xref: /gem5/src/systemc/tests/verify.py

#!/usr/bin/env python2.7
#
# Copyright 2018 Google, Inc.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are
# met: redistributions of source code must retain the above copyright
# notice, this list of conditions and the following disclaimer;
# redistributions in binary form must reproduce the above copyright
# notice, this list of conditions and the following disclaimer in the
# documentation and/or other materials provided with the distribution;
# neither the name of the copyright holders nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
#
# Authors: Gabe Black

from __future__ import print_function

import argparse
import collections
import difflib
import functools
import inspect
import itertools
import json
import multiprocessing.pool
import os
import re
import subprocess
import sys

script_path = os.path.abspath(inspect.getfile(inspect.currentframe()))
script_dir = os.path.dirname(script_path)
config_path = os.path.join(script_dir, 'config.py')
# Parent directories if checked out as part of gem5.
systemc_dir = os.path.dirname(script_dir)
src_dir = os.path.dirname(systemc_dir)
checkout_dir = os.path.dirname(src_dir)

systemc_rel_path = 'systemc'
tests_rel_path = os.path.join(systemc_rel_path, 'tests')
json_rel_path = os.path.join(tests_rel_path, 'tests.json')



def scons(*args):
    args = ['scons'] + list(args)
    subprocess.check_call(args)



class Test(object):
    def __init__(self, target, suffix, build_dir, props):
        self.target = target
        self.suffix = suffix
        self.build_dir = build_dir
        self.props = {}

        for key, val in props.iteritems():
            self.set_prop(key, val)

    def set_prop(self, key, val):
        setattr(self, key, val)
        self.props[key] = val

    def dir(self):
        return os.path.join(self.build_dir, tests_rel_path, self.path)

    def src_dir(self):
        return os.path.join(script_dir, self.path)

    def expected_returncode_file(self):
        return os.path.join(self.src_dir(), 'expected_returncode')

    def golden_dir(self):
        return os.path.join(self.src_dir(), 'golden')

    def bin(self):
        return '.'.join([self.name, self.suffix])

    def full_path(self):
        return os.path.join(self.dir(), self.bin())

    def m5out_dir(self):
        return os.path.join(self.dir(), 'm5out.' + self.suffix)

    def returncode_file(self):
        return os.path.join(self.m5out_dir(), 'returncode')



test_phase_classes = {}

class TestPhaseMeta(type):
    def __init__(cls, name, bases, d):
        if not d.pop('abstract', False):
            test_phase_classes[d['name']] = cls

        super(TestPhaseMeta, cls).__init__(name, bases, d)

class TestPhaseBase(object):
    __metaclass__ = TestPhaseMeta
    abstract = True

    def __init__(self, main_args, *args):
        self.main_args = main_args
        self.args = args

    def __lt__(self, other):
        return self.number < other.number

class CompilePhase(TestPhaseBase):
    name = 'compile'
    number = 1

    def run(self, tests):
        targets = list([test.full_path() for test in tests])

        parser = argparse.ArgumentParser()
        parser.add_argument('-j', type=int, default=0)
        args, leftovers = parser.parse_known_args(self.args)
        if args.j == 0:
            self.args = ('-j', str(self.main_args.j)) + self.args

        scons_args = [ '--directory', self.main_args.scons_dir,
                       'USE_SYSTEMC=1' ] + list(self.args) + targets
        scons(*scons_args)

class RunPhase(TestPhaseBase):
    name = 'execute'
    number = 2

    def run(self, tests):
        parser = argparse.ArgumentParser()
        parser.add_argument('--timeout', type=int, metavar='SECONDS',
                            help='Time limit for each run in seconds, '
                            '0 to disable.',
                            default=60)
        parser.add_argument('-j', type=int, default=0,
                help='How many tests to run in parallel.')
        args = parser.parse_args(self.args)

        timeout_cmd = [
            'timeout',
            '--kill-after', str(args.timeout * 2),
            str(args.timeout)
        ]
        def run_test(test):
            cmd = []
            if args.timeout:
                cmd.extend(timeout_cmd)
            cmd.extend([
                os.path.abspath(test.full_path()),
                '-rd', os.path.abspath(test.m5out_dir()),
                '--listener-mode=off',
                '--quiet',
                os.path.abspath(config_path),
            ])
            # Ensure the output directory exists.
            if not os.path.exists(test.m5out_dir()):
                os.makedirs(test.m5out_dir())
            try:
                subprocess.check_call(cmd, cwd=os.path.dirname(test.dir()))
            except subprocess.CalledProcessError, error:
                returncode = error.returncode
            else:
                returncode = 0
            with open(test.returncode_file(), 'w') as rc:
                rc.write('%d\n' % returncode)

        j = self.main_args.j if args.j == 0 else args.j

        runnable = filter(lambda t: not t.compile_only, tests)
        if j == 1:
            map(run_test, runnable)
        else:
            tp = multiprocessing.pool.ThreadPool(j)
            map(lambda t: tp.apply_async(run_test, (t,)), runnable)
            tp.close()
            tp.join()

class Checker(object):
    def __init__(self, ref, test, tag):
        self.ref = ref
        self.test = test
        self.tag = tag

    def check(self):
        with open(self.test) as test_f, open(self.ref) as ref_f:
            return test_f.read() == ref_f.read()

def tagged_filt(tag, num):
    return (r'\n{}: \({}{}\) .*\n(In file: .*\n)?'
            r'(In process: [\w.]* @ .*\n)?').format(tag, tag[0], num)

def error_filt(num):
    return tagged_filt('Error', num)

def warning_filt(num):
    return tagged_filt('Warning', num)

def info_filt(num):
    return tagged_filt('Info', num)

class DiffingChecker(Checker):
    def __init__(self, ref, test, tag, out_dir):
        super(DiffingChecker, self).__init__(ref, test, tag)
        self.out_dir = out_dir

    def diffing_check(self, ref_lines, test_lines):
        test_file = os.path.basename(self.test)
        ref_file = os.path.basename(self.ref)

        diff_file = '.'.join([ref_file, 'diff'])
        diff_path = os.path.join(self.out_dir, diff_file)
        if test_lines != ref_lines:
            with open(diff_path, 'w') as diff_f:
                for line in difflib.unified_diff(
                        ref_lines, test_lines,
                        fromfile=ref_file,
                        tofile=test_file):
                    diff_f.write(line)
            return False
        else:
            if os.path.exists(diff_path):
                os.unlink(diff_path)
            return True

class LogChecker(DiffingChecker):
    def merge_filts(*filts):
        filts = map(lambda f: '(' + f + ')', filts)
        filts = '|'.join(filts)
        return re.compile(filts, flags=re.MULTILINE)

    # The reporting mechanism will print the actual filename when running in
    # gem5, and the "golden" output will say "<removed by verify.py>". We want
    # to strip out both versions to make comparing the output sensible.
    in_file_filt = r'^In file: ((<removed by verify\.pl>)|([a-zA-Z0-9.:_/]*))$'

    ref_filt = merge_filts(
        r'^\nInfo: /OSCI/SystemC: Simulation stopped by user.\n',
        r'^SystemC Simulation\n',
        r'^\nInfo: \(I804\) /IEEE_Std_1666/deprecated: ' +
        r'You can turn off(.*\n){7}',
        r'^\nInfo: \(I804\) /IEEE_Std_1666/deprecated: \n' +
        r'    sc_clock\(const char(.*\n){3}',
        warning_filt(540),
        warning_filt(571),
        info_filt(804),
        info_filt(704),
        in_file_filt,
    )
    test_filt = merge_filts(
        r'^Global frequency set at \d* ticks per second\n',
        r'^info: Entering event queue @ \d*\.  Starting simulation\.\.\.\n',
        r'warn: Ignoring request to set stack size\.\n',
        info_filt(804),
        in_file_filt,
    )

    def apply_filters(self, data, filts):
        re.sub(filt, '', data)

    def check(self):
        with open(self.test) as test_f, open(self.ref) as ref_f:
            test = re.sub(self.test_filt, '', test_f.read())
            ref = re.sub(self.ref_filt, '', ref_f.read())
            return self.diffing_check(ref.splitlines(True),
                                      test.splitlines(True))

class VcdChecker(DiffingChecker):
    def check(self):
        with open (self.test) as test_f, open(self.ref) as ref_f:
            ref = ref_f.read().splitlines(True)
            test = test_f.read().splitlines(True)
            # Strip off the first seven lines of the test output which are
            # date and version information.
            test = test[7:]

            return self.diffing_check(ref, test)

class GoldenDir(object):
    def __init__(self, path, platform):
        self.path = path
        self.platform = platform

        contents = os.listdir(path)
        suffix = '.' + platform
        suffixed = filter(lambda c: c.endswith(suffix), contents)
        bases = map(lambda t: t[:-len(platform)], suffixed)
        common = filter(lambda t: not t.startswith(tuple(bases)), contents)

        self.entries = {}
        class Entry(object):
            def __init__(self, e_path):
                self.used = False
                self.path = os.path.join(path, e_path)

            def use(self):
                self.used = True

        for entry in contents:
            self.entries[entry] = Entry(entry)

    def entry(self, name):
        def match(n):
            return (n == name) or n.startswith(name + '.')
        matches = { n: e for n, e in self.entries.items() if match(n) }

        for match in matches.values():
            match.use()

        platform_name = '.'.join([ name, self.platform ])
        if platform_name in matches:
            return matches[platform_name].path
        if name in matches:
            return matches[name].path
        else:
            return None

    def unused(self):
        items = self.entries.items()
        items = filter(lambda i: not i[1].used, items)

        items.sort()
        sources = []
        i = 0
        while i < len(items):
            root = items[i][0]
            sources.append(root)
            i += 1
            while i < len(items) and items[i][0].startswith(root):
                i += 1
        return sources

class VerifyPhase(TestPhaseBase):
    name = 'verify'
    number = 3

    def reset_status(self):
        self._passed = []
        self._failed = {}

    def passed(self, test):
        self._passed.append(test)

    def failed(self, test, cause, note=''):
        test.set_prop('note', note)
        self._failed.setdefault(cause, []).append(test)

    def print_status(self):
        total_passed = len(self._passed)
        total_failed = sum(map(len, self._failed.values()))
        print()
        print('Passed: {passed:4} - Failed: {failed:4}'.format(
                  passed=total_passed, failed=total_failed))

    def write_result_file(self, path):
        results = {
            'passed': map(lambda t: t.props, self._passed),
            'failed': {
                cause: map(lambda t: t.props, tests) for
                       cause, tests in self._failed.iteritems()
            }
        }
        with open(path, 'w') as rf:
            json.dump(results, rf)

    def print_results(self):
        print()
        print('Passed:')
        for path in sorted(list([ t.path for t in self._passed ])):
            print('    ', path)

        print()
        print('Failed:')

        causes = []
        for cause, tests in sorted(self._failed.items()):
            block = '  ' + cause.capitalize() + ':\n'
            for test in sorted(tests, key=lambda t: t.path):
                block += '    ' + test.path
                if test.note:
                    block += ' - ' + test.note
                block += '\n'
            causes.append(block)

        print('\n'.join(causes))

    def run(self, tests):
        parser = argparse.ArgumentParser()
        result_opts = parser.add_mutually_exclusive_group()
        result_opts.add_argument('--result-file', action='store_true',
                help='Create a results.json file in the current directory.')
        result_opts.add_argument('--result-file-at', metavar='PATH',
                help='Create a results json file at the given path.')
        parser.add_argument('--no-print-results', action='store_true',
                help='Don\'t print a list of tests that passed or failed')
        args = parser.parse_args(self.args)

        self.reset_status()

        runnable = filter(lambda t: not t.compile_only, tests)
        compile_only = filter(lambda t: t.compile_only, tests)

        for test in compile_only:
            if os.path.exists(test.full_path()):
                self.passed(test)
            else:
                self.failed(test, 'compile failed')

        for test in runnable:
            with open(test.returncode_file()) as rc:
                returncode = int(rc.read())

            expected_returncode = 0
            if os.path.exists(test.expected_returncode_file()):
                with open(test.expected_returncode_file()) as erc:
                    expected_returncode = int(erc.read())

            if returncode == 124:
                self.failed(test, 'time out')
                continue
            elif returncode != expected_returncode:
                if expected_returncode == 0:
                    self.failed(test, 'abort')
                else:
                    self.failed(test, 'missed abort')
                continue

            out_dir = test.m5out_dir()

            Diff = collections.namedtuple(
                    'Diff', 'ref, test, tag, ref_filter')

            diffs = []

            gd = GoldenDir(test.golden_dir(), 'linux64')

            missing = []
            log_file = '.'.join([test.name, 'log'])
            log_path = gd.entry(log_file)
            simout_path = os.path.join(out_dir, 'simout')
            if not os.path.exists(simout_path):
                missing.append('log output')
            elif log_path:
                diffs.append(LogChecker(log_path, simout_path,
                                        log_file, out_dir))

            for name in gd.unused():
                test_path = os.path.join(out_dir, name)
                ref_path = gd.entry(name)
                if not os.path.exists(test_path):
                    missing.append(name)
                elif name.endswith('.vcd'):
                    diffs.append(VcdChecker(ref_path, test_path,
                                            name, out_dir))
                else:
                    diffs.append(Checker(ref_path, test_path, name))

            if missing:
                self.failed(test, 'missing output', ' '.join(missing))
                continue

            failed_diffs = filter(lambda d: not d.check(), diffs)
            if failed_diffs:
                tags = map(lambda d: d.tag, failed_diffs)
                self.failed(test, 'failed diffs', ' '.join(tags))
                continue

            self.passed(test)

        if not args.no_print_results:
            self.print_results()

        self.print_status()

        result_path = None
        if args.result_file:
            result_path = os.path.join(os.getcwd(), 'results.json')
        elif args.result_file_at:
            result_path = args.result_file_at

        if result_path:
            self.write_result_file(result_path)


parser = argparse.ArgumentParser(description='SystemC test utility')

parser.add_argument('build_dir', metavar='BUILD_DIR',
                    help='The build directory (ie. build/ARM).')

parser.add_argument('--update-json', action='store_true',
                    help='Update the json manifest of tests.')

parser.add_argument('--flavor', choices=['debug', 'opt', 'fast'],
                    default='opt',
                    help='Flavor of binary to test.')

parser.add_argument('--list', action='store_true',
                    help='List the available tests')

parser.add_argument('-j', type=int, default=1,
                    help='Default level of parallelism, can be overriden '
                    'for individual stages')

parser.add_argument('-C', '--scons-dir', metavar='SCONS_DIR',
                    default=checkout_dir,
                    help='Directory to run scons from')

filter_opts = parser.add_mutually_exclusive_group()
filter_opts.add_argument('--filter', default='True',
                         help='Python expression which filters tests based '
                         'on their properties')
filter_opts.add_argument('--filter-file', default=None,
                         type=argparse.FileType('r'),
                         help='Same as --filter, but read from a file')

def collect_phases(args):
    phase_groups = [list(g) for k, g in
                    itertools.groupby(args, lambda x: x != '--phase') if k]
    main_args = parser.parse_args(phase_groups[0][1:])
    phases = []
    names = []
    for group in phase_groups[1:]:
        name = group[0]
        if name in names:
            raise RuntimeException('Phase %s specified more than once' % name)
        phase = test_phase_classes[name]
        phases.append(phase(main_args, *group[1:]))
    phases.sort()
    return main_args, phases

main_args, phases = collect_phases(sys.argv)

if len(phases) == 0:
    phases = [
        CompilePhase(main_args),
        RunPhase(main_args),
        VerifyPhase(main_args)
    ]



json_path = os.path.join(main_args.build_dir, json_rel_path)

if main_args.update_json:
    scons('--directory', main_args.scons_dir, os.path.join(json_path))

with open(json_path) as f:
    test_data = json.load(f)

    if main_args.filter_file:
        f = main_args.filter_file
        filt = compile(f.read(), f.name, 'eval')
    else:
        filt = compile(main_args.filter, '<string>', 'eval')

    filtered_tests = {
        target: props for (target, props) in
                    test_data.iteritems() if eval(filt, dict(props))
    }

    if len(filtered_tests) == 0:
        print('All tests were filtered out.')
        exit()

    if main_args.list:
        for target, props in sorted(filtered_tests.iteritems()):
            print('%s.%s' % (target, main_args.flavor))
            for key, val in props.iteritems():
                print('    %s: %s' % (key, val))
        print('Total tests: %d' % len(filtered_tests))
    else:
        tests_to_run = list([
            Test(target, main_args.flavor, main_args.build_dir, props) for
                target, props in sorted(filtered_tests.iteritems())
        ])

        for phase in phases:
            phase.run(tests_to_run)