gecko-dev/build/mobile/remoteautomation.py

496 lines
20 KiB
Python

# This Source Code Form is subject to the terms of the Mozilla Public
# License, v. 2.0. If a copy of the MPL was not distributed with this
# file, You can obtain one at http://mozilla.org/MPL/2.0/.
import datetime
import glob
import time
import re
import os
import posixpath
import tempfile
import shutil
import sys
from automation import Automation
from mozdevice import ADBTimeoutError
from mozlog import get_default_logger
from mozscreenshot import dump_screen, dump_device_screen
import mozcrash
# signatures for logcat messages that we don't care about much
fennecLogcatFilters = ["The character encoding of the HTML document was not declared",
"Use of Mutation Events is deprecated. Use MutationObserver instead.",
"Unexpected value from nativeGetEnabledTags: 0"]
class RemoteAutomation(Automation):
def __init__(self, device, appName='', remoteProfile=None, remoteLog=None,
processArgs=None):
self.device = device
self.appName = appName
self.remoteProfile = remoteProfile
self.remoteLog = remoteLog
self.processArgs = processArgs or {}
self.lastTestSeen = "remoteautomation.py"
Automation.__init__(self)
def runApp(self, testURL, env, app, profileDir, extraArgs,
utilityPath=None, xrePath=None, debuggerInfo=None, symbolsPath=None,
timeout=-1, maxTime=None, e10s=True, **kwargs):
"""
Run the app, log the duration it took to execute, return the status code.
Kills the app if it runs for longer than |maxTime| seconds, or outputs nothing
for |timeout| seconds.
"""
if self.device.is_file(self.remoteLog):
self.device.rm(self.remoteLog, root=True)
self.log.info("remoteautomation.py | runApp deleted %s" % self.remoteLog)
if utilityPath is None:
utilityPath = self.DIST_BIN
if xrePath is None:
xrePath = self.DIST_BIN
if timeout == -1:
timeout = self.DEFAULT_TIMEOUT
self.utilityPath = utilityPath
cmd, args = self.buildCommandLine(app, debuggerInfo, profileDir, testURL, extraArgs)
startTime = datetime.datetime.now()
self.lastTestSeen = "remoteautomation.py"
self.launchApp([cmd] + args,
env=self.environment(env=env, crashreporter=not debuggerInfo),
e10s=e10s, **self.processArgs)
self.log.info("remoteautomation.py | Application pid: %d", self.pid)
status = self.waitForFinish(timeout, maxTime)
self.log.info("remoteautomation.py | Application ran for: %s",
str(datetime.datetime.now() - startTime))
crashed = self.checkForCrashes(symbolsPath)
if crashed:
status = 1
return status, self.lastTestSeen
# Set up what we need for the remote environment
def environment(self, env=None, crashreporter=True, **kwargs):
# Because we are running remote, we don't want to mimic the local env
# so no copying of os.environ
if env is None:
env = {}
if crashreporter:
env['MOZ_CRASHREPORTER_NO_REPORT'] = '1'
env['MOZ_CRASHREPORTER'] = '1'
env['MOZ_CRASHREPORTER_SHUTDOWN'] = '1'
else:
env['MOZ_CRASHREPORTER_DISABLE'] = '1'
# Crash on non-local network connections by default.
# MOZ_DISABLE_NONLOCAL_CONNECTIONS can be set to "0" to temporarily
# enable non-local connections for the purposes of local testing.
# Don't override the user's choice here. See bug 1049688.
env.setdefault('MOZ_DISABLE_NONLOCAL_CONNECTIONS', '1')
# Send an env var noting that we are in automation. Passing any
# value except the empty string will declare the value to exist.
#
# This may be used to disabled network connections during testing, e.g.
# Switchboard & telemetry uploads.
env.setdefault('MOZ_IN_AUTOMATION', '1')
# Set WebRTC logging in case it is not set yet.
# On Android, environment variables cannot contain ',' so the
# standard WebRTC setting for NSPR_LOG_MODULES is not available.
# env.setdefault('NSPR_LOG_MODULES', 'signaling:5,mtransport:5,datachannel:5,jsep:5,MediaPipelineFactory:5') # NOQA: E501
env.setdefault('R_LOG_LEVEL', '6')
env.setdefault('R_LOG_DESTINATION', 'stderr')
env.setdefault('R_LOG_VERBOSE', '1')
return env
def waitForFinish(self, timeout, maxTime):
""" Wait for tests to finish.
If maxTime seconds elapse or no output is detected for timeout
seconds, kill the process and fail the test.
"""
# maxTime is used to override the default timeout, we should honor that
status = self.wait(timeout=maxTime, noOutputTimeout=timeout)
topActivity = self.device.get_top_activity(timeout=60)
if topActivity == self.procName:
self.log.info("%s unexpectedly found running. Killing..." % self.procName)
self.kill(True)
if status == 1:
if maxTime:
self.log.error("TEST-UNEXPECTED-FAIL | %s | "
"application ran for longer than allowed maximum time "
"of %s seconds" % (self.lastTestSeen, maxTime))
else:
self.log.error("TEST-UNEXPECTED-FAIL | %s | "
"application ran for longer than allowed maximum time"
% self.lastTestSeen)
if status == 2:
self.log.error("TEST-UNEXPECTED-FAIL | %s | "
"application timed out after %d seconds with no output"
% (self.lastTestSeen, int(timeout)))
return status
def deleteANRs(self):
# Remove files from the dalvik stack-trace directory.
if not self.device.is_dir(self.device.stack_trace_dir, root=True):
return
try:
for trace_file in self.device.ls(self.device.stack_trace_dir, root=True):
trace_path = posixpath.join(self.device.stack_trace_dir, trace_file)
self.device.chmod(trace_path, root=True)
self.device.rm(trace_path, root=True)
except Exception as e:
print("Error deleting %s: %s" % (self.device.stack_trace_dir, str(e)))
def checkForANRs(self):
if not self.device.is_dir(self.device.stack_trace_dir):
print("%s not found" % self.device.stack_trace_dir)
return
try:
for trace_file in self.device.ls(self.device.stack_trace_dir, root=True):
trace_path = posixpath.join(self.device.stack_trace_dir, trace_file)
t = self.device.get_file(trace_path)
if t:
stripped = t.strip()
if len(stripped) > 0:
print("Contents of %s:" % trace_path)
print(t)
# Once reported, delete traces
self.deleteANRs()
except Exception as e:
print("Error pulling %s: %s" % (self.device.stack_trace_dir, str(e)))
def deleteTombstones(self):
# delete any tombstone files from device
self.device.rm("/data/tombstones", force=True, recursive=True, root=True)
def checkForTombstones(self):
# pull any tombstones from device and move to MOZ_UPLOAD_DIR
remoteDir = "/data/tombstones"
uploadDir = os.environ.get('MOZ_UPLOAD_DIR', None)
if uploadDir:
if not os.path.exists(uploadDir):
os.mkdir(uploadDir)
if self.device.is_dir(remoteDir):
# copy tombstone files from device to local upload directory
self.device.chmod(remoteDir, recursive=True, root=True)
self.device.pull(remoteDir, uploadDir)
self.deleteTombstones()
for f in glob.glob(os.path.join(uploadDir, "tombstone_??")):
# add a unique integer to the file name, in case there are
# multiple tombstones generated with the same name, for
# instance, after multiple robocop tests
for i in xrange(1, sys.maxint):
newname = "%s.%d.txt" % (f, i)
if not os.path.exists(newname):
os.rename(f, newname)
break
else:
print("%s does not exist; tombstone check skipped" % remoteDir)
else:
print("MOZ_UPLOAD_DIR not defined; tombstone check skipped")
def checkForCrashes(self, symbolsPath):
self.checkForANRs()
self.checkForTombstones()
logcat = self.device.get_logcat(
filter_out_regexps=fennecLogcatFilters)
javaException = mozcrash.check_for_java_exception(
logcat, test_name=self.lastTestSeen)
if javaException:
return True
# If crash reporting is disabled (MOZ_CRASHREPORTER!=1), we can't say
# anything.
if not self.CRASHREPORTER:
return False
try:
dumpDir = tempfile.mkdtemp()
remoteCrashDir = posixpath.join(self.remoteProfile, 'minidumps')
if not self.device.is_dir(remoteCrashDir):
# If crash reporting is enabled (MOZ_CRASHREPORTER=1), the
# minidumps directory is automatically created when Fennec
# (first) starts, so its lack of presence is a hint that
# something went wrong.
print("Automation Error: No crash directory (%s) found on remote device" %
remoteCrashDir)
return True
self.device.pull(remoteCrashDir, dumpDir)
logger = get_default_logger()
crashed = mozcrash.log_crashes(
logger, dumpDir, symbolsPath, test=self.lastTestSeen)
finally:
try:
shutil.rmtree(dumpDir)
except Exception as e:
print("WARNING: unable to remove directory %s: %s" % (
dumpDir, str(e)))
return crashed
def buildCommandLine(self, app, debuggerInfo, profileDir, testURL, extraArgs):
# If remote profile is specified, use that instead
if self.remoteProfile:
profileDir = self.remoteProfile
# Hack for robocop, if app is "am" and extraArgs contains the rest of the stuff, lets
# assume extraArgs is all we need
if app == "am" and extraArgs[0] in ('instrument', 'start'):
return app, extraArgs
cmd, args = Automation.buildCommandLine(
self, app, debuggerInfo, profileDir, testURL, extraArgs)
try:
args.remove('-foreground')
except Exception:
pass
return app, args
def launchApp(self, cmd, env=None, e10s=True, messageLogger=None, counts=None):
self.messageLogger = messageLogger
self.stdoutlen = 0
if self.appName and self.device.process_exist(self.appName):
print("remoteautomation.py %s is already running. Stopping..." % self.appName)
self.device.stop_application(self.appName, root=True)
self.counts = counts
if self.counts is not None:
self.counts['pass'] = 0
self.counts['fail'] = 0
self.counts['todo'] = 0
if cmd[0] == 'am':
cmd = ' '.join(cmd)
self.procName = self.appName
if not self.device.shell_bool(cmd):
print("remote_automation.py failed to launch %s" % cmd)
else:
self.procName = cmd[0].split(posixpath.sep)[-1]
args = cmd
if args[0] == self.appName:
args = args[1:]
url = args[-1:][0]
if url.startswith('/'):
# this is probably a reftest profile directory, not a url
url = None
else:
args = args[:-1]
if 'geckoview' in self.appName:
activity = "TestRunnerActivity"
self.device.launch_activity(self.appName, activity_name=activity, e10s=e10s,
moz_env=env, extra_args=args, url=url)
else:
self.device.launch_fennec(self.appName, moz_env=env, extra_args=args, url=url)
# Setting timeout at 1 hour since on a remote device this takes much longer.
# Temporarily increased to 110 minutes because no more chunks can be created.
self.timeout = 6600
# Used to buffer log messages until we meet a line break
self.logBuffer = ""
@property
def pid(self):
procs = self.device.get_process_list()
# limit the comparison to the first 75 characters due to a
# limitation in processname length in android.
pids = [proc[0] for proc in procs if proc[1] == self.procName[:75]]
if pids is None or len(pids) < 1:
return 0
return pids[0]
def read_stdout(self):
"""
Fetch the full remote log file, log any new content and return True if new
content processed.
"""
if not self.device.is_file(self.remoteLog):
return False
try:
newLogContent = self.device.get_file(self.remoteLog, offset=self.stdoutlen)
except ADBTimeoutError:
raise
except Exception as e:
self.log.info("remoteautomation.py | exception reading log: %s" % str(e))
return False
if not newLogContent:
return False
self.stdoutlen += len(newLogContent)
if self.messageLogger is None:
testStartFilenames = re.findall(r"TEST-START \| ([^\s]*)", newLogContent)
if testStartFilenames:
self.lastTestSeen = testStartFilenames[-1]
print(newLogContent)
return True
self.logBuffer += newLogContent
lines = self.logBuffer.split('\n')
lines = [l for l in lines if l]
if lines:
if self.logBuffer.endswith('\n'):
# all lines are complete; no need to buffer
self.logBuffer = ""
else:
# keep the last (unfinished) line in the buffer
self.logBuffer = lines[-1]
del lines[-1]
if not lines:
return False
for line in lines:
# This passes the line to the logger (to be logged or buffered)
parsed_messages = self.messageLogger.write(line)
for message in parsed_messages:
if isinstance(message, dict) and message.get('action') == 'test_start':
self.lastTestSeen = message['test']
if isinstance(message, dict) and message.get('action') == 'log':
line = message['message'].strip()
if self.counts:
m = re.match(".*:\s*(\d*)", line)
if m:
try:
val = int(m.group(1))
if "Passed:" in line:
self.counts['pass'] += val
elif "Failed:" in line:
self.counts['fail'] += val
elif "Todo:" in line:
self.counts['todo'] += val
except ADBTimeoutError:
raise
except Exception:
pass
return True
# Wait for the remote process to end (or for its activity to go to background).
# While waiting, periodically retrieve the process output and print it.
# If the process is still running after *timeout* seconds, return 1;
# If the process is still running but no output is received in *noOutputTimeout*
# seconds, return 2;
# Else, once the process exits/goes to background, return 0.
def wait(self, timeout=None, noOutputTimeout=None):
timer = 0
noOutputTimer = 0
interval = 10
if timeout is None:
timeout = self.timeout
status = 0
top = self.procName
slowLog = False
endTime = datetime.datetime.now() + datetime.timedelta(seconds=timeout)
while top == self.procName:
# Get log updates on each interval, but if it is taking
# too long, only do it every 60 seconds
hasOutput = False
if (not slowLog) or (timer % 60 == 0):
startRead = datetime.datetime.now()
hasOutput = self.read_stdout()
if (datetime.datetime.now() - startRead) > datetime.timedelta(seconds=5):
slowLog = True
if hasOutput:
noOutputTimer = 0
if self.counts and 'pass' in self.counts and self.counts['pass'] > 0:
interval = 0.5
time.sleep(interval)
timer += interval
noOutputTimer += interval
if datetime.datetime.now() > endTime:
status = 1
break
if (noOutputTimeout and noOutputTimer > noOutputTimeout):
status = 2
break
if not hasOutput:
top = self.device.get_top_activity(timeout=60)
if top is None:
print("Failed to get top activity, retrying, once...")
top = self.device.get_top_activity(timeout=60)
# Flush anything added to stdout during the sleep
self.read_stdout()
print("wait for %s complete; top activity=%s" % (self.procName, top))
return status
def kill(self, stagedShutdown=False):
# Take a screenshot to capture the screen state just before
# the application is killed.
# Do not use the on-device screenshot options since
# they rarely work well with Firefox on the Android
# emulator. dump_screen provides an effective
# screenshot of the emulator and its host desktop.
if not self.device._device_serial.startswith('emulator-'):
dump_device_screen(self.device, get_default_logger())
elif self.utilityPath:
dump_screen(self.utilityPath, get_default_logger())
if stagedShutdown:
# Trigger an ANR report with "kill -3" (SIGQUIT)
try:
self.device.pkill(self.procName, sig=3, attempts=1, root=True)
except ADBTimeoutError:
raise
except: # NOQA: E722
pass
time.sleep(3)
# Trigger a breakpad dump with "kill -6" (SIGABRT)
try:
self.device.pkill(self.procName, sig=6, attempts=1, root=True)
except ADBTimeoutError:
raise
except: # NOQA: E722
pass
# Wait for process to end
retries = 0
while retries < 3:
if self.device.process_exist(self.procName):
print("%s still alive after SIGABRT: waiting..." % self.procName)
time.sleep(5)
else:
break
retries += 1
if self.device.process_exist(self.procName):
try:
self.device.pkill(self.procName, sig=9, attempts=1, root=True)
except ADBTimeoutError:
raise
except: # NOQA: E722
print("%s still alive after SIGKILL!" % self.procName)
if self.device.process_exist(self.procName):
self.device.stop_application(self.procName)
else:
self.device.stop_application(self.procName)
# Test harnesses use the MOZ_CRASHREPORTER environment variables to suppress
# the interactive crash reporter, but that may not always be effective;
# check for and cleanup errant crashreporters.
crashreporter = "%s.CrashReporter" % self.procName
if self.device.process_exist(crashreporter):
print("Warning: %s unexpectedly found running. Killing..." % crashreporter)
try:
self.device.pkill(crashreporter, root=True)
except ADBTimeoutError:
raise
except: # NOQA: E722
pass
if self.device.process_exist(crashreporter):
print("ERROR: %s still running!!" % crashreporter)