test suite reviews and discussions
 help / color / mirror / Atom feed
From: hanyingya <yingyax.han@intel.com>
To: dts@dpdk.org, yuan.peng@intel.com
Cc: hanyingya <yingyax.han@intel.com>
Subject: [dts] [PATCH V1]tests: add performance_thread test suite
Date: Thu,  5 Sep 2019 14:26:01 +0000	[thread overview]
Message-ID: <20190905142601.25547-1-yingyax.han@intel.com> (raw)

Signed-off-by: hanyingya <yingyax.han@intel.com>
---
 conf/performance_thread.cfg           |   4 +
 tests/TestSuite_performance_thread.py | 255 ++++++++++++++++++++++++++
 2 files changed, 259 insertions(+)
 create mode 100644 conf/performance_thread.cfg
 create mode 100644 tests/TestSuite_performance_thread.py

diff --git a/conf/performance_thread.cfg b/conf/performance_thread.cfg
new file mode 100644
index 0000000..2bc76ea
--- /dev/null
+++ b/conf/performance_thread.cfg
@@ -0,0 +1,4 @@
+[suite]
+# specify cores to test based on dut cpu specification.
+frame_size=[64, 128, 256, 512, 1024, 2000]
+cores=[2, 4, 8, 16]
diff --git a/tests/TestSuite_performance_thread.py b/tests/TestSuite_performance_thread.py
new file mode 100644
index 0000000..4d4cfe0
--- /dev/null
+++ b/tests/TestSuite_performance_thread.py
@@ -0,0 +1,255 @@
+# BSD LICENSE
+#
+# Copyright(c) 2010-2014 Intel Corporation. All rights reserved.
+# All rights reserved.
+#
+# 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 Intel Corporation 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.
+
+"""
+DPDK Test suite.
+Performance-Thread test script.
+"""
+import os
+import string
+import utils
+from test_case import TestCase
+from settings import HEADER_SIZE
+from pktgen import PacketGeneratorHelper
+
+
+class TestPerformanceThread(TestCase):
+
+    def set_up_all(self):
+        """
+        Run at the start of each test suite.
+        """
+        self.tester.extend_external_packet_generator(TestPerformanceThread, self)
+        self.dut_ports = self.dut.get_ports(self.nic)
+        global valports
+        valports = [_ for _ in self.dut_ports if self.tester.get_local_port(_) != -1]
+
+        # Verify that enough ports are available
+        self.verify(len(valports) >= 2, "Insufficent Ports")
+
+        self.port_mask = utils.create_mask(valports)
+        self.socket = self.dut.get_numa_id(self.dut_ports[0])
+
+        self.frame_sizes = self.get_suite_cfg()["frame_size"]
+        self.nb_cores = self.get_suite_cfg()["cores"]
+        self.headers_size = HEADER_SIZE['eth'] + HEADER_SIZE['ip'] + HEADER_SIZE['tcp']
+        self.path = "examples/performance-thread/l3fwd-thread/build/l3fwd-thread"
+
+        # compile performance_thread app
+        out = self.dut.build_dpdk_apps("./examples/performance-thread/l3fwd-thread")
+        self.verify("Error" not in out, "compilation error 1")
+        self.verify("No such file" not in out, "compilation error 2")
+
+        # results table header
+        self.test_results = {'header': [], 'data': []}
+
+        # get dts output path
+        if self.logger.log_path.startswith(os.sep):
+            self.output_path = self.logger.log_path
+        else:
+            cur_path = os.path.dirname(
+                os.path.dirname(os.path.realpath(__file__)))
+            self.output_path = os.sep.join([cur_path, self.logger.log_path])
+        # create an instance to set stream field setting
+        self.pktgen_helper = PacketGeneratorHelper()
+
+    def set_up(self):
+        """
+        Run before each test case.
+        """
+        pass
+
+    def flows(self):
+        """
+        Return a list of packets that implements the flows described.
+        """
+        return [
+            'IP(src="1.2.3.4",dst="192.18.1.0")',
+            'IP(src="1.2.3.4",dst="192.18.1.1")',
+            'IP(src="1.2.3.4",dst="192.18.0.0")',
+            'IP(src="1.2.3.4",dst="192.18.0.1")',
+            'IP(src="1.2.3.4",dst="192.18.3.0")',
+            'IP(src="1.2.3.4",dst="192.18.3.1")',
+            'IP(src="1.2.3.4",dst="192.18.2.0")',
+            'IP(src="1.2.3.4",dst="192.18.2.1")']
+
+    def create_cores(self, cores):
+
+        core_config = "1S/{}C/1T".format(cores)
+        core_list = self.dut.get_core_list(core_config, socket=self.socket)
+        core_mask = utils.create_mask(core_list)
+        return core_list, core_mask
+
+    def config_rx_tx(self, cores, core_list):
+
+        # config --tx and --tx params for performace thread app
+        if cores == 2:
+            rx = "({},{},{},{})".format(valports[0], 0, core_list[0], 0) + "," + "({},{},{},{})".format(valports[1], 0,
+                                                                                                        core_list[0], 0)
+            tx = "({},{})".format(core_list[1], 0)
+        elif cores == 4:
+            rx = "({},{},{},{})".format(valports[0], 0, core_list[0], 0) + "," + "({},{},{},{})".format(valports[1], 0,
+                                                                                                        core_list[1], 1)
+            tx = "({},{})".format(core_list[2], 0) + "," + "({},{})".format(core_list[3], 1)
+        elif cores == 8:
+            rx = "({},{},{},{})".format(valports[0], 0, core_list[0], 0) + "," + \
+                 "({},{},{},{})".format(valports[0], 1, core_list[1], 1) + "," + \
+                 "({},{},{},{})".format(valports[1], 0, core_list[2], 2) + "," + \
+                 "({},{},{},{})".format(valports[1], 1, core_list[3], 3)
+            tx = "({},{})".format(core_list[4], 0) + "," + "({},{})".format(core_list[5], 1) + "," + \
+                 "({},{})".format(core_list[6], 2) + "," + "({},{})".format(core_list[7], 3)
+        elif cores == 16:
+            rx = "({},{},{},{})".format(valports[0], 0, core_list[0], 0) + "," + \
+                 "({},{},{},{})".format(valports[0], 1, core_list[1], 1) + "," + \
+                 "({},{},{},{})".format(valports[0], 2, core_list[2], 2) + "," + \
+                 "({},{},{},{})".format(valports[0], 3, core_list[3], 3) + "," + \
+                 "({},{},{},{})".format(valports[1], 0, core_list[4], 4) + "," + \
+                 "({},{},{},{})".format(valports[1], 1, core_list[5], 5) + "," + \
+                 "({},{},{},{})".format(valports[1], 2, core_list[6], 6) + "," + \
+                 "({},{},{},{})".format(valports[1], 3, core_list[7], 7)
+            tx = "({},{})".format(core_list[8], 0) + "," + "({},{})".format(core_list[9], 1) + "," + \
+                 "({},{})".format(core_list[10], 2) + "," + "({},{})".format(core_list[11], 3) + "," + \
+                 "({},{})".format(core_list[12], 4) + "," + "({},{})".format(core_list[13], 5) + "," + \
+                 "({},{})".format(core_list[14], 6) + "," + "({},{})".format(core_list[15], 7)
+
+        lcore_config = "(%s-%s)@%s" % (core_list[0], core_list[-1], core_list[0])
+        return lcore_config, rx, tx
+
+    def create_pacap_file(self, frame_size):
+        """
+        Prepare traffic flow
+        """
+        dmac = [self.dut.get_mac_address(self.dut_ports[i]) for i in valports]
+        smac = ["02:00:00:00:00:0%d" % i for i in valports]
+        payload_size = frame_size - HEADER_SIZE['ip'] - HEADER_SIZE['eth']
+        pcaps = {}
+        for _port in valports:
+            index = valports[_port]
+            cnt = 0
+            for layer in self.flows()[_port * 2:(_port + 1) * 2]:
+                flow = ['Ether(dst="%s", src="%s")/%s/("X"*%d)' % (dmac[index], smac[index], layer, payload_size)]
+                pcap = os.sep.join([self.output_path, "dst{0}_{1}.pcap".format(index, cnt)])
+                self.tester.scapy_append('wrpcap("%s", [%s])' % (pcap, string.join(flow, ',')))
+                self.tester.scapy_execute()
+                if index not in pcaps:
+                    pcaps[index] = []
+                pcaps[index].append(pcap)
+                cnt += 1
+        return pcaps
+
+    def prepare_stream(self, pcaps):
+        """
+        create streams for ports,one port one stream
+        """
+        tgen_input = []
+        for rxPort in valports:
+            if rxPort % len(valports) == 0 or len(valports) % rxPort == 2:
+                txIntf = self.tester.get_local_port(valports[rxPort + 1])
+                port_id = valports[rxPort + 1]
+            else:
+                txIntf = self.tester.get_local_port(valports[rxPort - 1])
+                port_id = valports[rxPort - 1]
+            rxIntf = self.tester.get_local_port(valports[rxPort])
+            for pcap in pcaps[port_id]:
+                tgen_input.append((txIntf, rxIntf, pcap))
+        return tgen_input
+
+    def perf_test(self, params):
+
+        # create result table
+        header_row = ["Frame size", "S/C/T", "Throughput(Mpps)", "Line Rate(%)"]
+        self.test_results["header"] = header_row
+        self.result_table_create(header_row)
+        self.test_results["data"] = []
+
+        for cores in self.nb_cores:
+            core_list, core_mask = self.create_cores(cores)
+            lcore_config, rx, tx = self.config_rx_tx(cores, core_list)
+            if self.running_case is "test_perf_n_lcore_per_pcore":
+                cmdline = "{} -n 4 --lcores='{}' {} --rx='{}' --tx='{}'".format(self.path, lcore_config, params, rx, tx)
+            else:
+                cmdline = "{} -c {} {} --rx='{}' --tx='{}'".format(self.path, core_mask, params, rx, tx)
+            self.dut.send_expect(cmdline, "L3FWD:", 120)
+
+            for frame_size in self.frame_sizes:
+                pcaps = self.create_pacap_file(frame_size)
+                # send the traffic and Measure test
+                tgenInput = self.prepare_stream(pcaps)
+
+                vm_config = self.set_fields()
+                # clear streams before add new streams
+                self.tester.pktgen.clear_streams()
+                # run packet generator
+                streams = self.pktgen_helper.prepare_stream_from_tginput(tgenInput, 100, vm_config, self.tester.pktgen)
+                # set traffic option
+                traffic_opt = {'delay': 15}
+                _, pps = self.tester.pktgen.measure_throughput(stream_ids=streams, options=traffic_opt)
+                self.verify(pps > 0, "No traffic detected")
+                pps /= 1000000.0
+                linerate = self.wirespeed(self.nic, frame_size, len(valports))
+                percentage = pps * 100 / linerate
+                data_row = [frame_size, "1S/{}C/1T".format(cores), str(pps), str(percentage)]
+                self.result_table_add(data_row)
+                self.test_results["data"].append(data_row)
+            # stop application
+            self.dut.send_expect("^C", "# ", 15)
+        self.result_table_print()
+
+    def test_perf_one_lcore_per_pcore(self):
+        params = "-n 4 -- -P -p {} --enable-jumbo --max-pkt-len=2500 --no-lthread" .format(self.port_mask)
+        self.perf_test(params)
+
+    def test_perf_n_lthreads_per_pcore(self):
+        params = "-n 4 -- -P -p {} --enable-jumbo --max-pkt-len=2500" .format(self.port_mask)
+        self.perf_test(params)
+
+    def test_perf_n_lcore_per_pcore(self):
+        params = "-- -P -p {} --enable-jumbo --max-pkt-len 2500 --no-lthread" .format(self.port_mask)
+        self.perf_test(params)
+
+    def set_fields(self):
+        """
+        set ip protocol field behavior
+        """
+        fields_config = {'ip':  {'src': {'action': 'random'}, }, }
+        return fields_config
+
+    def tear_down(self):
+        """
+        Run after each test case.
+        """
+        self.dut.kill_all()
+
+    def tear_down_all(self):
+        """
+        Run after each test suite.
+        """
+        pass
-- 
2.17.1


             reply	other threads:[~2019-09-05  6:26 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-09-05 14:26 hanyingya [this message]
2019-09-18  5:37 ` Peng, Yuan
2019-09-20  6:41 ` Tu, Lijuan

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20190905142601.25547-1-yingyax.han@intel.com \
    --to=yingyax.han@intel.com \
    --cc=dts@dpdk.org \
    --cc=yuan.peng@intel.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).