# Copyright 2015-2017 Intel Corporation. # # 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. """Automation of QEMU hypervisor for launching guests. """ import os import logging import locale import re import subprocess import time import pexpect from conf import settings as S from vnfs.vnf.vnf import IVnf class IVnfQemu(IVnf): """ Abstract class for controling an instance of QEMU """ _cmd = None _expect = None _proc_name = 'qemu' class GuestCommandFilter(logging.Filter): """ Filter out strings beginning with 'guestcmd :'. """ def filter(self, record): return record.getMessage().startswith(self.prefix) def __init__(self): """ Initialisation function. """ super(IVnfQemu, self).__init__() self._expect = S.getValue('GUEST_PROMPT_LOGIN')[self._number] self._logger = logging.getLogger(__name__) self._logfile = os.path.join( S.getValue('LOG_DIR'), S.getValue('LOG_FILE_QEMU')) + str(self._number) self._timeout = S.getValue('GUEST_TIMEOUT')[self._number] self._monitor = '%s/vm%dmonitor' % ('/tmp', self._number) # read GUEST NICs configuration and use only defined NR of NICS nics_nr = S.getValue('GUEST_NICS_NR')[self._number] # and inform user about missconfiguration if nics_nr < 1: raise RuntimeError('At least one VM NIC is mandotory, but {} ' 'NICs are configured'.format(nics_nr)) elif nics_nr > 1 and nics_nr % 2: nics_nr = int(nics_nr / 2) * 2 self._logger.warning('Odd number of NICs is configured, only ' '%s NICs will be used', nics_nr) self._nics = S.getValue('GUEST_NICS')[self._number][:nics_nr] # set guest loopback application based on VNF configuration self._guest_loopback = S.getValue('GUEST_LOOPBACK')[self._number] self._testpmd_fwd_mode = S.getValue('GUEST_TESTPMD_FWD_MODE')[self._number] # in case of SRIOV we must ensure, that MAC addresses are not swapped if S.getValue('SRIOV_ENABLED') and self._testpmd_fwd_mode.startswith('mac') and \ not str(S.getValue('VNF')).endswith('PciPassthrough'): self._logger.info("SRIOV detected, forwarding mode of testpmd was changed from '%s' to '%s'", self._testpmd_fwd_mode, 'io') self._testpmd_fwd_mode = 'io' name = 'Client%d' % self._number vnc = ':%d' % self._number # NOTE: affinization of main qemu process can cause hangup of 2nd VM # in case of DPDK usage. It can also slow down VM response time. cpumask = ",".join(S.getValue('GUEST_CORE_BINDING')[self._number]) self._cmd = ['sudo', '-E', 'taskset', '-c', cpumask, S.getValue('TOOLS')['qemu-system'], '-m', S.getValue('GUEST_MEMORY')[self._number], '-smp', str(S.getValue('GUEST_SMP')[self._number]), '-cpu', str(S.getValue('GUEST_CPU_OPTIONS')[self._number]), '-drive', 'if={},file='.format(S.getValue( 'GUEST_BOOT_DRIVE_TYPE')[self._number]) + S.getValue('GUEST_IMAGE')[self._number], '-boot', 'c', '--enable-kvm', '-monitor', 'unix:%s,server,nowait' % self._monitor, '-object', 'memory-backend-file,id=mem,size=' + str(S.getValue('GUEST_MEMORY')[self._number]) + 'M,' + 'mem-path=' + S.getValue('HUGEPAGE_DIR') + ',share=on', '-numa', 'node,memdev=mem -mem-prealloc', '-nographic', '-vnc', str(vnc), '-name', name, '-snapshot', '-net none', '-no-reboot', '-drive', 'if=%s,format=raw,file=fat:rw:%s,snapshot=off' % (S.getValue('GUEST_SHARED_DRIVE_TYPE')[self._number], S.getValue('GUEST_SHARE_DIR')[self._number]), ] self._configure_logging() def _configure_logging(self): """ Configure logging. """ self.GuestCommandFilter.prefix = self._log_prefix logger = logging.getLogger() cmd_logger = logging.FileHandler( filename=os.path.join(S.getValue('LOG_DIR'), S.getValue('LOG_FILE_GUEST_CMDS')) + str(self._number)) cmd_logger.setLevel(logging.DEBUG) cmd_logger.addFilter(self.GuestCommandFilter()) logger.addHandler(cmd_logger) # startup/Shutdown def start(self): """ Start QEMU instance, login and prepare for commands. """ super(IVnfQemu, self).start() if S.getValue('VNF_AFFINITIZATION_ON'): self._affinitize() if S.getValue('VSWITCH_VHOST_NET_AFFINITIZATION') and S.getValue( 'VNF') == 'QemuVirtioNet': self._affinitize_vhost_net() if self._timeout: self._config_guest_loopback() def stop(self): """ Stops VNF instance gracefully first. """ if self.is_running(): try: if self._login_active: # exit testpmd if needed if self._guest_loopback == 'testpmd': self.execute_and_wait('stop', 120, "Done") self.execute_and_wait('quit', 120, "[bB]ye") # turn off VM self.execute_and_wait('poweroff', 120, "Power down") except pexpect.TIMEOUT: self.kill() # wait until qemu shutdowns self._logger.debug('Wait for QEMU to terminate') for dummy in range(30): time.sleep(1) if not self.is_running(): break # just for case that graceful shutdown failed super(IVnfQemu, self).stop() # helper functions def login(self, timeout=120): """ Login to QEMU instance. This can be used immediately after booting the machine, provided a sufficiently long ``timeout`` is given. :param timeout: Timeout to wait for login to complete. :returns: True if login is active """ if self._login_active: return self._login_active # if no timeout was set, we likely started QEMU without waiting for it # to boot. This being the case, we best check that it has finished # first. if not self._timeout: self._expect_process(timeout=timeout) self._child.sendline(S.getValue('GUEST_USERNAME')[self._number]) self._child.expect(S.getValue('GUEST_PROMPT_PASSWORD')[self._number], timeout=5) self._child.sendline(S.getValue('GUEST_PASSWORD')[self._number]) self._expect_process(S.getValue('GUEST_PROMPT')[self._number], timeout=5) self._login_active = True return self._login_active def _affinitize(self): """ Affinitize the SMP cores of a QEMU instance. This is a bit of a hack. The 'socat' utility is used to interact with the QEMU HMP. This is necessary due to the lack of QMP in older versions of QEMU, like v1.6.2. In future releases, this should be replaced with calls to libvirt or another Python-QEMU wrapper library. :returns: None """ thread_id = (r'.* CPU #%d:.*? thread_id=(\d+)') self._logger.info('Affinitizing guest...') cur_locale = locale.getdefaultlocale()[1] proc = subprocess.Popen( ('echo', 'info cpus'), stdout=subprocess.PIPE) output = subprocess.check_output( ('sudo', 'socat', '-', 'UNIX-CONNECT:%s' % self._monitor), stdin=proc.stdout) proc.wait() # calculate the number of CPUs in SMP topology specified by GUEST_SMP # e.g. "sockets=2,cores=3", "4", etc. cpu_nr = 1 for i in re.findall(r'\d', S.getValue('GUEST_SMP')[self._number]): cpu_nr = cpu_nr * int(i) # pin each GUEST's core to host core based on configured BINDING for cpu in range(0, cpu_nr): match = None guest_thread_binding = S.getValue('GUEST_THREAD_BINDING')[self._number] if guest_thread_binding is None: guest_thread_binding = S.getValue('GUEST_CORE_BINDING')[self._number] for line in output.decode(cur_locale).split('\n'): match = re.search(thread_id % cpu, line) if match: self._affinitize_pid(guest_thread_binding[cpu], match.group(1)) break if not match: self._logger.error('Failed to affinitize guest core #%d. Could' ' not parse tid.', cpu) def _affinitize_vhost_net(self): """ Affinitize the vhost net threads for Vanilla OVS and guest nic queues. :return: None """ self._logger.info('Affinitizing VHOST Net threads.') args1 = ['pgrep', 'vhost-'] process1 = subprocess.Popen(args1, stdout=subprocess.PIPE, shell=False) out = process1.communicate()[0] processes = out.decode(locale.getdefaultlocale()[1]).split('\n') if processes[-1] == '': processes.pop() # pgrep may return an extra line with no data self._logger.info('Found %s vhost net threads...', len(processes)) cpumap = S.getValue('VSWITCH_VHOST_CPU_MAP') mapcount = 0 for proc in processes: self._affinitize_pid(cpumap[mapcount], proc) mapcount += 1 if mapcount + 1 > len(cpumap): # Not enough cpus were given in the mapping to cover all the # threads on a 1 to 1 ratio with cpus so reset the list counter # to 0. mapcount = 0 def _config_guest_loopback(self): """ Configure VM to run VNF, e.g. port forwarding application based on the configuration """ if self._guest_loopback == 'buildin': return self.login() if self._guest_loopback == 'testpmd': self._configure_testpmd() elif self._guest_loopback == 'l2fwd': self._configure_l2fwd() elif self._guest_loopback == 'linux_bridge': self._configure_linux_bridge() elif self._guest_loopback != 'clean': raise RuntimeError('Unsupported guest loopback method "%s" was specified.' % self._guest_loopback) def wait(self, prompt=None, timeout=30): if prompt is None: prompt = S.getValue('GUEST_PROMPT')[self._number] return super(IVnfQemu, self).wait(prompt=prompt, timeout=timeout) def execute_and_wait(self, cmd, timeout=30, prompt=None): if prompt is None: prompt = S.getValue('GUEST_PROMPT')[self._number] return super(IVnfQemu, self).execute_and_wait(cmd, timeout=timeout, prompt=prompt) def _modify_dpdk_makefile(self): """ Modifies DPDK makefile in Guest before compilation if needed """ pass def _configure_copy_sources(self, dirname): """ Mount shared directory and copy DPDK and l2fwd sources """ # mount shared directory self.execute_and_wait('umount /dev/sdb1') self.execute_and_wait('rm -rf ' + S.getValue('GUEST_OVS_DPDK_DIR')[self._number]) self.execute_and_wait('mkdir -p ' + S.getValue('GUEST_OVS_DPDK_SHARE')[self._number]) self.execute_and_wait('mount -o ro,iocharset=utf8 /dev/sdb1 ' + S.getValue('GUEST_OVS_DPDK_SHARE')[self._number]) self.execute_and_wait('mkdir -p ' + S.getValue('GUEST_OVS_DPDK_DIR')[self._number]) self.execute_and_wait('cp -r ' + os.path.join(S.getValue('GUEST_OVS_DPDK_SHARE')[self._number], dirname) + ' ' + S.getValue('GUEST_OVS_DPDK_DIR')[self._number]) self.execute_and_wait('umount /dev/sdb1') def _configure_disable_firewall(self): """ Disable firewall in VM """ for iptables in ['iptables', 'ip6tables']: # filter table for chain in ['INPUT', 'FORWARD', 'OUTPUT']: self.execute_and_wait("{} -t filter -P {} ACCEPT".format(iptables, chain)) # mangle table for chain in ['PREROUTING', 'INPUT', 'FORWARD', 'OUTPUT', 'POSTROUTING']: self.execute_and_wait("{} -t mangle -P {} ACCEPT".format(iptables, chain)) # nat table for chain in ['PREROUTING', 'INPUT', 'OUTPUT', 'POSTROUTING']: self.execute_and_wait("{} -t nat -P {} ACCEPT".format(iptables, chain)) # flush rules and delete chains created by user for table in ['filter', 'mangle', 'nat']: self.execute_and_wait("{} -t {} -F".format(iptables, table)) self.execute_and_wait("{} -t {} -X".format(iptables, table)) def _configure_testpmd(self): """ Configure VM to perform L2 forwarding between NICs by DPDK's testpmd """ self._configure_copy_sources('DPDK') self._configure_disable_firewall() # Guest images _should_ have 1024 hugepages by default, # but just in case:''' self.execute_and_wait('sysctl vm.nr_hugepages={}'.format(S.getValue('GUEST_HUGEPAGES_NR')[self._number])) # Mount hugepages self.execute_and_wait('mkdir -p /dev/hugepages') self.execute_and_wait( 'mount -t hugetlbfs hugetlbfs /dev/hugepages') # build and configure system for dpdk self.execute_and_wait('cd ' + S.getValue('GUEST_OVS_DPDK_DIR')[self._number] + '/DPDK') self.execute_and_wait('export CC=gcc') self.execute_and_wait('export RTE_SDK=' + S.getValue('GUEST_OVS_DPDK_DIR')[self._number] + '/DPDK') self.execute_and_wait('export RTE_TARGET=%s' % S.getValue('RTE_TARGET')) # modify makefile if needed self._modify_dpdk_makefile() # disable network interfaces, so DPDK can take care of them for nic in self._nics: self.execute_and_wait('ifdown ' + nic['device']) self.execute_and_wait('./*tools/dpdk*bind.py --status') pci_list = ' '.join([nic['pci'] for nic in self._nics]) self.execute_and_wait('./*tools/dpdk*bind.py -u ' + pci_list) self._bind_dpdk_driver(S.getValue( 'GUEST_DPDK_BIND_DRIVER')[self._number], pci_list) self.execute_and_wait('./*tools/dpdk*bind.py --status') # build and run 'test-pmd' self.execute_and_wait('cd ' + S.getValue('GUEST_OVS_DPDK_DIR')[self._number] + '/DPDK/app/test-pmd') self.execute_and_wait('make clean') self.execute_and_wait('make') # get testpmd settings from CLI testpmd_params = S.getValue('GUEST_TESTPMD_PARAMS')[self._number] if S.getValue('VSWITCH_JUMBO_FRAMES_ENABLED'): testpmd_params += ' --max-pkt-len={}'.format(S.getValue( 'VSWITCH_JUMBO_FRAMES_SIZE')) self.execute_and_wait('./testpmd {}'.format(testpmd_params), 60, "Done") self.execute_and_wait('set fwd ' + self._testpmd_fwd_mode, 20, 'testpmd>') self.execute_and_wait('start', 20, 'testpmd>') def _configure_l2fwd(self): """ Configure VM to perform L2 forwarding between NICs by l2fwd module """ if int(S.getValue('GUEST_NIC_QUEUES')[self._number]): self._set_multi_queue_nic() self._configure_copy_sources('l2fwd') self._configure_disable_firewall() # configure all interfaces for nic in self._nics: self.execute_and_wait('ip addr add ' + nic['ip'] + ' dev ' + nic['device']) if S.getValue('VSWITCH_JUMBO_FRAMES_ENABLED'): self.execute_and_wait('ifconfig {} mtu {}'.format( nic['device'], S.getValue('VSWITCH_JUMBO_FRAMES_SIZE'))) self.execute_and_wait('ip link set dev ' + nic['device'] + ' up') # build and configure system for l2fwd self.execute_and_wait('cd ' + S.getValue('GUEST_OVS_DPDK_DIR')[self._number] + '/l2fwd') self.execute_and_wait('export CC=gcc') self.execute_and_wait('make') if len(self._nics) == 2: self.execute_and_wait('insmod ' + S.getValue('GUEST_OVS_DPDK_DIR')[self._number] + '/l2fwd' + '/l2fwd.ko net1=' + self._nics[0]['device'] + ' net2=' + self._nics[1]['device']) else: raise RuntimeError('l2fwd can forward only between 2 NICs, but {} NICs are ' 'configured inside GUEST'.format(len(self._nics))) def _configure_linux_bridge(self): """ Configure VM to perform L2 forwarding between NICs by linux bridge """ if int(S.getValue('GUEST_NIC_QUEUES')[self._number]): self._set_multi_queue_nic() self._configure_disable_firewall() # configure linux bridge self.execute_and_wait('brctl addbr br0') # add all NICs into the bridge for nic in self._nics: self.execute_and_wait('ip addr add ' + nic['ip'] + ' dev ' + nic['device']) if S.getValue('VSWITCH_JUMBO_FRAMES_ENABLED'): self.execute_and_wait('ifconfig {} mtu {}'.format( nic['device'], S.getValue('VSWITCH_JUMBO_FRAMES_SIZE'))) self.execute_and_wait('ip link set dev ' + nic['device'] + ' up') self.execute_and_wait('brctl addif br0 ' + nic['device']) self.execute_and_wait('ip addr add {} dev br0'.format( S.getValue('GUEST_BRIDGE_IP')[self._number])) self.execute_and_wait('ip link set dev br0 up') # Add the arp entries for the IXIA ports and the bridge you are using. # Use command line values if provided. trafficgen_mac = S.getValue('VANILLA_TGEN_PORT1_MAC') trafficgen_ip = S.getValue('VANILLA_TGEN_PORT1_IP') self.execute_and_wait('arp -s ' + trafficgen_ip + ' ' + trafficgen_mac) trafficgen_mac = S.getValue('VANILLA_TGEN_PORT2_MAC') trafficgen_ip = S.getValue('VANILLA_TGEN_PORT2_IP') self.execute_and_wait('arp -s ' + trafficgen_ip + ' ' + trafficgen_mac) # Enable forwarding self.execute_and_wait('sysctl -w net.ipv4.ip_forward=1') # Controls source route verification # 0 means no source validation self.execute_and_wait('sysctl -w net.ipv4.conf.all.rp_filter=0') for nic in self._nics: self.execute_and_wait('sysctl -w net.ipv4.conf.' + nic['device'] + '.rp_filter=0') def _bind_dpdk_driver(self, driver, pci_slots): """ Bind the virtual nics to the driver specific in the conf file :return: None """ if driver == 'uio_pci_generic': if S.getValue('VNF') == 'QemuPciPassthrough': # unsupported config, bind to igb_uio instead and exit the # outer function after completion. self._logger.error('SR-IOV does not support uio_pci_generic. ' 'Igb_uio will be used instead.') self._bind_dpdk_driver('igb_uio_from_src', pci_slots) return self.execute_and_wait('modprobe uio_pci_generic') self.execute_and_wait('./*tools/dpdk*bind.py -b uio_pci_generic '+ pci_slots) elif driver == 'vfio_no_iommu': self.execute_and_wait('modprobe -r vfio') self.execute_and_wait('modprobe -r vfio_iommu_type1') self.execute_and_wait('modprobe vfio enable_unsafe_noiommu_mode=Y') self.execute_and_wait('modprobe vfio-pci') self.execute_and_wait('./*tools/dpdk*bind.py -b vfio-pci ' + pci_slots) elif driver == 'igb_uio_from_src': # build and insert igb_uio and rebind interfaces to it self.execute_and_wait('make RTE_OUTPUT=$RTE_SDK/$RTE_TARGET -C ' '$RTE_SDK/lib/librte_eal/linuxapp/igb_uio') self.execute_and_wait('modprobe uio') self.execute_and_wait('insmod %s/kmod/igb_uio.ko' % S.getValue('RTE_TARGET')) self.execute_and_wait('./*tools/dpdk*bind.py -b igb_uio ' + pci_slots) else: self._logger.error( 'Unknown driver for binding specified, defaulting to igb_uio') self._bind_dpdk_driver('igb_uio_from_src', pci_slots) def _set_multi_queue_nic(self): """ Enable multi-queue in guest kernel with ethool. :return: None """ for nic in self._nics: self.execute_and_wait('ethtool -L {} combined {}'.format( nic['device'], S.getValue('GUEST_NIC_QUEUES')[self._number])) self.execute_and_wait('ethtool -l {}'.format(nic['device']))