hv_check_cpu_utilization: cpu utilization check with all hv flags

Check host cpu utilization for each vcpu is less than 5% when
guest is idle with all hv flags
Signed-off-by: NYu Wang <wyu@redhat.com>

qemu/tests/cfg/hv_check_cpu_utilization.cfg

+- hv_check_cpu_utilization:
+    only Windows
+    type = hv_check_cpu_utilization
+    clone_master = yes
+    master_images_clone = image1
+    remove_image_image1 =yes
+    nics = ""
+    cdroms = ""
+    cpu_model_flags = "hv_crash"
+    host_check_times = 900
+    host_check_interval = 2
+    serives_to_stop = wuauserv bits dosvc SysMain
+    service_check_cmd = sc query %s
+    service_stop_cmd = sc stop {0} & sc config {0} start=disabled
+    reg_cmd = reg add "HKLM\SYSTEM\CurrentControlSet\Services\SecurityHealthService" /v Start /d 4 /t REG_DWORD /f
+    host_check_cmd = top -H -p %s -n ${host_check_times} -d ${host_check_interval} -b > fixed-top-pc-result
+    vcpn_thread_pattern = r'thread_id.?[:|=]\s*(\d+)'
+    thread_process_cmd = "cat fixed-top-pc-result |grep %s|awk -F ' ' '    {print $9;}'|awk '{sum+=$1} END {print sum/NR}'"
+    thread_cpu_level = 5
+    post_command = rm -f fixed-top-pc-result

qemu/tests/hv_check_cpu_utilization.py

+import time
+import threading
+import re
+import logging
+
+from virttest import utils_misc
+from virttest import utils_test
+from virttest import error_context
+from avocado.utils import process
+
+
+def _check_cpu_usage(session):
+    """
+    Check windows guest cpu usage by wmic. This function is used within
+    utils_misc.wait_for(), to check cpu usage repeatedly.
+
+    param session: a session object to send wmic commands
+    """
+    status, output = session.cmd_status_output(
+        "wmic cpu get loadpercentage /value")
+    if not status:
+        result = re.search(r'LoadPercentage=(\d+)', output)
+        if result:
+            percent = int(result.group(1))
+            if percent > 1:
+                logging.warning("Guest cpu usage :%s%%", percent)
+
+
+def _check_cpu_thread_func(session, timeout):
+    """
+    Repeatedlly checking guest cpu usage, until timeout has arrived.
+
+    param session: a session object to send commands
+    param timeout: total checking time
+    """
+    utils_misc.wait_for(lambda: _check_cpu_usage(session), timeout, 0, 5)
+
+
+def _pin_vm_threads(vm, node):
+    """
+    Pin guest to certain numa node.
+
+    param vm: a vm object
+    param node: a numa node to pin to
+    """
+    node = utils_misc.NumaNode(node)
+    utils_test.qemu.pin_vm_threads(vm, node)
+
+
+def _stop_service(test, params, session, service):
+    """
+    Check & stop windows system service
+
+    param session: a session to send commands
+    param service: the name of the service to stop
+    """
+    service_check_cmd = params.get("service_check_cmd")
+    service_stop_cmd = params.get("service_stop_cmd")
+    s, o = session.cmd_status_output("sc query")
+    if s:
+        test.error("Failed to query service list, "
+                   "status=%s, output=%s" % (s, o))
+    service_item = re.search(
+        r'SERVICE_NAME:\s+%s' % service, o, re.I | re.M)
+    if not service_item:
+        return
+
+    s, o = session.cmd_status_output(service_check_cmd % service)
+    if s:
+        test.error("Failed to get status for service: %s, "
+                   "status=%s, output=%s" % (service, s, o))
+    if re.search(r'STOPPED', o, re.I | re.M):
+        return
+    session.cmd(service_stop_cmd.format(service))
+
+
+@error_context.context_aware
+def run(test, params, env):
+    """
+    Cpu utilization test with hv flags.
+
+    1)Start a Windows guest vm.
+    2)Pin the vm to certain numa node, to keep accuracy.
+    3)Stop serval Windows services & background processes on guest.
+      to lower the cpu usage to minimum.
+    4)Reboot vm to apply changes, then wait for serveral minutes to make
+      sure the cpu is chill down.
+    5)Start both checking the guest&host's cpu usage, monitoring the value.
+    6)Compare the average utilization value to standard values.
+
+    param test: the test object
+    param params: the params of the test
+    param env: the testing environment object
+    """
+    vm = env.get_vm(params["main_vm"])
+
+    # pin guest vcpus/memory/vhost threads to last numa node of host
+    _pin_vm_threads(vm, params.get_numeric("numa_node", -1))
+
+    vm.verify_alive()
+
+    timeout = params.get_numeric("login_timeout", 240)
+    host_check_times = params.get_numeric("host_check_times", 900)
+    host_check_interval = params.get_numeric("host_check_interval", 2)
+    guest_check_timeout = host_check_times * host_check_interval
+    thread_cpu_level = params.get_numeric("thread_cpu_level", 5)
+    session = vm.wait_for_serial_login(timeout=timeout)
+
+    service_names = params.get("serives_to_stop").split()
+
+    # check and stop services
+    for service in service_names:
+        _stop_service(test, params, session, service)
+
+    # stop windows defender
+    session.cmd(params["reg_cmd"])
+
+    session = vm.reboot(session, timeout=timeout, serial=True)
+
+    # wait for the guest to chill
+    time.sleep(900)
+
+    # start background checking guest cpu usage
+    thread = threading.Thread(target=_check_cpu_thread_func,
+                              args=(session, guest_check_timeout))
+    thread.start()
+    time.sleep(60)
+
+    # start checking host cpu usage
+    pid = vm.get_pid()
+    process.system(params["host_check_cmd"] % pid, shell=True)
+    thread.join(guest_check_timeout + 360)
+
+    vcpu_thread_pattern = params.get("vcpu_thread_pattern",
+                                     r'thread_id.?[:|=]\s*(\d+)')
+    vcpu_ids = vm.get_vcpu_pids(vcpu_thread_pattern)
+    for thread_id in vcpu_ids:
+        # output result
+        host_cpu_usage = process.system_output(
+            params["thread_process_cmd"] % thread_id, shell=True)
+        host_cpu_usage = float(host_cpu_usage.decode())
+        if host_cpu_usage > thread_cpu_level:
+            test.fail("The cpu usage of thread %s is %s"
+                      " > %s" % (thread_id, host_cpu_usage, thread_cpu_level))