[dpdk-dev] Questions about vm2vm vhost-user/virtio-net test

[Min Hu (Connor)]

Hi, all,
	I got the answer: testpmd fwd mode should be set "io", then ping
OK.

	IO fwd mode, it will not change packet MAC address info, and
ping OK
	MAC fwd mode, it will change packet MAC address info, like:
	"rte_ether_addr_copy(&peer_eth_addrs[fs->peer_addr],
				&eth_hdr->d_addr);
	rte_ether_addr_copy(&ports[fs->tx_port].eth_addr,
				&eth_hdr->s_addr)"
	Then, ping failed.

	So, everyone, I got one question: how could this happen?
	router change packet src MAC and dst MAC, but ping OK,
	testpmd vhost, treated as switch, also changes packert src MAC
	and dst MAC, but ping failed ?	


在 2021/9/29 20:09, Min Hu (Connor) 写道:
> Hi, Coquelin, Chenbo, all,
>      I  want to seek help about vm2vm vhost-user/virtio-net test from
> you.
>      When I set up vm2vm vhost-user/virtio-net test, I cannot ping
> vm2 in vm1. That is, ping failed between vm1 and vm2.
> 
>      Detailed description are as flows:
> host configuration:
> Linux localhost 5.14.0-rc4+ #1 SMP PREEMPT Sun Sep 26 16:52:13 CST 2021 
> aarch64 aarch64 aarch64 GNU/Linux
> 
> 1, set up vhost on host
> cd /home/humin/virtio_test
> rm -rf ./vhost-net*
> 
> /usr/app/testpmd -l 2-4 -n 4 --no-pci --file-prefix=vhost --vdev 
> 'net_vhost0,iface=vhost-net0,queues=1' --vdev 
> 'net_vhost1,iface=vhost-net1,queues=1'  -- -i --nb-cores=2 --txd=1024 
> --rxd=1024
> 
> testpmd> set fwd mac
> testpmd> start
> 
> 2, create vm1 and vm2
> taskset -c 13 qemu-system-aarch64 \
> -name us-vhost-vm1 \
> -kernel /home/humin/virtio_test/Image \
> -initrd /home/humin/virtio_test/rootfs.cpio.gz \
> -machine virt,gic-version=3 -nographic \
> -cpu host -enable-kvm -m 4096 \
> -object 
> memory-backend-file,id=mem,size=4096M,mem-path=/dev/hugepages,share=on \
> -numa node,memdev=mem \
> -mem-prealloc \
> -smp cores=4,sockets=1 \
> -monitor unix:/tmp/vm2_monitor.sock,server,nowait \
> -net user,hostfwd=tcp:127.0.0.1:6004-:22 \
> -chardev socket,id=char0,path=/home/humin/virtio_test/vhost-net0 \
> -netdev type=vhost-user,id=mynet1,chardev=char0,vhostforce \
> -device 
> virtio-net-pci,mac=52:54:00:00:00:01,netdev=mynet1,mrg_rxbuf=on,csum=on,guest_csum=on,guest_ecn=on 
> \
> -vnc :15
> 
> 
> taskset -c 15 qemu-system-aarch64 \
> -name us-vhost-vm2 \
> -kernel /home/humin/virtio_test/Image \
> -initrd /home/humin/virtio_test/rootfs.cpio.gz \
> -machine virt,gic-version=3 -nographic \
> -cpu host -enable-kvm -m 4096 \
> -object 
> memory-backend-file,id=mem,size=4096M,mem-path=/dev/hugepages,share=on \
> -numa node,memdev=mem \
> -mem-prealloc \
> -smp cores=4,sockets=1 \
> -monitor unix:/tmp/vm2_monitor.sock,server,nowait \
> -net user,hostfwd=tcp:127.0.0.1:6005-:22 \
> -chardev socket,id=char1,path=/home/humin/virtio_test/vhost-net1 \
> -netdev type=vhost-user,id=mynet2,chardev=char1,vhostforce \
> -device 
> virtio-net-pci,mac=52:54:00:00:00:02,netdev=mynet2,mrg_rxbuf=on,csum=on,guest_csum=on,guest_ecn=on 
> \
> -vnc :16
> 
> 
> 3, do test
> in vm1:
> ifconfig eth0 up
> ifconfig eth0 1.1.1.2
> arp -s 1.1.1.8 52:54:00:00:00:02
> 
> in vm2:
> ifconfig eth0 up
> ifconfig eth0 1.1.1.8
> arp -s 1.1.1.2 52:54:00:00:00:01
> 
> Then in vm1:
> ping 1.1.1.8
> 
> the result:
> no icmp reply, ping failed. same for vm2.
> 
> 4, Try to debug
> using tcpdump to capture packets in vm2,  we can get only the ICMP
> request packet which is from vm1, but no ICMP reply.
> It looks like ping task does not send ICMP reply, but it is just guess
> for it.
> 
> BTW, I also use iperf test:
> UDP packet test:
> vm1:
> iperf -s  -i 1
> 
> vm2:
> iperf -c 1.1.1.2 -u -i 1 -t 60
> 
> Then vm1 can get the packet. Same result if vm1 is used as client.
> 
> TCP packet test:
> vm1:
> iperf -s  -i 1
> 
> vm2:
> iperf -c 1.1.1.2 -i 1 -t 60
> 
> Then vm1 can NOT get the packet. Same result if vm1 is used as client.
> 
> 5, reference
> I refre to the website
> https://doc.dpdk.org/dts/test_plans/vm2vm_virtio_net_perf_test_plan.html
> 
> In 199.7. Test Case 5:
> scp test success.
> 
> But my scp test also failed, any one could help me?  thanks very much.
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> .