workload-automation/wa/workloads/lmbench/__init__.py

#    Copyright 2015-2018 ARM Limited
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#

# pylint: disable=E1101,W0201

import os

from wa import Workload, Parameter, Executable
from wa.utils.types import list_or_integer, list_or_string
from wa.utils.types import cpu_mask


class Lmbench(Workload):

    name = 'lmbench'

    # Define supported tests. Each requires a _setup_{name} routine below
    test_names = ['lat_mem_rd', 'bw_mem']

    description = """
    Run a subtest from lmbench, a suite of portable ANSI/C microbenchmarks for
    UNIX/POSIX.
    In general, lmbench measures two key features: latency and bandwidth. This
    workload supports a subset of lmbench tests.  lat_mem_rd can be used to
    measure latencies to memory (including caches). bw_mem can be used to
    measure bandwidth to/from memory over a range of operations.
    Further details, and source code are available from:

        http://sourceforge.net/projects/lmbench/.

    See lmbench/bin/README for license details.
    """

    parameters = [
        Parameter('test', default='lat_mem_rd', allowed_values=test_names,
                  description='''
                  Specifies an lmbench test to run.
                  '''),
        Parameter('stride', kind=list_or_integer, default=[128],
                  description='''
                  Stride for lat_mem_rd test. Workload will iterate over one or
                  more integer values.
                  '''),
        Parameter('thrash', kind=bool, default=True,
                  description='Sets -t flag for lat_mem_rd_test'),
        Parameter('size', kind=list_or_string, default="4m",
                  description='Data set size for lat_mem_rd bw_mem tests.'),
        Parameter('mem_category', kind=list_or_string,
                  default=('rd', 'wr', 'cp', 'frd', 'fwr', 'fcp', 'bzero', 'bcopy'),
                  description='List of memory catetories for bw_mem test.'),
        Parameter('parallelism', kind=int, default=None,
                  description='Parallelism flag for tests that accept it.'),
        Parameter('warmup', kind=int, default=None,
                  description='Warmup flag for tests that accept it.'),
        Parameter('repetitions', kind=int, default=None,
                  description='Repetitions flag for tests that accept it.'),
        Parameter('force_abi', kind=str, default=None,
                  description='''
                  Override device abi with this value. Can be used to force
                  arm32 on 64-bit devices.
                  '''),
        Parameter('run_timeout', kind=int, default=900,
                  description="""
                  Timeout for execution of the test.
                  """),
        Parameter('loops', kind=int, default=1, constraint=lambda x: x > 0,
                  description="""
                  Specifies the number of times the benchmark will be run in a
                  "tight loop", i.e. without performaing setup/teardown
                  inbetween. This parameter is distinct from "repetitions", as
                  the latter takes place within the benchmark and produces a
                  single result.
                  """),
        Parameter('cpus', kind=cpu_mask, default=0, aliases=['taskset_mask'],
                  description="""
                  Specifies the CPU mask the benchmark process will be pinned to.
                  """),
    ]

    def setup(self, context):

        abi = self.target.abi
        if self.force_abi:
            abi = self.force_abi

        # self.test has been pre-validated, so this _should_ only fail if there's an abi mismatch
        host_exe = context.resolver.get(Executable(self, abi, self.test))
        self.target_exe = self.target.install(host_exe)
        self.commands = []

        setup_test = getattr(self, '_setup_{}'.format(self.test))
        setup_test()

    def run(self, context):
        for _ in range(self.loops):
            for command in self.commands:
                self.target.execute(command, timeout=self.run_timeout)

    def extract_results(self, context):
        results_path = self.target.get_workpath("lmbench.output")
        result_file = self.target.list_directory(results_path)[-1]
        self.result_file = result_file.rstrip()
        result = self.target.path.join(results_path, result_file)
        self.target.pull(result, context.output_directory)
        context.add_artifact('lmbench-result', "lmbench.output", kind='raw')

    def teardown(self, context):
        if self.uninstall:
            self.target.uninstall(self.test)

    #
    # Test setup routines
    #
    def _setup_lat_mem_rd(self):
        target_file = self.target.get_workpath('lmbench.output')
        self.target.execute('rm -f {}'.format(target_file))

        command_stub = self._setup_common()
        if self.thrash:
            command_stub = '{} -t'.format(command_stub)

        for size in self.size:
            command = '{} {}'.format(command_stub, size)
            for stride in self.stride:
                self.commands.append('{} {} >> {} 2>&1'.format(command, stride, target_file))

    def _setup_bw_mem(self):
        target_file = self.target.get_workpath('lmbench.output')
        self.target.execute('rm -f {}'.format(target_file))

        command_stub = self._setup_common()

        for size in self.size:
            command = '{} {}'.format(command_stub, size)
            for category in self.mem_category:
                self.commands.append('{} {} >> {} 2>&1'.format(command, category, target_file))

    def _setup_common(self):
        parts = []
        if self.cpus:
            parts.append('{} taskset {} {}'.format(self.target.busybox,
                                                   self.cpus.mask(),
                                                   self.target_exe))
        else:
            parts.append(self.target_exe)
        if self.parallelism is not None:
            parts.append('-P {}'.format(self.parallelism))
        if self.warmup is not None:
            parts.append('-W {}'.format(self.warmup))
        if self.repetitions is not None:
            parts.append('-N {}'.format(self.repetitions))
        return ' '.join(parts) + ' '