # Remove the lock if the servohost has been locked. |
Garry Wang | 79e9af6 | 2019-06-12 15:19:19 -0700 | [diff] [blame] | 611 | try: |
| 612 | self._unlock() |
| 613 | except error.AutoservSSHTimeout: |
| 614 | logging.error('Unlock servohost failed due to ssh timeout.' |
| 615 | ' It may caused by servohost went down during' |
| 616 | ' the task.') |
Garry Wang | c1288cf | 2019-12-17 14:58:00 -0800 | [diff] [blame] | 617 | # We want always stop servod after task to minimum the impact of bad |
| 618 | # servod process interfere other servods.(see crbug.com/1028665) |
Garry Wang | 4c624bc | 2020-01-27 16:34:43 -0800 | [diff] [blame] | 619 | try: |
| 620 | self.stop_servod() |
| 621 | except error.AutoservRunError as e: |
| 622 | logging.info("Failed to stop servod due to:\n%s\n" |
Ruben Rodriguez Buchillon | 93084d0 | 2020-01-21 15:17:36 -0800 | [diff] [blame] | 623 | "This error is forgiven.", str(e)) |
Garry Wang | c1288cf | 2019-12-17 14:58:00 -0800 | [diff] [blame] | 624 | |
Congbin Guo | a1f9cba | 2018-07-03 11:36:59 -0700 | [diff] [blame] | 625 | super(ServoHost, self).close() |
Ruben Rodriguez Buchillon | 93084d0 | 2020-01-21 15:17:36 -0800 | [diff] [blame] | 626 | # Mark closed. |
| 627 | self._closed = True |
Congbin Guo | a1f9cba | 2018-07-03 11:36:59 -0700 | [diff] [blame] | 628 | |
| 629 | |
Otabek Kasimov | cc9738e | 2020-02-14 16:17:15 -0800 | [diff] [blame] | 630 | def get_servo_state(self): |
Otabek Kasimov | a7ba91a | 2020-03-09 08:31:01 -0700 | [diff] [blame] | 631 | if self._servo_state is None: |
Garry Wang | 11b5e87 | 2020-03-11 15:14:08 -0700 | [diff] [blame^] | 632 | return servo_constants.SERVO_STATE_UNKNOWN |
Otabek Kasimov | a7ba91a | 2020-03-09 08:31:01 -0700 | [diff] [blame] | 633 | return self._servo_state |
Otabek Kasimov | cc9738e | 2020-02-14 16:17:15 -0800 | [diff] [blame] | 634 | |
| 635 | |
Richard Barnette | ea3e460 | 2016-06-10 12:36:41 -0700 | [diff] [blame] | 636 | def make_servo_hostname(dut_hostname): |
| 637 | """Given a DUT's hostname, return the hostname of its servo. |
| 638 | |
| 639 | @param dut_hostname: hostname of a DUT. |
| 640 | |
| 641 | @return hostname of the DUT's servo. |
| 642 | |
| 643 | """ |
| 644 | host_parts = dut_hostname.split('.') |
| 645 | host_parts[0] = host_parts[0] + '-servo' |
| 646 | return '.'.join(host_parts) |
| 647 | |
| 648 | |
| 649 | def servo_host_is_up(servo_hostname): |
Kevin Cheng | 5f2ba6c | 2016-09-28 10:20:05 -0700 | [diff] [blame] | 650 | """Given a servo host name, return if it's up or not. |
Richard Barnette | ea3e460 | 2016-06-10 12:36:41 -0700 | [diff] [blame] | 651 | |
| 652 | @param servo_hostname: hostname of the servo host. |
| 653 | |
| 654 | @return True if it's up, False otherwise |
| 655 | """ |
| 656 | # Technically, this duplicates the SSH ping done early in the servo |
| 657 | # proxy initialization code. However, this ping ends in a couple |
| 658 | # seconds when if fails, rather than the 60 seconds it takes to decide |
| 659 | # that an SSH ping has timed out. Specifically, that timeout happens |
| 660 | # when our servo DNS name resolves, but there is no host at that IP. |
| 661 | logging.info('Pinging servo host at %s', servo_hostname) |
| 662 | ping_config = ping_runner.PingConfig( |
| 663 | servo_hostname, count=3, |
| 664 | ignore_result=True, ignore_status=True) |
| 665 | return ping_runner.PingRunner().ping(ping_config).received > 0 |
| 666 | |
| 667 | |
Richard Barnette | e519dcd | 2016-08-15 17:37:17 -0700 | [diff] [blame] | 668 | def _map_afe_board_to_servo_board(afe_board): |
| 669 | """Map a board we get from the AFE to a servo appropriate value. |
| 670 | |
| 671 | Many boards are identical to other boards for servo's purposes. |
| 672 | This function makes that mapping. |
| 673 | |
| 674 | @param afe_board string board name received from AFE. |
| 675 | @return board we expect servo to have. |
| 676 | |
| 677 | """ |
| 678 | KNOWN_SUFFIXES = ['-freon', '_freon', '_moblab', '-cheets'] |
| 679 | BOARD_MAP = {'gizmo': 'panther'} |
| 680 | mapped_board = afe_board |
| 681 | if afe_board in BOARD_MAP: |
| 682 | mapped_board = BOARD_MAP[afe_board] |
| 683 | else: |
| 684 | for suffix in KNOWN_SUFFIXES: |
| 685 | if afe_board.endswith(suffix): |
| 686 | mapped_board = afe_board[0:-len(suffix)] |
| 687 | break |
| 688 | if mapped_board != afe_board: |
| 689 | logging.info('Mapping AFE board=%s to %s', afe_board, mapped_board) |
| 690 | return mapped_board |
| 691 | |
| 692 | |
Prathmesh Prabhu | b481023 | 2018-09-07 13:24:08 -0700 | [diff] [blame] | 693 | def get_servo_args_for_host(dut_host): |
Kevin Cheng | 5f2ba6c | 2016-09-28 10:20:05 -0700 | [diff] [blame] | 694 | """Return servo data associated with a given DUT. |
Richard Barnette | ea3e460 | 2016-06-10 12:36:41 -0700 | [diff] [blame] | 695 | |
Richard Barnette | ea3e460 | 2016-06-10 12:36:41 -0700 | [diff] [blame] | 696 | @param dut_host Instance of `Host` on which to find the servo |
| 697 | attributes. |
Prathmesh Prabhu | f605dd3 | 2018-08-28 17:09:04 -0700 | [diff] [blame] | 698 | @return `servo_args` dict with host and an optional port. |
Richard Barnette | ea3e460 | 2016-06-10 12:36:41 -0700 | [diff] [blame] | 699 | """ |
Prathmesh Prabhu | cba4429 | 2018-08-28 17:44:45 -0700 | [diff] [blame] | 700 | info = dut_host.host_info_store.get() |
| 701 | servo_args = {k: v for k, v in info.attributes.iteritems() |
Garry Wang | 11b5e87 | 2020-03-11 15:14:08 -0700 | [diff] [blame^] | 702 | if k in servo_constants.SERVO_ATTR_KEYS} |
Richard Barnette | ea3e460 | 2016-06-10 12:36:41 -0700 | [diff] [blame] | 703 | |
Garry Wang | 11b5e87 | 2020-03-11 15:14:08 -0700 | [diff] [blame^] | 704 | if servo_constants.SERVO_PORT_ATTR in servo_args: |
Prathmesh Prabhu | cba4429 | 2018-08-28 17:44:45 -0700 | [diff] [blame] | 705 | try: |
Garry Wang | 11b5e87 | 2020-03-11 15:14:08 -0700 | [diff] [blame^] | 706 | servo_args[servo_constants.SERVO_PORT_ATTR] = int( |
| 707 | servo_args[servo_constants.SERVO_PORT_ATTR]) |
Prathmesh Prabhu | cba4429 | 2018-08-28 17:44:45 -0700 | [diff] [blame] | 708 | except ValueError: |
| 709 | logging.error('servo port is not an int: %s', |
Garry Wang | 11b5e87 | 2020-03-11 15:14:08 -0700 | [diff] [blame^] | 710 | servo_args[servo_constants.SERVO_PORT_ATTR]) |
Prathmesh Prabhu | cba4429 | 2018-08-28 17:44:45 -0700 | [diff] [blame] | 711 | # Reset servo_args because we don't want to use an invalid port. |
Garry Wang | 11b5e87 | 2020-03-11 15:14:08 -0700 | [diff] [blame^] | 712 | servo_args.pop(servo_constants.SERVO_HOST_ATTR, None) |
Prathmesh Prabhu | cba4429 | 2018-08-28 17:44:45 -0700 | [diff] [blame] | 713 | |
| 714 | if info.board: |
Garry Wang | 11b5e87 | 2020-03-11 15:14:08 -0700 | [diff] [blame^] | 715 | servo_board = _map_afe_board_to_servo_board(info.board) |
| 716 | servo_args[servo_constants.SERVO_BOARD_ATTR] = servo_board |
Nick Sanders | 2f3c985 | 2018-10-24 12:10:24 -0700 | [diff] [blame] | 717 | if info.model: |
Garry Wang | 11b5e87 | 2020-03-11 15:14:08 -0700 | [diff] [blame^] | 718 | servo_args[servo_constants.SERVO_MODEL_ATTR] = info.model |
| 719 | return servo_args if servo_constants.SERVO_HOST_ATTR in servo_args else None |
Richard Barnette | ea3e460 | 2016-06-10 12:36:41 -0700 | [diff] [blame] | 720 | |
| 721 | |
Prathmesh Prabhu | efb1b48 | 2018-08-28 17:15:05 -0700 | [diff] [blame] | 722 | def _tweak_args_for_ssp_moblab(servo_args): |
Garry Wang | 11b5e87 | 2020-03-11 15:14:08 -0700 | [diff] [blame^] | 723 | if (servo_args[servo_constants.SERVO_HOST_ATTR] |
| 724 | in ['localhost', '127.0.0.1']): |
| 725 | servo_args[servo_constants.SERVO_HOST_ATTR] = _CONFIG.get_config_value( |
Prathmesh Prabhu | efb1b48 | 2018-08-28 17:15:05 -0700 | [diff] [blame] | 726 | 'SSP', 'host_container_ip', type=str, default=None) |
| 727 | |
| 728 | |
Dan Shi | 023aae3 | 2016-05-25 11:13:01 -0700 | [diff] [blame] | 729 | def create_servo_host(dut, servo_args, try_lab_servo=False, |
Gregory Nisbet | de13e2a | 2019-12-09 22:44:00 -0800 | [diff] [blame] | 730 | try_servo_repair=False, dut_host_info=None): |
Kevin Cheng | 5f2ba6c | 2016-09-28 10:20:05 -0700 | [diff] [blame] | 731 | """Create a ServoHost object for a given DUT, if appropriate. |
Dan Shi | 4d47852 | 2014-02-14 13:46:32 -0800 | [diff] [blame] | 732 | |
Richard Barnette | 9a26ad6 | 2016-06-10 12:03:08 -0700 | [diff] [blame] | 733 | This function attempts to create and verify or repair a `ServoHost` |
| 734 | object for a servo connected to the given `dut`, subject to various |
| 735 | constraints imposed by the parameters: |
| 736 | * When the `servo_args` parameter is not `None`, a servo |
| 737 | host must be created, and must be checked with `repair()`. |
| 738 | * Otherwise, if a servo exists in the lab and `try_lab_servo` is |
| 739 | true: |
| 740 | * If `try_servo_repair` is true, then create a servo host and |
| 741 | check it with `repair()`. |
| 742 | * Otherwise, if the servo responds to `ping` then create a |
| 743 | servo host and check it with `verify()`. |
Fang Deng | e545abb | 2014-12-30 18:43:47 -0800 | [diff] [blame] | 744 | |
Richard Barnette | 9a26ad6 | 2016-06-10 12:03:08 -0700 | [diff] [blame] | 745 | In cases where `servo_args` was not `None`, repair failure |
| 746 | exceptions are passed back to the caller; otherwise, exceptions |
Richard Barnette | 07c2e1d | 2016-10-26 14:24:28 -0700 | [diff] [blame] | 747 | are logged and then discarded. Note that this only happens in cases |
| 748 | where we're called from a test (not special task) control file that |
| 749 | has an explicit dependency on servo. In that case, we require that |
| 750 | repair not write to `status.log`, so as to avoid polluting test |
| 751 | results. |
| 752 | |
| 753 | TODO(jrbarnette): The special handling for servo in test control |
| 754 | files is a thorn in my flesh; I dearly hope to see it cut out before |
| 755 | my retirement. |
Richard Barnette | 9a26ad6 | 2016-06-10 12:03:08 -0700 | [diff] [blame] | 756 | |
| 757 | Parameters for a servo host consist of a host name, port number, and |
| 758 | DUT board, and are determined from one of these sources, in order of |
| 759 | priority: |
Richard Barnette | ea3e460 | 2016-06-10 12:36:41 -0700 | [diff] [blame] | 760 | * Servo attributes from the `dut` parameter take precedence over |
| 761 | all other sources of information. |
| 762 | * If a DNS entry for the servo based on the DUT hostname exists in |
| 763 | the CrOS lab network, that hostname is used with the default |
Richard Barnette | 9a26ad6 | 2016-06-10 12:03:08 -0700 | [diff] [blame] | 764 | port and the DUT's board. |
Richard Barnette | ea3e460 | 2016-06-10 12:36:41 -0700 | [diff] [blame] | 765 | * If no other options are found, the parameters will be taken |
Richard Barnette | 9a26ad6 | 2016-06-10 12:03:08 -0700 | [diff] [blame] | 766 | from the `servo_args` dict passed in from the caller. |
Richard Barnette | ea3e460 | 2016-06-10 12:36:41 -0700 | [diff] [blame] | 767 | |
| 768 | @param dut An instance of `Host` from which to take |
| 769 | servo parameters (if available). |
| 770 | @param servo_args A dictionary with servo parameters to use if |
| 771 | they can't be found from `dut`. If this |
| 772 | argument is supplied, unrepaired exceptions |
| 773 | from `verify()` will be passed back to the |
| 774 | caller. |
| 775 | @param try_lab_servo If not true, servo host creation will be |
| 776 | skipped unless otherwise required by the |
| 777 | caller. |
Richard Barnette | 9a26ad6 | 2016-06-10 12:03:08 -0700 | [diff] [blame] | 778 | @param try_servo_repair If true, check a servo host with |
| 779 | `repair()` instead of `verify()`. |
Dan Shi | 4d47852 | 2014-02-14 13:46:32 -0800 | [diff] [blame] | 780 | |
| 781 | @returns: A ServoHost object or None. See comments above. |
| 782 | |
| 783 | """ |
Richard Barnette | 07c2e1d | 2016-10-26 14:24:28 -0700 | [diff] [blame] | 784 | servo_dependency = servo_args is not None |
Richard Barnette | 07c2e1d | 2016-10-26 14:24:28 -0700 | [diff] [blame] | 785 | if dut is not None and (try_lab_servo or servo_dependency): |
Prathmesh Prabhu | b481023 | 2018-09-07 13:24:08 -0700 | [diff] [blame] | 786 | servo_args_override = get_servo_args_for_host(dut) |
Richard Barnette | ea3e460 | 2016-06-10 12:36:41 -0700 | [diff] [blame] | 787 | if servo_args_override is not None: |
Prathmesh Prabhu | efb1b48 | 2018-08-28 17:15:05 -0700 | [diff] [blame] | 788 | if utils.in_moblab_ssp(): |
| 789 | _tweak_args_for_ssp_moblab(servo_args_override) |
Prathmesh Prabhu | 88bf605 | 2018-08-28 16:21:26 -0700 | [diff] [blame] | 790 | logging.debug( |
| 791 | 'Overriding provided servo_args (%s) with arguments' |
| 792 | ' determined from the host (%s)', |
| 793 | servo_args, |
| 794 | servo_args_override, |
| 795 | ) |
Richard Barnette | ea3e460 | 2016-06-10 12:36:41 -0700 | [diff] [blame] | 796 | servo_args = servo_args_override |
Prathmesh Prabhu | cba4429 | 2018-08-28 17:44:45 -0700 | [diff] [blame] | 797 | |
Richard Barnette | ea3e460 | 2016-06-10 12:36:41 -0700 | [diff] [blame] | 798 | if servo_args is None: |
Prathmesh Prabhu | 88bf605 | 2018-08-28 16:21:26 -0700 | [diff] [blame] | 799 | logging.debug('No servo_args provided, and failed to find overrides.') |
Otabek Kasimov | a7ba91a | 2020-03-09 08:31:01 -0700 | [diff] [blame] | 800 | if try_lab_servo or servo_dependency: |
Garry Wang | 11b5e87 | 2020-03-11 15:14:08 -0700 | [diff] [blame^] | 801 | return None, servo_constants.SERVO_STATE_NOT_CONNECTED |
Otabek Kasimov | a7ba91a | 2020-03-09 08:31:01 -0700 | [diff] [blame] | 802 | else: |
| 803 | # For regular test case which not required the servo |
| 804 | return None, None |
| 805 | |
Garry Wang | 11b5e87 | 2020-03-11 15:14:08 -0700 | [diff] [blame^] | 806 | servo_hostname = servo_args.get(servo_constants.SERVO_HOST_ATTR) |
| 807 | servo_port = servo_args.get(servo_constants.SERVO_PORT_ATTR) |
Otabek Kasimov | a7ba91a | 2020-03-09 08:31:01 -0700 | [diff] [blame] | 808 | if not _is_servo_host_information_exist(servo_hostname, servo_port): |
| 809 | logging.debug( |
| 810 | 'Servo connection info missed hostname: %s , port: %s', |
| 811 | servo_hostname, servo_port) |
Garry Wang | 11b5e87 | 2020-03-11 15:14:08 -0700 | [diff] [blame^] | 812 | return None, servo_constants.SERVO_STATE_NOT_CONNECTED |
Otabek Kasimov | a7ba91a | 2020-03-09 08:31:01 -0700 | [diff] [blame] | 813 | if not is_servo_host_information_valid(servo_hostname, servo_port): |
| 814 | logging.debug( |
| 815 | 'Servo connection info is incorrect hostname: %s , port: %s', |
| 816 | servo_hostname, servo_port) |
Garry Wang | 11b5e87 | 2020-03-11 15:14:08 -0700 | [diff] [blame^] | 817 | return None, servo_constants.SERVO_STATE_WRONG_CONFIG |
Richard Barnette | 07c2e1d | 2016-10-26 14:24:28 -0700 | [diff] [blame] | 818 | if (not servo_dependency and not try_servo_repair and |
Otabek Kasimov | a7ba91a | 2020-03-09 08:31:01 -0700 | [diff] [blame] | 819 | not servo_host_is_up(servo_hostname)): |
Prathmesh Prabhu | 88bf605 | 2018-08-28 16:21:26 -0700 | [diff] [blame] | 820 | logging.debug('ServoHost is not up.') |
Garry Wang | 11b5e87 | 2020-03-11 15:14:08 -0700 | [diff] [blame^] | 821 | return None, servo_constants.SERVO_STATE_BROKEN |
Prathmesh Prabhu | 88bf605 | 2018-08-28 16:21:26 -0700 | [diff] [blame] | 822 | |
Garry Wang | ebc015b | 2019-06-06 17:45:06 -0700 | [diff] [blame] | 823 | newhost = ServoHost(**servo_args) |
Garry Wang | cdd27b2 | 2020-01-13 14:59:11 -0800 | [diff] [blame] | 824 | try: |
| 825 | newhost.restart_servod(quick_startup=True) |
| 826 | except error.AutoservSSHTimeout: |
| 827 | logging.warning("Restart servod failed due ssh connection " |
| 828 | "to servohost timed out. This error is forgiven" |
| 829 | " here, we will retry in servo repair process.") |
| 830 | except error.AutoservRunError as e: |
| 831 | logging.warning("Restart servod failed due to:\n%s\n" |
| 832 | "This error is forgiven here, we will retry" |
| 833 | " in servo repair process.", str(e)) |
Garry Wang | ebc015b | 2019-06-06 17:45:06 -0700 | [diff] [blame] | 834 | |
Gregory Nisbet | de13e2a | 2019-12-09 22:44:00 -0800 | [diff] [blame] | 835 | # TODO(gregorynisbet): Clean all of this up. |
| 836 | logging.debug('create_servo_host: attempt to set info store on ' |
| 837 | 'servo host') |
| 838 | try: |
| 839 | if dut_host_info is None: |
| 840 | logging.debug('create_servo_host: dut_host_info is ' |
| 841 | 'None, skipping') |
| 842 | else: |
| 843 | newhost.set_dut_host_info(dut_host_info) |
| 844 | logging.debug('create_servo_host: successfully set info ' |
| 845 | 'store') |
| 846 | except Exception: |
| 847 | logging.error("create_servo_host: (%s)", traceback.format_exc()) |
| 848 | |
Richard Barnette | 9a26ad6 | 2016-06-10 12:03:08 -0700 | [diff] [blame] | 849 | # Note that the logic of repair() includes everything done |
| 850 | # by verify(). It's sufficient to call one or the other; |
| 851 | # we don't need both. |
Richard Barnette | 07c2e1d | 2016-10-26 14:24:28 -0700 | [diff] [blame] | 852 | if servo_dependency: |
| 853 | newhost.repair(silent=True) |
Otabek Kasimov | a7ba91a | 2020-03-09 08:31:01 -0700 | [diff] [blame] | 854 | return newhost, newhost.get_servo_state() |
Prathmesh Prabhu | 88bf605 | 2018-08-28 16:21:26 -0700 | [diff] [blame] | 855 | |
| 856 | if try_servo_repair: |
| 857 | try: |
| 858 | newhost.repair() |
| 859 | except Exception: |
| 860 | logging.exception('servo repair failed for %s', newhost.hostname) |
Richard Barnette | 9a26ad6 | 2016-06-10 12:03:08 -0700 | [diff] [blame] | 861 | else: |
| 862 | try: |
Prathmesh Prabhu | 88bf605 | 2018-08-28 16:21:26 -0700 | [diff] [blame] | 863 | newhost.verify() |
Kevin Cheng | 5f2ba6c | 2016-09-28 10:20:05 -0700 | [diff] [blame] | 864 | except Exception: |
Prathmesh Prabhu | 88bf605 | 2018-08-28 16:21:26 -0700 | [diff] [blame] | 865 | logging.exception('servo verify failed for %s', newhost.hostname) |
Otabek Kasimov | a7ba91a | 2020-03-09 08:31:01 -0700 | [diff] [blame] | 866 | return newhost, newhost.get_servo_state() |
Otabek Kasimov | 7267a7a | 2020-03-04 11:18:45 -0800 | [diff] [blame] | 867 | |
| 868 | |
Otabek Kasimov | a7ba91a | 2020-03-09 08:31:01 -0700 | [diff] [blame] | 869 | def _is_servo_host_information_exist(hostname, port): |
Otabek Kasimov | 7267a7a | 2020-03-04 11:18:45 -0800 | [diff] [blame] | 870 | if hostname is None or len(hostname.strip()) == 0: |
| 871 | return False |
Otabek Kasimov | a7ba91a | 2020-03-09 08:31:01 -0700 | [diff] [blame] | 872 | if port is None: |
Otabek Kasimov | 7267a7a | 2020-03-04 11:18:45 -0800 | [diff] [blame] | 873 | return False |
Otabek Kasimov | a7ba91a | 2020-03-09 08:31:01 -0700 | [diff] [blame] | 874 | if not type(port) is int: |
| 875 | try: |
| 876 | int(port) |
| 877 | except ValueError: |
| 878 | return False |
| 879 | |
Otabek Kasimov | 7267a7a | 2020-03-04 11:18:45 -0800 | [diff] [blame] | 880 | return True |
| 881 | |
| 882 | |
Otabek Kasimov | a7ba91a | 2020-03-09 08:31:01 -0700 | [diff] [blame] | 883 | def is_servo_host_information_valid(hostname, port): |
| 884 | if not _is_servo_host_information_exist(hostname, port): |
Otabek Kasimov | 7267a7a | 2020-03-04 11:18:45 -0800 | [diff] [blame] | 885 | return False |
| 886 | # checking range and correct of the port |
Otabek Kasimov | a7ba91a | 2020-03-09 08:31:01 -0700 | [diff] [blame] | 887 | port_int = int(port) |
Otabek Kasimov | 7267a7a | 2020-03-04 11:18:45 -0800 | [diff] [blame] | 888 | if port_int < 1 or port_int > 65000: |
| 889 | return False |
| 890 | # we expecting host contain only latters, digits and '-' or '_' |
Otabek Kasimov | a7ba91a | 2020-03-09 08:31:01 -0700 | [diff] [blame] | 891 | if not re.match('[a-zA-Z0-9-_\.]*$', hostname) or len(hostname) < 5: |
Otabek Kasimov | 7267a7a | 2020-03-04 11:18:45 -0800 | [diff] [blame] | 892 | return False |
| 893 | return True |