使用OVS DPDK (by quqi99)_综合

作者：张华发表于：2016-04-07
版权声明：可以任意转载，转载时请务必以超链接形式标明文章原始出处和作者信息及本版权声明

( http://blog.csdn.net/quqi99 )

硬件要求

网卡得支持DPDK，见：http://dpdk.org/doc/nics

CPU得支持DPDK, 测试命令：cat /proc/cpuinfo |grep pdpe1gb

不一定非要支持DPDK硬件的网卡才能做实验，因为DPDK也支持virtio dpdk driver.

打开大页支持

hua@node1:~$ cat /etc/default/grub |grep GRUB_CMDLINE_LINUX
GRUB_CMDLINE_LINUX_DEFAULT="quiet splash intel_iommu=on pci=assign-busses"
GRUB_CMDLINE_LINUX="transparent_hugepage=never hugepagesz=2M hugepages=64 default_hugepagesz=2M"

vi /etc/fstab
nodev /mnt/huge hugetlbfs pagesize=2MB 0 0

sudo update-grub
sudo mkdir -p /mnt/huge
sudo reboot
hua@node1:~$ cat /proc/meminfo |grep HugePages_
HugePages_Total: 64
HugePages_Free: 64
HugePages_Rsvd: 0

HugePages_Surp: 0

hua@node1:~$ grep Hugepagesize /proc/meminfo
Hugepagesize: 2048 kB

配置网卡使用uio_pci_generic驱动, #注意：实验发现，对于sr-iov网卡，此处必须是PF，PF试验成功，VF试验不成功。所以虚机可以使用PF去做DPDK port，也可以直接直通方式使用VF。

至于PF的驱动，可以使用ixgbe，也可以使用DPDK PF driver；对于VF的驱动，可以使用ixgbefv，也可以使用DPDK VF driver. 如果要使用DPDK PF driver需在grub中加上iommu=pt.

它们之间也可以通信：

hua@node1:~$ sudo modprobe uio_pci_generic

hua@node1:~$ sudo dpdk_nic_bind --status

Network devices using DPDK-compatible driver
============================================
<none>

Network devices using kernel driver
===================================
0000:00:19.0 'Ethernet Connection I217-V' if=eth0 drv=e1000e unused=uio_pci_generic *Active*
0000:05:00.0 '82576 Gigabit Network Connection' if=eth1 drv=igb unused=uio_pci_generic *Active*
0000:05:00.1 '82576 Gigabit Network Connection' if=eth2 drv=igb unused=uio_pci_generic *Active*

Other network devices
=====================
0000:06:10.0 '82576 Virtual Function' unused=uio_pci_generic
0000:06:10.1 '82576 Virtual Function' unused=uio_pci_generic
0000:06:10.2 '82576 Virtual Function' unused=uio_pci_generic
0000:06:10.3 '82576 Virtual Function' unused=uio_pci_generic

hua@node1:~$ sudo dpdk_nic_bind -b uio_pci_generic 0000:05:00.0
Routing table indicates that interface 0000:05:00.0 is active. Not modifying

hua@node1:~$ sudo ifconfig eth1 down
hua@node1:~$ sudo dpdk_nic_bind -b uio_pci_generic 0000:05:00.0

hua@node1:~$ sudo dpdk_nic_bind --status
Network devices using DPDK-compatible driver
============================================
0000:05:00.0 '82576 Gigabit Network Connection' drv=uio_pci_generic unused=

Network devices using kernel driver
===================================
0000:00:19.0 'Ethernet Connection I217-V' if=eth0 drv=e1000e unused=uio_pci_generic *Active*
0000:05:00.1 '82576 Gigabit Network Connection' if=eth2 drv=igb unused=uio_pci_generic *Active*

Other network devices
=====================
0000:06:10.1 '82576 Virtual Function' unused=uio_pci_generic
0000:06:10.3 '82576 Virtual Function' unused=uio_pci_generic

配置OVS使用DPDK

hua@node1:~$ cat /etc/default/openvswitch-switch |grep DPDK
DPDK_OPTS='--dpdk -c 0x1 -n 2'

hua@node1:~$ sudo update-alternatives --set ovs-vswitchd /usr/lib/openvswitch-switch-dpdk/ovs-vswitchd-dpdk
update-alternatives: using /usr/lib/openvswitch-switch-dpdk/ovs-vswitchd-dpdk to provide /usr/sbin/ovs-vswitchd (ovs-vswitchd) in manual mode

hua@node1:~$ sudo stop openvswitch-switch; sudo start openvswitch-switch

hua@node1:~$ cat /var/log/openvswitch/ovs-ctl.log |grep dpdk

有一个bug设置了DPDK_OPTS不生效，见：https://bugs.launchpad.net/ubuntu/+source/openvswitch-dpdk/+bug/1547463，做了这些改动：

NOTE: (还有一个bug - https://bugs.launchpad.net/ubuntu/+source/openvswitch/+bug/1836713 )

vi /etc/default/openvswitch-switch
DPDK_OPTS='--dpdk -c 0x1 -n 4'

vi /etc/init/openvswitch-switch.conf
# To enable openvswitch-dpdk
if test X"$DPDK_OPTS" != X; then
set "$@" DPDK_OPTS="$DPDK_OPTS"
fi

vi /usr/share/openvswitch/scripts/ovs-ctl
set ovs-vswitchd
if test X"$DPDK_OPTS" != X; then
set "$@" $DPDK_OPTS --
fi
set "$@" unix:"$DB_SOCK"

但在/var/log/openvswitch/ovs-ctl.log中仍然发现了下列错误，估计是ovs-vswitchd (Open vSwitch) 2.4.0版本（dpkg -l openvswitch-switch-dpdk）不支持--dpdk-opts参数吧。

/usr/share/openvswitch/scripts/ovs-ctl: unknown option "--dpdk-opts=-c 0x1 -n 4" (use --help for help)

为避免上述ovs的bug，我们改用下列三个命令启动：

sudo restart openvswitch-switch

sudo killall ovs-vswitchd

sudo ovs-vswitchd --dpdk -c 0x1 -n 2 -- unix:/var/run/openvswitch/db.sock --pidfile --detach

sudo ovs-vswitchd --dpdk -c 0x1 -n 2 --socket-mem 128,0 -- unix:/var/run/openvswitch/db.sock -vconsole:emer -vsyslog:err -vfile:info --mlockall --no-chdir --log-file=/var/log/openvswitch/ovs-vswitchd.log --pidfile=/var/run/openvswitch/ovs-vswitchd.pid --detach --monitor

但仍然有下列错误。估计是保留的大页数目太少所致，后来将GRUB_CMDLINE_LINUX="transparent_hugepage=never hugepagesz=2M hugepages=64 default_hugepagesz=2M"中的hugepages从4改成64这个错误就翻过去了。

hua@node1:~$ sudo ovs-vswitchd --dpdk -c 0x1 -n 2 -- unix:/var/run/openvswitch/db.sock --pidfile --detach
2016-04-07T11:48:03Z|00001|dpdk|INFO|No -cuse_dev_name provided - defaulting to vhost-net
EAL: Detected lcore 0 as core 0 on socket 0
EAL: Detected lcore 1 as core 1 on socket 0
EAL: Detected lcore 2 as core 2 on socket 0
EAL: Detected lcore 3 as core 3 on socket 0
EAL: Support maximum 128 logical core(s) by configuration.
EAL: Detected 4 lcore(s)
EAL: No free hugepages reported in hugepages-1048576kB
EAL: Setting up memory...
EAL: Ask a virtual area of 0x200000 bytes
EAL: Virtual area found at 0x7f8609a00000 (size = 0x200000)
EAL: Ask a virtual area of 0x400000 bytes
EAL: Virtual area found at 0x7f8609400000 (size = 0x400000)
EAL: Ask a virtual area of 0x200000 bytes
EAL: Virtual area found at 0x7f8609000000 (size = 0x200000)
EAL: Requesting 4 pages of size 2MB from socket 0
EAL: rte_eal_common_log_init(): cannot create log_history mempool
PANIC in rte_eal_init():
Cannot init logs
7: [ovs-vswitchd() [0x40b053]]
6: [/lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xf5) [0x7f860acc9ec5]]
5: [ovs-vswitchd() [0x40a059]]
4: [ovs-vswitchd() [0x639db1]]
3: [ovs-vswitchd() [0x43f7ad]]
2: [ovs-vswitchd() [0x407bad]]
1: [ovs-vswitchd() [0x444e58]]

测试

sudo ovs-vsctl ovsbr0 br-int
sudo ovs-vsctl set bridge ovsbr0 datapath_type=netdev

sudo ovs-vsctl add-port ovsbr0 dpdk0 -- set Interface dpdk0 type=dpdk #Port name shoud begin with dpdk

给虚机创建普通ovs port:

sudo ovs-vsctl add-port ovsbr0 intP1 -- set Interface intP1 type=internal
sudo ip addr add 192.168.10.129/24 dev intP1
sudo ip link set dev intP1 up
sudo tcpdump -i intP1?

或给虚机创建vhost-user port:

sudo ovs-vsctl add-port ovsbr0 vhost-user2 -- set Interface vhost-user2 type=dpdkvhostuser

虚机使用vhost-user port

sudo qemu-system-x86_64 -enable-kvm -m 128 -smp 2 \
-chardev socket,id=char0,path=/var/run/openvswitch/vhost-user1 \
-netdev type=vhost-user,id=mynet1,chardev=char0,vhostforce \
-device virtio-net-pci,netdev=mynet1,mac=52:54:00:02:d9:01 \
-object memory-backend-file,id=mem,size=128M,mem-path=/mnt/huge,share=on \
-numa node,memdev=mem -mem-prealloc \
-net user,hostfwd=tcp::10021-:22 -net nic \
/bak/images/openstack_demo.img

Inside VM: sudo ifconfig eth0 192.168.9.108/24

使用vhost-user方式可如下图所示：

dpdk-virtio-net

与Neutron的集成

nova已经支持vhost-user与hugepages特性，显然在nova.conf应该打开。

#https://bugs.launchpad.net/neutron/+bug/1506127

# then the OVS ML2 driver will use the vhost-user VIF type and pass the necessary binding details to use OVS+DPDK and vhost-user sockets.

[OVS]
datapath_type=netdev
vhostuser_socket_dir=/var/run/openvswitch
[libvirt]
use_hugepages = True

# https://review.opendev.org/c/openstack/neutron/+/237264/def get_supported_vif_type(self, agent):caps = agent['configurations'].get('ovs_capabilities', {})if (any(x in caps.get('iface_types', []) for xin [a_const.OVS_DPDK_VHOST_USER,a_const.OVS_DPDK_VHOST_USER_CLIENT]) andagent['configurations'].get('datapath_type') ==a_const.OVS_DATAPATH_NETDEV):return portbindings.VIF_TYPE_VHOST_USER                                                                                                                                    return self.vif_type$ sudo ovs-vsctl list Open_vSwitch |grep iface_types
iface_types         : [geneve, gre, internal, lisp, patch, stt, system, tap, vxlan]

另外，应该手工创建上述的ovsbr0，然后在/etc/neutron/plugins/ml2/ml2_conf.ini指定使用ovsbr0作为provider network.

[ovs]
bridge_mappings = default:ovsbr0

dpdk与neutron集成的工程叫networking-ovs-dpdk，在devstack加下列一行可使用devstack创建ovs dpdk环境。

enable_plugin networking-ovs-dpdk https://github.com/openstack/networking-ovs-dpdk master

整个运行命令

sudo add-apt-repository cloud-archive:mitaka

sudo apt-get install openvswitch-switch-dpdk

sudo modprobe vfio-pci
sudo chmod a+x /dev/vfio
sudo chmod 0666 /dev/vfio/*
sudo ifconfig eth1 down
sudo dpdk_nic_bind -b vfio-pci eth1 #注意：实验发现，对于sr-iov网卡，此处必须是PF，PF试验成功，VF试验不成功。

find /sys/kernel/iommu_groups/ -type l
sudo sysctl -w vm.nr_hugepages=256
cat /proc/meminfo |grep HugePages_

sudo restart openvswitch-switch
sudo update-alternatives --set ovs-vswitchd /usr/lib/openvswitch-switch-dpdk/ovs-vswitchd-dpdk
sudo killall ovs-vswitchd
sudo ovs-vswitchd --dpdk -c 0x1 -n 2 --socket-mem 128,0 -- unix:/var/run/openvswitch/db.sock -vconsole:emer -vsyslog:err -vfile:info --mlockall --no-chdir --log-file=/var/log/openvswitch/ovs-vswitchd.log --pidfile=/var/run/openvswitch/ovs-vswitchd.pid --detach --monitor
ps -ef|grep dpdk

sudo qemu-system-x86_64 -enable-kvm -m 128 -smp 2 \
-chardev socket,id=char0,path=/var/run/openvswitch/vhost-user1 \
-netdev type=vhost-user,id=mynet1,chardev=char0,vhostforce \
-device virtio-net-pci,netdev=mynet1,mac=52:54:00:02:d9:01 \
-object memory-backend-file,id=mem,size=128M,mem-path=/mnt/huge,share=on \
-numa node,memdev=mem -mem-prealloc \
-net user,hostfwd=tcp::10021-:22 -net nic \
/bak/images/openstack_demo.img

Inside VM: sudo ifconfig eth0 192.168.9.108/24

附录1,编译DPDK源码并运行单个应用

git clone git://dpdk.org/dpdk
export DPDK_DIR=/bak/linux/dpdk
export DPDK_BUILD=$DPDK_DIR/x86_64-ivshmem-linuxapp-gcc/
export RTE_TARGET=x86_64-ivshmem-linuxapp-gcc
export EXTRA_CFLAGS="-ggdb -ffunction-sections -O0" #设置调试选项
echo 'CONFIG_RTE_BUILD_COMBINE_LIBS=y' >> ./config/common_linuxapp #编译成单个文件

echo 'CONFIG_RTE_LIBRTE_VHOST=y' >> ./config/common_linuxapp

sudo make config T=x86_64-ivshmem-linuxapp-gcc

sudo make install T=x86_64-ivshmem-linuxapp-gcc #使用共享内存ivshmem模式

cd x86_64-ivshmem-linuxapp-gcc/
EXTRA_CFLAGS="-g -O0" make -j4

sudo modprobe uio
sudo insmod x86_64-ivshmem-linuxapp-gcc/kmod/igb_uio.ko #DPDK只能使用igb_uio和vfio-pci两种驱动

cd lib/librte_vhost/eventfd_link/
make
sudo insmod eventfd_link.ko
sudo python tools/dpdk_nic_bind.py --status
sudo python tools/dpdk_nic_bind.py -b igb_uio 0000:05:00.0
#sudo python tools/dpdk_nic_bind.py -b uio_pci_generic 0000:05:00.0

cd examples/helloworld/
make
sudo ./build/helloworld -c 0xf -n 4 --socket-mem=32

gdb ./build/helloworld
(gdb) run -c 0xf -n 4 --socket-mem=32

#integrate ovs with dpdk

git clone https://github.com/openvswitch/ovs.git
cd /bak/linux/ovs
export OVS_DIR=`pwd`
./boot.sh
./configure --with-dpdk="$DPDK_DIR/x86_64-ivshmem-linuxapp-gcc/" --enable-ssl CFLAGS="-g -O0"
make 'CFLAGS=-g -O0 -march=native' -j4

参考

[1] https://software.intel.com/en-us/blogs/2015/06/09/building-vhost-user-for-ovs-today-using-dpdk-200

[2] http://dpdk.readthedocs.org/en/v2.2.0/linux_gsg/index.html

[3] http://blog.csdn.net/xy010902100449/article/details/47282937

[4] http://openvswitch.org/support/dist-docs/INSTALL.DPDK.md.txt

[5] https://software.intel.com/en-us/articles/using-open-vswitch-with-dpdk-for-inter-vm-nfv-applications

[6] https://raw.githubusercontent.com/xsited/ssg/master/scripts/build_ovs_dpdk.sh

[7] http://www.ran-lifshitz.com/2015/06/17/open-vswitch-netdev-dpdk-with-vhost-user-support/

[8] DPDK编程指南

[9] set up ovn based sr-iov test env - https://zhhuabj.blog.csdn.net/article/details/116893909

[10] 通过源码安装dpdk并运行examples - https://zhhuabj.blog.csdn.net/article/details/119610026

使用OVS DPDK (by quqi99)

作者：张华 发表于：2016-04-07 版权声明：可以任意转载，转载时请务必以超链接形式标明文章原始出处和作者信息及本版权声明

硬件要求

打开大页支持

配置OVS使用DPDK

测试

附录1,编译DPDK源码并运行单个应用

参考

作者：张华发表于：2016-04-07
版权声明：可以任意转载，转载时请务必以超链接形式标明文章原始出处和作者信息及本版权声明