[Bug 1863200] [NEW] Reconnect failed with loopback virtio1.1 server mode test

[xuan]

Public bug reported:

Issue discription：
Packed ring server mode is a new feature to enable the virtio-user or 
virtio-pmd(in VM) as the server, vhost as the client, then when the vhost-user 
is killed then re-launched, the vhost-user can connect back to 
virtio-user/virtio-pmd again. Test with dpdk 20.02 ,virtio-pmd loopback 
reconnect from vhost-user failed.

Test Environment:
DPDK version: DPDK v20.02
Other software versions: virtio1.1
Qemu versions:4.2.0
OS: Linux 4.15.0-20-generic
Compiler: gcc (Ubuntu 7.5.0-3ubuntu1~18.04) 7.5.0
Hardware platform: R2208WFTZSR.

The reproduce step is :
Test Case: vhost-user/virtio-pmd loopback reconnect from vhost-user
===============================================================
Flow: Vhost-user --> Virtio --> Vhost-user

1. Launch vhost-user with client mode by below commands::

    ./testpmd -c 0x30 -n 4 --socket-mem 1024,1024 --legacy-mem --vdev 
'eth_vhost0,iface=/tmp/vhost-net,client=1,queues=1' -- -i --nb-cores=1
    testpmd>set fwd mac

2. Start VM with 1 virtio device, and set the qemu as server mode::

    ./qemu-system-x86_64 -name vm2 -enable-kvm -cpu host -smp 100 -m 8G \
    -object memory-backend-file,id=mem,size=8192M,mem-path=/mnt/huge,share=on \
    -numa node,memdev=mem -mem-prealloc -drive 
file=/home/xuan/dpdk_project/shell/u18.img  \
    -chardev socket,path=/tmp/vm2_qga0.sock,server,nowait,id=vm2_qga0 -device 
virtio-serial \
    -device virtserialport,chardev=vm2_qga0,name=org.qemu.guest_agent.2 
-daemonize \
    -monitor unix:/tmp/vm2_monitor.sock,server,nowait -net 
nic,macaddr=00:00:00:08:e8:aa,addr=1f \
    -net user,hostfwd=tcp:127.0.0.1:6002-:22 \
    -chardev socket,id=char0,path=/tmp/vhost-net,server \
    -netdev type=vhost-user,id=netdev0,chardev=char0,vhostforce \
    -device 
virtio-net-pci,netdev=netdev0,mac=52:54:00:00:00:01,mrg_rxbuf=on,rx_queue_size=1024,tx_queue_size=1024,packed=on
 \
    -vnc :10

3. On VM, bind virtio net to igb_uio and run testpmd::

    ./testpmd -c 0x3 -n 4 -- -i --nb-cores=1 --txd=1024 --rxd=1024
    testpmd>set fwd mac
    testpmd>start

4. Send packets by vhost-user, check if packets can be RX/TX with
virtio-pmd::

    testpmd>start tx_first 32
    testpmd>show port stats all

5. On host, quit vhost-user, then re-launch the vhost-user with below
command::

    testpmd>quit
    ./testpmd -c 0x30 -n 4 --socket-mem 1024,1024 --legacy-mem --vdev 
'eth_vhost0,iface=/tmp/vhost-net,client=1,queues=1' -- -i --nb-cores=1
    testpmd>set fwd mac
    testpmd>start tx_first 32

6. Check if the reconnection can work, still send packets by vhost-user,
check if packets can be RX/TX with virtio-pmd::

    testpmd>show port stats all

Expected result::

After the vhost-user is killed then re-launched, the VM can connect back
to vhost-user again with throughput.

Real result::

After the vhost-user is killed then re-launched, no throughput with PVP.

Analysis::

QEMU has its own way to handle reconnect function for virtio server
mode. However, for packed ring, when reconnecting to virtio, vhost
cannot get the status of descriptors via the descriptor ring. This bug
is caused since the reconnection for packed ring need additional reset
operation.

** Affects: qemu
     Importance: Undecided
         Status: New


** Tags: mode packed ring server

-- 
You received this bug notification because you are a member of qemu-
devel-ml, which is subscribed to QEMU.
https://bugs.launchpad.net/bugs/1863200

Title:
  Reconnect failed with loopback virtio1.1 server mode test

Status in QEMU:
  New

Bug description:
  Issue discription：
  Packed ring server mode is a new feature to enable the virtio-user or 
virtio-pmd(in VM) as the server, vhost as the client, then when the vhost-user 
is killed then re-launched, the vhost-user can connect back to 
virtio-user/virtio-pmd again. Test with dpdk 20.02 ,virtio-pmd loopback 
reconnect from vhost-user failed.

  Test Environment:
  DPDK version: DPDK v20.02
  Other software versions: virtio1.1
  Qemu versions:4.2.0
  OS: Linux 4.15.0-20-generic
  Compiler: gcc (Ubuntu 7.5.0-3ubuntu1~18.04) 7.5.0
  Hardware platform: R2208WFTZSR.

  The reproduce step is :
  Test Case: vhost-user/virtio-pmd loopback reconnect from vhost-user
  ===============================================================
  Flow: Vhost-user --> Virtio --> Vhost-user

  1. Launch vhost-user with client mode by below commands::

      ./testpmd -c 0x30 -n 4 --socket-mem 1024,1024 --legacy-mem --vdev 
'eth_vhost0,iface=/tmp/vhost-net,client=1,queues=1' -- -i --nb-cores=1
      testpmd>set fwd mac

  2. Start VM with 1 virtio device, and set the qemu as server mode::

      ./qemu-system-x86_64 -name vm2 -enable-kvm -cpu host -smp 100 -m 8G \
      -object memory-backend-file,id=mem,size=8192M,mem-path=/mnt/huge,share=on 
\
      -numa node,memdev=mem -mem-prealloc -drive 
file=/home/xuan/dpdk_project/shell/u18.img  \
      -chardev socket,path=/tmp/vm2_qga0.sock,server,nowait,id=vm2_qga0 -device 
virtio-serial \
      -device virtserialport,chardev=vm2_qga0,name=org.qemu.guest_agent.2 
-daemonize \
      -monitor unix:/tmp/vm2_monitor.sock,server,nowait -net 
nic,macaddr=00:00:00:08:e8:aa,addr=1f \
      -net user,hostfwd=tcp:127.0.0.1:6002-:22 \
      -chardev socket,id=char0,path=/tmp/vhost-net,server \
      -netdev type=vhost-user,id=netdev0,chardev=char0,vhostforce \
      -device 
virtio-net-pci,netdev=netdev0,mac=52:54:00:00:00:01,mrg_rxbuf=on,rx_queue_size=1024,tx_queue_size=1024,packed=on
 \
      -vnc :10

  3. On VM, bind virtio net to igb_uio and run testpmd::

      ./testpmd -c 0x3 -n 4 -- -i --nb-cores=1 --txd=1024 --rxd=1024
      testpmd>set fwd mac
      testpmd>start

  4. Send packets by vhost-user, check if packets can be RX/TX with
  virtio-pmd::

      testpmd>start tx_first 32
      testpmd>show port stats all

  5. On host, quit vhost-user, then re-launch the vhost-user with below
  command::

      testpmd>quit
      ./testpmd -c 0x30 -n 4 --socket-mem 1024,1024 --legacy-mem --vdev 
'eth_vhost0,iface=/tmp/vhost-net,client=1,queues=1' -- -i --nb-cores=1
      testpmd>set fwd mac
      testpmd>start tx_first 32

  6. Check if the reconnection can work, still send packets by vhost-
  user, check if packets can be RX/TX with virtio-pmd::

      testpmd>show port stats all

  Expected result::

  After the vhost-user is killed then re-launched, the VM can connect
  back to vhost-user again with throughput.

  Real result::

  After the vhost-user is killed then re-launched, no throughput with
  PVP.

  Analysis::

  QEMU has its own way to handle reconnect function for virtio server
  mode. However, for packed ring, when reconnecting to virtio, vhost
  cannot get the status of descriptors via the descriptor ring. This bug
  is caused since the reconnection for packed ring need additional reset
  operation.

To manage notifications about this bug go to:
https://bugs.launchpad.net/qemu/+bug/1863200/+subscriptions