mirror of
https://github.com/AsahiLinux/u-boot
synced 2024-11-16 17:58:23 +00:00
b0a928a15d
In pytest 3, runtestprotocol() may not call pytest_runtest_setup() if the test is skipped. That call is required to create a section for the test in the log file. If this is skipped, the call to log.end_section() at the tail of pytest_runtest_protocol() will throw an exception. This patch ensures that a log section always exists, both to avoid the exception and to ensure that a consistently structured log file is always created. Cc: Stefan Brüns <stefan.bruens@rwth-aachen.de> Reported-by: Stefan Brüns <stefan.bruens@rwth-aachen.de> Signed-off-by: Stephen Warren <swarren@nvidia.com> Tested-by: Tom Rini <trini@konsulko.com>
531 lines
16 KiB
Python
531 lines
16 KiB
Python
# Copyright (c) 2015 Stephen Warren
|
|
# Copyright (c) 2015-2016, NVIDIA CORPORATION. All rights reserved.
|
|
#
|
|
# SPDX-License-Identifier: GPL-2.0
|
|
|
|
# Implementation of pytest run-time hook functions. These are invoked by
|
|
# pytest at certain points during operation, e.g. startup, for each executed
|
|
# test, at shutdown etc. These hooks perform functions such as:
|
|
# - Parsing custom command-line options.
|
|
# - Pullilng in user-specified board configuration.
|
|
# - Creating the U-Boot console test fixture.
|
|
# - Creating the HTML log file.
|
|
# - Monitoring each test's results.
|
|
# - Implementing custom pytest markers.
|
|
|
|
import atexit
|
|
import errno
|
|
import os
|
|
import os.path
|
|
import pytest
|
|
from _pytest.runner import runtestprotocol
|
|
import ConfigParser
|
|
import re
|
|
import StringIO
|
|
import sys
|
|
|
|
# Globals: The HTML log file, and the connection to the U-Boot console.
|
|
log = None
|
|
console = None
|
|
|
|
def mkdir_p(path):
|
|
"""Create a directory path.
|
|
|
|
This includes creating any intermediate/parent directories. Any errors
|
|
caused due to already extant directories are ignored.
|
|
|
|
Args:
|
|
path: The directory path to create.
|
|
|
|
Returns:
|
|
Nothing.
|
|
"""
|
|
|
|
try:
|
|
os.makedirs(path)
|
|
except OSError as exc:
|
|
if exc.errno == errno.EEXIST and os.path.isdir(path):
|
|
pass
|
|
else:
|
|
raise
|
|
|
|
def pytest_addoption(parser):
|
|
"""pytest hook: Add custom command-line options to the cmdline parser.
|
|
|
|
Args:
|
|
parser: The pytest command-line parser.
|
|
|
|
Returns:
|
|
Nothing.
|
|
"""
|
|
|
|
parser.addoption('--build-dir', default=None,
|
|
help='U-Boot build directory (O=)')
|
|
parser.addoption('--result-dir', default=None,
|
|
help='U-Boot test result/tmp directory')
|
|
parser.addoption('--persistent-data-dir', default=None,
|
|
help='U-Boot test persistent generated data directory')
|
|
parser.addoption('--board-type', '--bd', '-B', default='sandbox',
|
|
help='U-Boot board type')
|
|
parser.addoption('--board-identity', '--id', default='na',
|
|
help='U-Boot board identity/instance')
|
|
parser.addoption('--build', default=False, action='store_true',
|
|
help='Compile U-Boot before running tests')
|
|
parser.addoption('--gdbserver', default=None,
|
|
help='Run sandbox under gdbserver. The argument is the channel '+
|
|
'over which gdbserver should communicate, e.g. localhost:1234')
|
|
|
|
def pytest_configure(config):
|
|
"""pytest hook: Perform custom initialization at startup time.
|
|
|
|
Args:
|
|
config: The pytest configuration.
|
|
|
|
Returns:
|
|
Nothing.
|
|
"""
|
|
|
|
global log
|
|
global console
|
|
global ubconfig
|
|
|
|
test_py_dir = os.path.dirname(os.path.abspath(__file__))
|
|
source_dir = os.path.dirname(os.path.dirname(test_py_dir))
|
|
|
|
board_type = config.getoption('board_type')
|
|
board_type_filename = board_type.replace('-', '_')
|
|
|
|
board_identity = config.getoption('board_identity')
|
|
board_identity_filename = board_identity.replace('-', '_')
|
|
|
|
build_dir = config.getoption('build_dir')
|
|
if not build_dir:
|
|
build_dir = source_dir + '/build-' + board_type
|
|
mkdir_p(build_dir)
|
|
|
|
result_dir = config.getoption('result_dir')
|
|
if not result_dir:
|
|
result_dir = build_dir
|
|
mkdir_p(result_dir)
|
|
|
|
persistent_data_dir = config.getoption('persistent_data_dir')
|
|
if not persistent_data_dir:
|
|
persistent_data_dir = build_dir + '/persistent-data'
|
|
mkdir_p(persistent_data_dir)
|
|
|
|
gdbserver = config.getoption('gdbserver')
|
|
if gdbserver and board_type != 'sandbox':
|
|
raise Exception('--gdbserver only supported with sandbox')
|
|
|
|
import multiplexed_log
|
|
log = multiplexed_log.Logfile(result_dir + '/test-log.html')
|
|
|
|
if config.getoption('build'):
|
|
if build_dir != source_dir:
|
|
o_opt = 'O=%s' % build_dir
|
|
else:
|
|
o_opt = ''
|
|
cmds = (
|
|
['make', o_opt, '-s', board_type + '_defconfig'],
|
|
['make', o_opt, '-s', '-j8'],
|
|
)
|
|
with log.section('make'):
|
|
runner = log.get_runner('make', sys.stdout)
|
|
for cmd in cmds:
|
|
runner.run(cmd, cwd=source_dir)
|
|
runner.close()
|
|
log.status_pass('OK')
|
|
|
|
class ArbitraryAttributeContainer(object):
|
|
pass
|
|
|
|
ubconfig = ArbitraryAttributeContainer()
|
|
ubconfig.brd = dict()
|
|
ubconfig.env = dict()
|
|
|
|
modules = [
|
|
(ubconfig.brd, 'u_boot_board_' + board_type_filename),
|
|
(ubconfig.env, 'u_boot_boardenv_' + board_type_filename),
|
|
(ubconfig.env, 'u_boot_boardenv_' + board_type_filename + '_' +
|
|
board_identity_filename),
|
|
]
|
|
for (dict_to_fill, module_name) in modules:
|
|
try:
|
|
module = __import__(module_name)
|
|
except ImportError:
|
|
continue
|
|
dict_to_fill.update(module.__dict__)
|
|
|
|
ubconfig.buildconfig = dict()
|
|
|
|
for conf_file in ('.config', 'include/autoconf.mk'):
|
|
dot_config = build_dir + '/' + conf_file
|
|
if not os.path.exists(dot_config):
|
|
raise Exception(conf_file + ' does not exist; ' +
|
|
'try passing --build option?')
|
|
|
|
with open(dot_config, 'rt') as f:
|
|
ini_str = '[root]\n' + f.read()
|
|
ini_sio = StringIO.StringIO(ini_str)
|
|
parser = ConfigParser.RawConfigParser()
|
|
parser.readfp(ini_sio)
|
|
ubconfig.buildconfig.update(parser.items('root'))
|
|
|
|
ubconfig.test_py_dir = test_py_dir
|
|
ubconfig.source_dir = source_dir
|
|
ubconfig.build_dir = build_dir
|
|
ubconfig.result_dir = result_dir
|
|
ubconfig.persistent_data_dir = persistent_data_dir
|
|
ubconfig.board_type = board_type
|
|
ubconfig.board_identity = board_identity
|
|
ubconfig.gdbserver = gdbserver
|
|
ubconfig.dtb = build_dir + '/arch/sandbox/dts/test.dtb'
|
|
|
|
env_vars = (
|
|
'board_type',
|
|
'board_identity',
|
|
'source_dir',
|
|
'test_py_dir',
|
|
'build_dir',
|
|
'result_dir',
|
|
'persistent_data_dir',
|
|
)
|
|
for v in env_vars:
|
|
os.environ['U_BOOT_' + v.upper()] = getattr(ubconfig, v)
|
|
|
|
if board_type.startswith('sandbox'):
|
|
import u_boot_console_sandbox
|
|
console = u_boot_console_sandbox.ConsoleSandbox(log, ubconfig)
|
|
else:
|
|
import u_boot_console_exec_attach
|
|
console = u_boot_console_exec_attach.ConsoleExecAttach(log, ubconfig)
|
|
|
|
re_ut_test_list = re.compile(r'_u_boot_list_2_(dm|env)_test_2_\1_test_(.*)\s*$')
|
|
def generate_ut_subtest(metafunc, fixture_name):
|
|
"""Provide parametrization for a ut_subtest fixture.
|
|
|
|
Determines the set of unit tests built into a U-Boot binary by parsing the
|
|
list of symbols generated by the build process. Provides this information
|
|
to test functions by parameterizing their ut_subtest fixture parameter.
|
|
|
|
Args:
|
|
metafunc: The pytest test function.
|
|
fixture_name: The fixture name to test.
|
|
|
|
Returns:
|
|
Nothing.
|
|
"""
|
|
|
|
fn = console.config.build_dir + '/u-boot.sym'
|
|
try:
|
|
with open(fn, 'rt') as f:
|
|
lines = f.readlines()
|
|
except:
|
|
lines = []
|
|
lines.sort()
|
|
|
|
vals = []
|
|
for l in lines:
|
|
m = re_ut_test_list.search(l)
|
|
if not m:
|
|
continue
|
|
vals.append(m.group(1) + ' ' + m.group(2))
|
|
|
|
ids = ['ut_' + s.replace(' ', '_') for s in vals]
|
|
metafunc.parametrize(fixture_name, vals, ids=ids)
|
|
|
|
def generate_config(metafunc, fixture_name):
|
|
"""Provide parametrization for {env,brd}__ fixtures.
|
|
|
|
If a test function takes parameter(s) (fixture names) of the form brd__xxx
|
|
or env__xxx, the brd and env configuration dictionaries are consulted to
|
|
find the list of values to use for those parameters, and the test is
|
|
parametrized so that it runs once for each combination of values.
|
|
|
|
Args:
|
|
metafunc: The pytest test function.
|
|
fixture_name: The fixture name to test.
|
|
|
|
Returns:
|
|
Nothing.
|
|
"""
|
|
|
|
subconfigs = {
|
|
'brd': console.config.brd,
|
|
'env': console.config.env,
|
|
}
|
|
parts = fixture_name.split('__')
|
|
if len(parts) < 2:
|
|
return
|
|
if parts[0] not in subconfigs:
|
|
return
|
|
subconfig = subconfigs[parts[0]]
|
|
vals = []
|
|
val = subconfig.get(fixture_name, [])
|
|
# If that exact name is a key in the data source:
|
|
if val:
|
|
# ... use the dict value as a single parameter value.
|
|
vals = (val, )
|
|
else:
|
|
# ... otherwise, see if there's a key that contains a list of
|
|
# values to use instead.
|
|
vals = subconfig.get(fixture_name+ 's', [])
|
|
def fixture_id(index, val):
|
|
try:
|
|
return val['fixture_id']
|
|
except:
|
|
return fixture_name + str(index)
|
|
ids = [fixture_id(index, val) for (index, val) in enumerate(vals)]
|
|
metafunc.parametrize(fixture_name, vals, ids=ids)
|
|
|
|
def pytest_generate_tests(metafunc):
|
|
"""pytest hook: parameterize test functions based on custom rules.
|
|
|
|
Check each test function parameter (fixture name) to see if it is one of
|
|
our custom names, and if so, provide the correct parametrization for that
|
|
parameter.
|
|
|
|
Args:
|
|
metafunc: The pytest test function.
|
|
|
|
Returns:
|
|
Nothing.
|
|
"""
|
|
|
|
for fn in metafunc.fixturenames:
|
|
if fn == 'ut_subtest':
|
|
generate_ut_subtest(metafunc, fn)
|
|
continue
|
|
generate_config(metafunc, fn)
|
|
|
|
@pytest.fixture(scope='function')
|
|
def u_boot_console(request):
|
|
"""Generate the value of a test's u_boot_console fixture.
|
|
|
|
Args:
|
|
request: The pytest request.
|
|
|
|
Returns:
|
|
The fixture value.
|
|
"""
|
|
|
|
console.ensure_spawned()
|
|
return console
|
|
|
|
anchors = {}
|
|
tests_not_run = []
|
|
tests_failed = []
|
|
tests_xpassed = []
|
|
tests_xfailed = []
|
|
tests_skipped = []
|
|
tests_passed = []
|
|
|
|
def pytest_itemcollected(item):
|
|
"""pytest hook: Called once for each test found during collection.
|
|
|
|
This enables our custom result analysis code to see the list of all tests
|
|
that should eventually be run.
|
|
|
|
Args:
|
|
item: The item that was collected.
|
|
|
|
Returns:
|
|
Nothing.
|
|
"""
|
|
|
|
tests_not_run.append(item.name)
|
|
|
|
def cleanup():
|
|
"""Clean up all global state.
|
|
|
|
Executed (via atexit) once the entire test process is complete. This
|
|
includes logging the status of all tests, and the identity of any failed
|
|
or skipped tests.
|
|
|
|
Args:
|
|
None.
|
|
|
|
Returns:
|
|
Nothing.
|
|
"""
|
|
|
|
if console:
|
|
console.close()
|
|
if log:
|
|
with log.section('Status Report', 'status_report'):
|
|
log.status_pass('%d passed' % len(tests_passed))
|
|
if tests_skipped:
|
|
log.status_skipped('%d skipped' % len(tests_skipped))
|
|
for test in tests_skipped:
|
|
anchor = anchors.get(test, None)
|
|
log.status_skipped('... ' + test, anchor)
|
|
if tests_xpassed:
|
|
log.status_xpass('%d xpass' % len(tests_xpassed))
|
|
for test in tests_xpassed:
|
|
anchor = anchors.get(test, None)
|
|
log.status_xpass('... ' + test, anchor)
|
|
if tests_xfailed:
|
|
log.status_xfail('%d xfail' % len(tests_xfailed))
|
|
for test in tests_xfailed:
|
|
anchor = anchors.get(test, None)
|
|
log.status_xfail('... ' + test, anchor)
|
|
if tests_failed:
|
|
log.status_fail('%d failed' % len(tests_failed))
|
|
for test in tests_failed:
|
|
anchor = anchors.get(test, None)
|
|
log.status_fail('... ' + test, anchor)
|
|
if tests_not_run:
|
|
log.status_fail('%d not run' % len(tests_not_run))
|
|
for test in tests_not_run:
|
|
anchor = anchors.get(test, None)
|
|
log.status_fail('... ' + test, anchor)
|
|
log.close()
|
|
atexit.register(cleanup)
|
|
|
|
def setup_boardspec(item):
|
|
"""Process any 'boardspec' marker for a test.
|
|
|
|
Such a marker lists the set of board types that a test does/doesn't
|
|
support. If tests are being executed on an unsupported board, the test is
|
|
marked to be skipped.
|
|
|
|
Args:
|
|
item: The pytest test item.
|
|
|
|
Returns:
|
|
Nothing.
|
|
"""
|
|
|
|
mark = item.get_marker('boardspec')
|
|
if not mark:
|
|
return
|
|
required_boards = []
|
|
for board in mark.args:
|
|
if board.startswith('!'):
|
|
if ubconfig.board_type == board[1:]:
|
|
pytest.skip('board not supported')
|
|
return
|
|
else:
|
|
required_boards.append(board)
|
|
if required_boards and ubconfig.board_type not in required_boards:
|
|
pytest.skip('board not supported')
|
|
|
|
def setup_buildconfigspec(item):
|
|
"""Process any 'buildconfigspec' marker for a test.
|
|
|
|
Such a marker lists some U-Boot configuration feature that the test
|
|
requires. If tests are being executed on an U-Boot build that doesn't
|
|
have the required feature, the test is marked to be skipped.
|
|
|
|
Args:
|
|
item: The pytest test item.
|
|
|
|
Returns:
|
|
Nothing.
|
|
"""
|
|
|
|
mark = item.get_marker('buildconfigspec')
|
|
if not mark:
|
|
return
|
|
for option in mark.args:
|
|
if not ubconfig.buildconfig.get('config_' + option.lower(), None):
|
|
pytest.skip('.config feature not enabled')
|
|
|
|
def start_test_section(item):
|
|
anchors[item.name] = log.start_section(item.name)
|
|
|
|
def pytest_runtest_setup(item):
|
|
"""pytest hook: Configure (set up) a test item.
|
|
|
|
Called once for each test to perform any custom configuration. This hook
|
|
is used to skip the test if certain conditions apply.
|
|
|
|
Args:
|
|
item: The pytest test item.
|
|
|
|
Returns:
|
|
Nothing.
|
|
"""
|
|
|
|
start_test_section(item)
|
|
setup_boardspec(item)
|
|
setup_buildconfigspec(item)
|
|
|
|
def pytest_runtest_protocol(item, nextitem):
|
|
"""pytest hook: Called to execute a test.
|
|
|
|
This hook wraps the standard pytest runtestprotocol() function in order
|
|
to acquire visibility into, and record, each test function's result.
|
|
|
|
Args:
|
|
item: The pytest test item to execute.
|
|
nextitem: The pytest test item that will be executed after this one.
|
|
|
|
Returns:
|
|
A list of pytest reports (test result data).
|
|
"""
|
|
|
|
reports = runtestprotocol(item, nextitem=nextitem)
|
|
|
|
# In pytest 3, runtestprotocol() may not call pytest_runtest_setup() if
|
|
# the test is skipped. That call is required to create the test's section
|
|
# in the log file. The call to log.end_section() requires that the log
|
|
# contain a section for this test. Create a section for the test if it
|
|
# doesn't already exist.
|
|
if not item.name in anchors:
|
|
start_test_section(item)
|
|
|
|
failure_cleanup = False
|
|
test_list = tests_passed
|
|
msg = 'OK'
|
|
msg_log = log.status_pass
|
|
for report in reports:
|
|
if report.outcome == 'failed':
|
|
if hasattr(report, 'wasxfail'):
|
|
test_list = tests_xpassed
|
|
msg = 'XPASSED'
|
|
msg_log = log.status_xpass
|
|
else:
|
|
failure_cleanup = True
|
|
test_list = tests_failed
|
|
msg = 'FAILED:\n' + str(report.longrepr)
|
|
msg_log = log.status_fail
|
|
break
|
|
if report.outcome == 'skipped':
|
|
if hasattr(report, 'wasxfail'):
|
|
failure_cleanup = True
|
|
test_list = tests_xfailed
|
|
msg = 'XFAILED:\n' + str(report.longrepr)
|
|
msg_log = log.status_xfail
|
|
break
|
|
test_list = tests_skipped
|
|
msg = 'SKIPPED:\n' + str(report.longrepr)
|
|
msg_log = log.status_skipped
|
|
|
|
if failure_cleanup:
|
|
console.drain_console()
|
|
|
|
test_list.append(item.name)
|
|
tests_not_run.remove(item.name)
|
|
|
|
try:
|
|
msg_log(msg)
|
|
except:
|
|
# If something went wrong with logging, it's better to let the test
|
|
# process continue, which may report other exceptions that triggered
|
|
# the logging issue (e.g. console.log wasn't created). Hence, just
|
|
# squash the exception. If the test setup failed due to e.g. syntax
|
|
# error somewhere else, this won't be seen. However, once that issue
|
|
# is fixed, if this exception still exists, it will then be logged as
|
|
# part of the test's stdout.
|
|
import traceback
|
|
print 'Exception occurred while logging runtest status:'
|
|
traceback.print_exc()
|
|
# FIXME: Can we force a test failure here?
|
|
|
|
log.end_section(item.name)
|
|
|
|
if failure_cleanup:
|
|
console.cleanup_spawn()
|
|
|
|
return reports
|