test suite reviews and discussions
 help / color / mirror / Atom feed
From: "Tu, Lijuan" <lijuan.tu@intel.com>
To: "Mo, YufengX" <yufengx.mo@intel.com>,
	"dts@dpdk.org" <dts@dpdk.org>,
	"Wang, Yinan" <yinan.wang@intel.com>
Subject: Re: [dts] [PATCH V2 1/1] framework/qemu_libvirt: add new features
Date: Wed, 28 Aug 2019 10:29:02 +0000	[thread overview]
Message-ID: <8CE3E05A3F976642AAB0F4675D0AD20E0BB06C1A@SHSMSX101.ccr.corp.intel.com> (raw)
In-Reply-To: <20190828074756.16519-2-yufengx.mo@intel.com>

Applied, thanks

> -----Original Message-----
> From: Mo, YufengX
> Sent: Wednesday, August 28, 2019 3:48 PM
> To: dts@dpdk.org; Wang, Yinan <yinan.wang@intel.com>; Tu, Lijuan
> <lijuan.tu@intel.com>
> Cc: Mo, YufengX <yufengx.mo@intel.com>
> Subject: [dts][PATCH V2 1/1] framework/qemu_libvirt: add new features
> 
> 
> *. support bridge setting in xml.
> *. support vhost-user setting in xml. support to add driver attribute by set
> option opt_queue and opt_setting.
> *. add __add_vm_net_tap method to add tap&bridge net device.
> *. add __generate_net_config_script method to set default br0 using script.
> *. add __write_config method to make recursion write config in xml config
> file.
> *. use dut logger to take the place of print.
> *. remove graphics vnc config in set_vm_default method, add add_vm_vnc
> method to do graphics vnc setting. This is aimed to keep the same as
> qemu_kvm method.
> *. add add_vm_daemon method to keep the same as qemu_kvm method.
> libvirt run virtual machine with daemon status by default.
> 
> Signed-off-by: yufengmx <yufengx.mo@intel.com>
> ---
>  framework/qemu_libvirt.py | 275 +++++++++++++++++++++++++++++++++----
> -
>  1 file changed, 241 insertions(+), 34 deletions(-)
> 
> diff --git a/framework/qemu_libvirt.py b/framework/qemu_libvirt.py index
> ab62ab8..d2edd31 100644
> --- a/framework/qemu_libvirt.py
> +++ b/framework/qemu_libvirt.py
> @@ -48,6 +48,21 @@ from xml.dom import minidom
> 
> 
>  class LibvirtKvm(VirtBase):
> +    DEFAULT_BRIDGE = 'br0'
> +    QEMU_IFUP = "#!/bin/sh\n\n" + \
> +                "set -x\n\n" + \
> +                "switch=%(switch)s\n\n" + \
> +                "if [ -n '$1' ];then\n" + \
> +                "   tunctl -t $1\n" + \
> +                "   ip link set $1 up\n" + \
> +                "   sleep 0.5s\n" + \
> +                "   brctl addif $switch $1\n" + \
> +                "   exit 0\n" + \
> +                "else\n" + \
> +                "   echo 'Error: no interface specified'\n" + \
> +                "   exit 1\n" + \
> +                "fi"
> +    QEMU_IFUP_PATH = '/etc/qemu-ifup'
> 
>      def __init__(self, dut, name, suite):
>          # initialize virtualization base module @@ -56,6 +71,7 @@ class
> LibvirtKvm(VirtBase):
>          # initialize qemu emulator, example: qemu-system-x86_64
>          self.qemu_emulator = self.get_qemu_emulator()
> 
> +        self.logger = dut.logger
>          # disk and pci device default index
>          self.diskindex = 'a'
>          self.controllerindex = 0
> @@ -258,9 +274,6 @@ class LibvirtKvm(VirtBase):
>          device = ET.SubElement(self.domain, 'devices')
>          ET.SubElement(device, 'emulator').text = self.qemu_emulator
> 
> -        # graphic device
> -        ET.SubElement(device, 'graphics', {
> -                      'type': 'vnc', 'port': '-1', 'autoport': 'yes'})
>          # qemu guest agent
>          self.add_vm_qga(None)
> 
> @@ -348,6 +361,38 @@ class LibvirtKvm(VirtBase):
>                   'slot': '0x00', 'function': '0x00'})
>              self.pciindex += 1
> 
> +    def add_vm_daemon(self, **options):
> +        pass
> +
> +    def add_vm_vnc(self, **options):
> +        """
> +        Add VM display option
> +        """
> +        disable = options.get('disable')
> +        if disable and disable == 'True':
> +            return
> +        else:
> +            displayNum = options.get('displayNum')
> +            port = \
> +                displayNum if displayNum else \
> +                self.virt_pool.alloc_port(self.vm_name, port_type="display")
> +        ip = self.host_dut.get_ip_address()
> +        # set main block
> +        graphics = {
> +            'type': 'vnc',
> +            'port': port,
> +            'autoport': 'yes',
> +            'listen': ip,
> +            'keymap': 'en-us', }
> +
> +        devices = self.domain.find('devices')
> +        graphics = ET.SubElement(devices, 'graphics', graphics)
> +        # set sub block
> +        listen = {
> +            'type': 'address',
> +            'address': ip, }
> +        ET.SubElement(graphics, 'listen', listen)
> +
>      def add_vm_serial_port(self, **options):
>          if 'enable' in options.keys():
>              if options['enable'].lower() == 'yes':
> @@ -370,8 +415,8 @@ class LibvirtKvm(VirtBase):
>                          {'mode': 'bind', 'path': self.serial_path})
>                      ET.SubElement(serial, 'target', {'port': '0'})
>                  else:
> -                    print utils.RED(
> -                        "Serial type %s is not supported!" % serial_type)
> +                    msg = "Serial type %s is not supported!" % serial_type
> +                    self.logger.error(msg)
>                      return False
>                  console = ET.SubElement(
>                      devices, 'console', {'type': serial_type}) @@ -417,54 +462,189
> @@ class LibvirtKvm(VirtBase):
>          return None
> 
>      def set_vm_device(self, driver='pci-assign', **opts):
> +        opts['driver'] = driver
>          self.add_vm_device(**opts)
> 
> -    def add_vm_device(self, **options):
> +    def __generate_net_config_script(self, switch=DEFAULT_BRIDGE):
>          """
> -        options:
> -            pf_idx: device index of pass-through device
> -            guestpci: assigned pci address in vm
> +        Generate a script for qemu emulator to build a tap device
> +        between host and guest.
>          """
> -        devices = self.domain.find('devices')
> -        hostdevice = ET.SubElement(devices, 'hostdev', {
> -                                   'mode': 'subsystem', 'type': 'pci',
> -                                   'managed': 'yes'})
> +        qemu_ifup = self.QEMU_IFUP % {'switch': switch}
> +        file_name = os.path.basename(self.QEMU_IFUP_PATH)
> +        tmp_file_path = '/tmp/%s' % file_name
> +        self.host_dut.create_file(qemu_ifup, tmp_file_path)
> +        self.host_session.send_expect(
> +            'mv -f ~/%s %s' % (file_name, self.QEMU_IFUP_PATH), '# ')
> +        self.host_session.send_expect(
> +            'chmod +x %s' % self.QEMU_IFUP_PATH, '# ')
> 
> -        if 'opt_host' in options.keys():
> -            pci_addr = options['opt_host']
> -        else:
> -            print utils.RED("Missing opt_host for device option!!!")
> +    def __parse_opt_setting(self, opt_settings):
> +        if '=' not in opt_settings:
> +            msg = 'wrong opt_settings setting'
> +            raise Exception(msg)
> +        setting = [item.split('=') for item in opt_settings.split(',')]
> +        return dict(setting)
> +
> +    def __get_pci_addr_config(self, pci):
> +        pci = self.__parse_pci(pci)
> +        if pci is None:
> +            msg = 'Invalid guestpci for host device pass-through !!!'
> +            self.logger.error(msg)
>              return False
> +        bus, slot, func, dom = pci
> +        config = {
> +            'type': 'pci', 'domain': '0x%s' % dom, 'bus': '0x%s' % bus,
> +            'slot': '0x%s' % slot, 'function': '0x%s' % func}
> +        return config
> +
> +    def __write_config(self, parent, configs):
> +        for config in configs:
> +            node_name = config[0]
> +            opt = config[1]
> +            node = ET.SubElement(parent, node_name, opt)
> +            if len(config) == 3:
> +                self.__write_config(node, config[2])
> +
> +    def __set_vm_bridge_interface(self, **options):
> +        mac = options.get('opt_mac')
> +        opt_br = options.get('opt_br')
> +        if not mac or not opt_br:
> +            msg = "Missing some bridge device option !!!"
> +            self.logger.error(msg)
> +            return False
> +        _config = [
> +            ['mac', {'address': mac}],
> +            ['source', {'bridge': opt_br, }],
> +            ['model', {'type': 'virtio', }]]
> +        config = [['interface', {'type': 'bridge'}, _config]]
> +        # set xml file
> +        parent = self.domain.find('devices')
> +        self.__write_config(parent, config)
> +
> +    def __add_vm_virtio_user_pci(self, **options):
> +        mac = options.get('opt_mac')
> +        mode = options.get('opt_server') or 'client'
> +        # unix socket path of character device
> +        sock_path = options.get('opt_path')
> +        queue = options.get('opt_queue')
> +        settings = options.get('opt_settings')
> +        # pci address in virtual machine
> +        pci = options.get('opt_host')
> +        if not mac or not sock_path:
> +            msg = "Missing some vhostuser device option !!!"
> +            self.logger.error(msg)
> +            return False
> +        node_name = 'interface'
> +        # basic options
> +        _config = [
> +            ['mac', {'address': mac}],
> +            ['source', {'type': 'unix',
> +                        'path': sock_path,
> +                        'mode': mode, }],
> +            ['model', {'type': 'virtio', }]]
> +        # append pci address
> +        if pci:
> +            _config.append(['address', self.__get_pci_addr_config(pci)])
> +        if queue or settings:
> +            drv_config = {'name': 'vhost'}
> +            if settings:
> +                _sub_opt = self.__parse_opt_setting(settings)
> +                drv_opt = {}
> +                guest_opt = {}
> +                host_opt = {}
> +                for key, value in _sub_opt.iteritems():
> +                    if key.startswith('host_'):
> +                        host_opt[key[5:]] = value
> +                        continue
> +                    if key.startswith('guest_'):
> +                        guest_opt[key[6:]] = value
> +                        continue
> +                    drv_opt[key] = value
> +                drv_config.update(drv_opt)
> +                sub_drv_config = []
> +                if host_opt:
> +                    sub_drv_config.append(['host', host_opt])
> +                if guest_opt:
> +                    sub_drv_config.append(['guest', guest_opt])
> +            # The optional queues attribute controls the number of queues to
> be
> +            # used for either Multiqueue virtio-net or vhost-user network
> +            # interfaces. Each queue will potentially be handled by a different
> +            # processor, resulting in much higher throughput. virtio-net since
> +            # 1.0.6 (QEMU and KVM only) vhost-user since 1.2.17(QEMU and
> KVM
> +            # only).
> +            if queue:
> +                drv_config.update({'queues': queue, })
> +            # set driver config
> +            if sub_drv_config:
> +                _config.append(['driver', drv_config, sub_drv_config])
> +            else:
> +                _config.append(['driver', drv_config])
> +        config = [[node_name, {'type': 'vhostuser'}, _config]]
> +        # set xml file
> +        parent = self.domain.find('devices')
> +        self.__write_config(parent, config)
> 
> +    def __add_vm_pci_assign(self, **options):
> +        devices = self.domain.find('devices')
> +        # add hostdev config block
> +        config = {
> +            'mode': 'subsystem',
> +            'type': 'pci',
> +            'managed': 'yes'}
> +        hostdevice = ET.SubElement(devices, 'hostdev', config)
> +        # add hostdev/source config block
> +        pci_addr = options.get('opt_host')
> +        if not pci_addr:
> +            msg = "Missing opt_host for device option!!!"
> +            self.logger.error(msg)
> +            return False
>          pci = self.__parse_pci(pci_addr)
>          if pci is None:
>              return False
>          bus, slot, func, dom = pci
> -
>          source = ET.SubElement(hostdevice, 'source')
> -        ET.SubElement(source, 'address', {
> -                      'domain': '0x%s' % dom, 'bus': '0x%s' % bus,
> -                      'slot': '0x%s' % slot,
> -                      'function': '0x%s' % func})
> -        if 'guestpci' in options.keys():
> -            guest_pci_addr = options['guestpci']
> -        else:
> +        config = {
> +            'domain': '0x%s' % dom,
> +            'bus': '0x%s' % bus,
> +            'slot': '0x%s' % slot,
> +            'function': '0x%s' % func}
> +        ET.SubElement(source, 'address', config)
> +        # add hostdev/source/address config block
> +        guest_pci_addr = options.get('guestpci')
> +        if not guest_pci_addr:
>              guest_pci_addr = '0000:%s:00.0' % hex(self.pciindex)[2:]
>              self.pciindex += 1
> -        pci = self.__parse_pci(guest_pci_addr)
> -        if pci is None:
> -            print utils.RED('Invalid guestpci for host device pass-through!!!')
> -            return False
> -        bus, slot, func, dom = pci
> -        ET.SubElement(hostdevice, 'address', {
> -            'type': 'pci', 'domain': '0x%s' % dom, 'bus': '0x%s' % bus,
> -            'slot': '0x%s' % slot, 'function': '0x%s' % func})
> +        config = self.__get_pci_addr_config(guest_pci_addr)
> +        ET.SubElement(hostdevice, 'address', config)
>          # save host and guest pci address mapping
>          pci_map = {}
>          pci_map['hostpci'] = pci_addr
>          pci_map['guestpci'] = guest_pci_addr
>          self.pci_maps.append(pci_map)
> 
> +    def add_vm_device(self, **options):
> +        """
> +        options:
> +            pf_idx: device index of pass-through device
> +            guestpci: assigned pci address in vm
> +        """
> +        driver_table = {
> +            'vhost-user':
> +                self.__add_vm_virtio_user_pci,
> +            'bridge':
> +                self.__set_vm_bridge_interface,
> +            'pci-assign':
> +                self.__add_vm_pci_assign,
> +        }
> +        driver = options.get('driver')
> +        if not driver or driver not in driver_table.keys():
> +            driver = 'pci-assign'
> +            msg = 'use {0} configuration as default driver'.format(driver)
> +            self.logger.warning(msg)
> +        func = driver_table.get(driver)
> +        func(**options)
> +
>      def add_vm_net(self, **options):
>          """
>          Options:
> @@ -473,6 +653,8 @@ class LibvirtKvm(VirtBase):
>          if 'type' in options.keys():
>              if options['type'] == 'nic':
>                  self.__add_vm_net_nic(**options)
> +            elif options['type'] == 'tap':
> +                self.__add_vm_net_tap(**options)
> 
>      def __add_vm_net_nic(self, **options):
>          """
> @@ -514,6 +696,30 @@ class LibvirtKvm(VirtBase):
>              ET.SubElement(qemu, 'qemu:arg', {'value': 'user,hostfwd='
>                                               'tcp:%s:%d-:22' % (dut_ip, port)})
> 
> +    def __add_vm_net_tap(self, **options):
> +        """
> +        type: tap
> +        opt_br: br0
> +            note: if choosing tap, need to specify bridge name,
> +                  else it will be br0.
> +        opt_script: QEMU_IFUP_PATH
> +            note: if not specified, default is self.QEMU_IFUP_PATH.
> +        """
> +        _config = [['target', {'dev': 'tap0'}]]
> +        # add bridge info
> +        opt_br = options.get('opt_br')
> +        bridge = opt_br if opt_br else self.DEFAULT_BRIDGE
> +        _config.append(['source', {'bridge': bridge}])
> +        self.__generate_net_config_script(str(bridge))
> +        # add network configure script path
> +        opt_script = options.get('opt_script')
> +        script_path = opt_script if opt_script else self.QEMU_IFUP_PATH
> +        _config.append(['script', {'path': script_path}])
> +        config = [['interface', {'type': 'bridge'}, _config]]
> +        # set xml file
> +        parent = self.domain.find('devices')
> +        self.__write_config(parent, config)
> +
>      def add_vm_virtio_serial_channel(self, **options):
>          """
>          Options:
> @@ -524,7 +730,8 @@ class LibvirtKvm(VirtBase):
>          channel = ET.SubElement(devices, 'channel', {'type': 'unix'})
>          for opt in ['path', 'name']:
>              if opt not in options.keys():
> -                print "invalid virtio serial channel setting"
> +                msg = "invalid virtio serial channel setting"
> +                self.logger.error(msg)
>                  return
> 
>          ET.SubElement(
> --
> 2.21.0


      reply	other threads:[~2019-08-28 10:29 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-08-28  7:47 [dts] [PATCH V2 0/1] " yufengmx
2019-08-28  7:47 ` [dts] [PATCH V2 1/1] " yufengmx
2019-08-28 10:29   ` Tu, Lijuan [this message]

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=8CE3E05A3F976642AAB0F4675D0AD20E0BB06C1A@SHSMSX101.ccr.corp.intel.com \
    --to=lijuan.tu@intel.com \
    --cc=dts@dpdk.org \
    --cc=yinan.wang@intel.com \
    --cc=yufengx.mo@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).