* [PATCH V7 3/3] tests/vhost_event_idx_interrupt_cbdma: add new testsuite
@ 2022-12-23 2:28 Wei Ling
2022-12-26 4:52 ` lijuan.tu
0 siblings, 1 reply; 2+ messages in thread
From: Wei Ling @ 2022-12-23 2:28 UTC (permalink / raw)
To: dts; +Cc: Wei Ling
Add TestSuite_vhost_event_idx_interrupt_cbdma.py in testsuite to test
the virtio enqueue with split ring and packed ring path and CBDMA.
Signed-off-by: Wei Ling <weix.ling@intel.com>
---
...stSuite_vhost_event_idx_interrupt_cbdma.py | 430 ++++++++++++++++++
1 file changed, 430 insertions(+)
create mode 100644 tests/TestSuite_vhost_event_idx_interrupt_cbdma.py
diff --git a/tests/TestSuite_vhost_event_idx_interrupt_cbdma.py b/tests/TestSuite_vhost_event_idx_interrupt_cbdma.py
new file mode 100644
index 00000000..adf27e01
--- /dev/null
+++ b/tests/TestSuite_vhost_event_idx_interrupt_cbdma.py
@@ -0,0 +1,430 @@
+# SPDX-License-Identifier: BSD-3-Clause
+# Copyright(c) 2022 Intel Corporation
+#
+
+import re
+import time
+
+from framework.test_case import TestCase
+from framework.virt_common import VM
+
+
+class TestVhostEventIdxInterruptCbdma(TestCase):
+ def set_up_all(self):
+ """
+ Run at the start of each test suite.
+
+ """
+ self.vm_num = 1
+ self.queues = 1
+ self.cores_num = len([n for n in self.dut.cores if int(n["socket"]) == 0])
+ self.prepare_l3fwd_power()
+ self.pci_info = self.dut.ports_info[0]["pci"]
+ self.base_dir = self.dut.base_dir.replace("~", "/root")
+ self.app_l3fwd_power_path = self.dut.apps_name["l3fwd-power"]
+ self.l3fwdpower_name = self.app_l3fwd_power_path.split("/")[-1]
+ self.dut_ports = self.dut.get_ports()
+ self.ports_socket = self.dut.get_numa_id(self.dut_ports[0])
+ self.cbdma_dev_infos = []
+ self.device_str = None
+
+ def set_up(self):
+ """
+ Run before each test case.
+ """
+ # Clean the execution ENV
+ self.verify_info = []
+ self.dut.send_expect(f"killall {self.l3fwdpower_name}", "#")
+ self.dut.send_expect("killall -s INT qemu-system-x86_64", "#")
+ self.dut.send_expect("rm -rf %s/vhost-net*" % self.base_dir, "#")
+ self.vhost = self.dut.new_session(suite="vhost-l3fwd")
+ self.vm_dut = []
+ self.vm = []
+ self.nopci = True
+
+ def get_core_mask(self):
+ self.core_config = "1S/%dC/1T" % (self.vm_num * self.queues)
+ self.verify(
+ self.cores_num >= self.queues * self.vm_num,
+ "There has not enought cores to test this case %s" % self.running_case,
+ )
+ self.core_list_l3fwd = self.dut.get_core_list(self.core_config)
+
+ def prepare_l3fwd_power(self):
+ out = self.dut.build_dpdk_apps("examples/l3fwd-power")
+ self.verify("Error" not in out, "compilation l3fwd-power error")
+
+ def list_split(self, items, n):
+ return [items[i : i + n] for i in range(0, len(items), n)]
+
+ @property
+ def check_2M_env(self):
+ out = self.dut.send_expect(
+ "cat /proc/meminfo |grep Hugepagesize|awk '{print($2)}'", "# "
+ )
+ return True if out == "2048" else False
+
+ def launch_l3fwd_power(self):
+ """
+ launch l3fwd-power with a virtual vhost device
+ """
+ res = True
+ self.logger.info("Launch l3fwd_sample sample:")
+ config_info = ""
+ core_index = 0
+ # config the interrupt cores info
+ for port in range(self.vm_num):
+ for queue in range(self.queues):
+ if config_info != "":
+ config_info += ","
+ config_info += "(%d,%d,%s)" % (
+ port,
+ queue,
+ self.core_list_l3fwd[core_index],
+ )
+ info = {
+ "core": self.core_list_l3fwd[core_index],
+ "port": port,
+ "queue": queue,
+ }
+ self.verify_info.append(info)
+ core_index = core_index + 1
+ # config the vdev info, if have 2 vms, it shoule have 2 vdev info
+ vdev_info = ""
+ self.cbdma_dev_infos_list = []
+ if self.vm_num >= 2:
+ self.cbdma_dev_infos_list = self.list_split(
+ self.cbdma_dev_infos, int(len(self.cbdma_dev_infos) / self.vm_num)
+ )
+ for i in range(self.vm_num):
+ dmas = ""
+ if self.vm_num == 1:
+ for queue in range(self.queues):
+ dmas += f"txq{queue}@{self.cbdma_dev_infos[queue]};"
+
+ else:
+ cbdma_dev_infos = self.cbdma_dev_infos_list[i]
+ for index, q in enumerate(cbdma_dev_infos):
+ dmas += f"txq{index}@{q};"
+ vdev_info += (
+ f"--vdev 'net_vhost%d,iface=%s/vhost-net%d,dmas=[{dmas}],queues=%d,client=1' "
+ % (i, self.base_dir, i, self.queues)
+ )
+
+ port_info = "0x1" if self.vm_num == 1 else "0x3"
+
+ example_para = self.app_l3fwd_power_path + " "
+ para = (
+ " --log-level=9 %s -- -p %s --parse-ptype 1 --config '%s' --interrupt-only"
+ % (vdev_info, port_info, config_info)
+ )
+ eal_params = self.dut.create_eal_parameters(
+ cores=self.core_list_l3fwd,
+ no_pci=self.nopci,
+ ports=self.used_cbdma,
+ )
+ command_line_client = example_para + eal_params + para
+ self.vhost.get_session_before(timeout=2)
+ self.vhost.send_expect(command_line_client, "POWER", 40)
+ time.sleep(10)
+ out = self.vhost.get_session_before()
+ if "Error" in out and "Error opening" not in out:
+ self.logger.error("Launch l3fwd-power sample error")
+ res = False
+ else:
+ self.logger.info("Launch l3fwd-power sample finished")
+ self.verify(res is True, "Lanuch l3fwd failed")
+
+ def relaunch_l3fwd_power(self):
+ """
+ relaunch l3fwd-power sample for port up
+ """
+ self.dut.send_expect("killall -s INT %s" % self.l3fwdpower_name, "#")
+ # make sure l3fwd-power be killed
+ pid = self.dut.send_expect(
+ "ps -ef |grep l3|grep -v grep |awk '{print $2}'", "#"
+ )
+ if pid:
+ self.dut.send_expect("kill -9 %s" % pid, "#")
+ self.launch_l3fwd_power()
+
+ def set_vm_cpu_number(self, vm_config):
+ # config the vcpu numbers when queue number greater than 1
+ if self.queues == 1:
+ return
+ params_number = len(vm_config.params)
+ for i in range(params_number):
+ if list(vm_config.params[i].keys())[0] == "cpu":
+ vm_config.params[i]["cpu"][0]["number"] = self.queues
+
+ def check_qemu_version(self, vm_config):
+ """
+ in this suite, the qemu version should greater 2.7
+ """
+ self.vm_qemu_version = vm_config.qemu_emulator
+ params_number = len(vm_config.params)
+ for i in range(params_number):
+ if list(vm_config.params[i].keys())[0] == "qemu":
+ self.vm_qemu_version = vm_config.params[i]["qemu"][0]["path"]
+
+ out = self.dut.send_expect("%s --version" % self.vm_qemu_version, "#")
+ result = re.search("QEMU\s*emulator\s*version\s*(\d*.\d*)", out)
+ self.verify(
+ result is not None,
+ "the qemu path may be not right: %s" % self.vm_qemu_version,
+ )
+ version = result.group(1)
+ index = version.find(".")
+ self.verify(
+ int(version[:index]) > 2
+ or (int(version[:index]) == 2 and int(version[index + 1 :]) >= 7),
+ "This qemu version should greater than 2.7 "
+ + "in this suite, please config it in vhost_sample.cfg file",
+ )
+
+ def start_vms(self, vm_num=1, packed=False):
+ """
+ start qemus
+ """
+ for i in range(vm_num):
+ vm_info = VM(self.dut, "vm%d" % i, "vhost_sample")
+ vm_info.load_config()
+ vm_params = {}
+ vm_params["driver"] = "vhost-user"
+ vm_params["opt_path"] = self.base_dir + "/vhost-net%d" % i
+ vm_params["opt_mac"] = "00:11:22:33:44:5%d" % i
+ vm_params["opt_server"] = "server"
+ if self.queues > 1:
+ vm_params["opt_queue"] = self.queues
+ opt_args = "csum=on,mq=on,vectors=%d" % (2 * self.queues + 2)
+ else:
+ opt_args = "csum=on"
+ if packed:
+ opt_args = opt_args + ",packed=on"
+ vm_params["opt_settings"] = opt_args
+ vm_info.set_vm_device(**vm_params)
+ self.set_vm_cpu_number(vm_info)
+ self.check_qemu_version(vm_info)
+ vm_dut = None
+ try:
+ vm_dut = vm_info.start(load_config=False, set_target=False)
+ if vm_dut is None:
+ raise Exception("Set up VM ENV failed")
+ except Exception as e:
+ self.logger.error("ERROR: Failure for %s" % str(e))
+ self.vm_dut.append(vm_dut)
+ self.vm.append(vm_info)
+
+ def config_virito_net_in_vm(self):
+ """
+ set vitio-net with 2 quques enable
+ """
+ for i in range(len(self.vm_dut)):
+ vm_intf = self.vm_dut[i].ports_info[0]["intf"]
+ self.vm_dut[i].send_expect(
+ "ethtool -L %s combined %d" % (vm_intf, self.queues), "#", 20
+ )
+
+ def check_vhost_core_status(self, vm_index, status):
+ """
+ check the cpu status
+ """
+ out = self.vhost.get_session_before()
+ for i in range(self.queues):
+ # because of the verify_info include all config(vm0 and vm1)
+ # so current index shoule vm_index + queue_index
+ verify_index = i + vm_index
+ if status == "waked up":
+ info = "lcore %s is waked up from rx interrupt on port %d queue %d"
+ info = info % (
+ self.verify_info[verify_index]["core"],
+ self.verify_info[verify_index]["port"],
+ self.verify_info[verify_index]["queue"],
+ )
+ elif status == "sleeps":
+ info = (
+ "lcore %s sleeps until interrupt triggers"
+ % self.verify_info[verify_index]["core"]
+ )
+ self.logger.info(info)
+ self.verify(info in out, "The CPU status not right for %s" % info)
+
+ def send_and_verify(self):
+ """
+ start to send packets and check the cpu status
+ stop and restart to send packets and check the cpu status
+ """
+ ping_ip = 3
+ for vm_index in range(self.vm_num):
+ session_info = []
+ vm_intf = self.vm_dut[vm_index].ports_info[0]["intf"]
+ self.vm_dut[vm_index].send_expect(
+ "ifconfig %s 1.1.1.%d" % (vm_intf, ping_ip), "#"
+ )
+ ping_ip = ping_ip + 1
+ self.vm_dut[vm_index].send_expect("ifconfig %s up" % vm_intf, "#")
+ for queue in range(self.queues):
+ session = self.vm_dut[vm_index].new_session(
+ suite="ping_info_%d" % queue
+ )
+ session.send_expect(
+ "taskset -c %d ping 1.1.1.%d" % (queue, ping_ip), "PING", 30
+ )
+ session_info.append(session)
+ ping_ip = ping_ip + 1
+ time.sleep(3)
+ self.check_vhost_core_status(vm_index=vm_index, status="waked up")
+ # close all sessions of ping in vm
+ for sess_index in range(len(session_info)):
+ session_info[sess_index].send_expect("^c", "#")
+ self.vm_dut[vm_index].close_session(session_info[sess_index])
+
+ def get_cbdma_ports_info_and_bind_to_dpdk(self):
+ """
+ get all cbdma ports
+ """
+ self.cbdma_dev_infos = []
+ self.used_cbdma = []
+ out = self.dut.send_expect(
+ "./usertools/dpdk-devbind.py --status-dev dma", "# ", 30
+ )
+ device_info = out.split("\n")
+ for device in device_info:
+ pci_info = re.search("\s*(0000:\S*:\d*.\d*)", device)
+ if pci_info is not None:
+ # dev_info = pci_info.group(1)
+ # the numa id of ioat dev, only add the device which
+ # on same socket with nic dev
+ self.cbdma_dev_infos.append(pci_info.group(1))
+ self.verify(
+ len(self.cbdma_dev_infos) >= self.queues,
+ "There no enough cbdma device to run this suite",
+ )
+ if self.queues == 1:
+ self.cbdma_dev_infos = [self.cbdma_dev_infos[0], self.cbdma_dev_infos[-1]]
+ self.used_cbdma = self.cbdma_dev_infos[0 : self.queues * self.vm_num]
+ self.device_str = " ".join(self.used_cbdma)
+ self.dut.send_expect(
+ "./usertools/dpdk-devbind.py --force --bind=%s %s"
+ % (self.drivername, self.device_str),
+ "# ",
+ 60,
+ )
+
+ def bind_cbdma_device_to_kernel(self):
+ if self.device_str is not None:
+ self.dut.send_expect("modprobe ioatdma", "# ")
+ self.dut.send_expect(
+ "./usertools/dpdk-devbind.py -u %s" % self.device_str, "# ", 30
+ )
+ self.dut.send_expect(
+ "./usertools/dpdk-devbind.py --force --bind=ioatdma %s"
+ % self.device_str,
+ "# ",
+ 60,
+ )
+
+ def stop_all_apps(self):
+ """
+ close all vms
+ """
+ for i in range(len(self.vm)):
+ self.vm[i].stop()
+ self.dut.send_expect("killall %s" % self.l3fwdpower_name, "#", timeout=2)
+
+ def test_wake_up_split_ring_vhost_user_cores_with_event_idx_interrupt_mode_16_queues_with_cbdma(
+ self,
+ ):
+ """
+ Test Case 1: wake up split ring vhost-user cores with event idx interrupt mode and cbdma enabled 16 queues test
+ """
+ self.vm_num = 1
+ self.bind_nic_driver(self.dut_ports)
+ self.queues = 16
+ self.get_core_mask()
+ self.nopci = False
+ self.get_cbdma_ports_info_and_bind_to_dpdk()
+ self.launch_l3fwd_power()
+ self.start_vms(
+ vm_num=self.vm_num,
+ )
+ self.relaunch_l3fwd_power()
+ self.config_virito_net_in_vm()
+ self.send_and_verify()
+ self.stop_all_apps()
+
+ def test_wake_up_split_ring_vhost_user_cores_by_multi_virtio_net_in_vms_with_event_idx_interrupt_with_cbdma(
+ self,
+ ):
+ """
+ Test Case 2: wake up split ring vhost-user cores by multi virtio-net in VMs with event idx interrupt mode and cbdma enabled test
+ """
+ self.vm_num = 2
+ self.bind_nic_driver(self.dut_ports)
+ self.queues = 1
+ self.get_core_mask()
+ self.nopci = False
+ self.get_cbdma_ports_info_and_bind_to_dpdk()
+ self.launch_l3fwd_power()
+ self.start_vms(
+ vm_num=self.vm_num,
+ )
+ self.relaunch_l3fwd_power()
+ self.config_virito_net_in_vm()
+ self.send_and_verify()
+ self.stop_all_apps()
+
+ def test_wake_up_packed_ring_vhost_user_cores_with_event_idx_interrupt_mode_16_queues_with_cbdma(
+ self,
+ ):
+ """
+ Test Case 3: wake up packed ring vhost-user cores with event idx interrupt mode and cbdma enabled 16 queues test
+ """
+ self.vm_num = 1
+ self.bind_nic_driver(self.dut_ports)
+ self.queues = 16
+ self.get_core_mask()
+ self.nopci = False
+ self.get_cbdma_ports_info_and_bind_to_dpdk()
+ self.launch_l3fwd_power()
+ self.start_vms(vm_num=self.vm_num, packed=True)
+ self.relaunch_l3fwd_power()
+ self.config_virito_net_in_vm()
+ self.send_and_verify()
+ self.stop_all_apps()
+
+ def test_wake_up_packed_ring_vhost_user_cores_by_multi_virtio_net_in_vms_with_event_idx_interrupt_with_cbdma(
+ self,
+ ):
+ """
+ Test Case 4: wake up packed ring vhost-user cores by multi virtio-net in VMs with event idx interrupt mode and cbdma enabled test
+ """
+ self.vm_num = 2
+ self.bind_nic_driver(self.dut_ports)
+ self.queues = 1
+ self.get_core_mask()
+ self.nopci = False
+ self.get_cbdma_ports_info_and_bind_to_dpdk()
+ self.launch_l3fwd_power()
+ self.start_vms(vm_num=self.vm_num, packed=True)
+ self.relaunch_l3fwd_power()
+ self.config_virito_net_in_vm()
+ self.send_and_verify()
+ self.stop_all_apps()
+
+ def tear_down(self):
+ """
+ Run after each test case.
+ """
+ self.dut.close_session(self.vhost)
+ self.dut.send_expect(f"killall {self.l3fwdpower_name}", "#")
+ self.dut.send_expect("killall -s INT qemu-system-x86_64", "#")
+ self.bind_cbdma_device_to_kernel()
+ if "cbdma" in self.running_case:
+ self.bind_nic_driver(self.dut_ports, self.drivername)
+
+ def tear_down_all(self):
+ """
+ Run after each test suite.
+ """
+ pass
--
2.25.1
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2022-12-26 4:52 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-12-23 2:28 [PATCH V7 3/3] tests/vhost_event_idx_interrupt_cbdma: add new testsuite Wei Ling
2022-12-26 4:52 ` lijuan.tu
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).