[PATCH v1 2/2] dts: add virtio forwarding test suite
Dean Marx
dmarx at iol.unh.edu
Fri Oct 3 20:14:37 CEST 2025
> Would it be worth introducing a TextParser class that will actually
> parse the forwarding stats and present them properly to the test?
Yes I think that would be useful moving forward, I'll add one to the
next version.
<snip>
> > + with TestPmdShell(
> > + prefix="vhost",
> > + no_pci=True,
> > + memory_channels=4,
> > + vdevs=[VirtualDevice("eth_vhost0,iface=/tmp/vhost-net,client=1")],
> > + ) as vhost:
> > + with TestPmdShell(
> > + prefix="virtio",
> > + no_pci=True,
> > + memory_channels=4,
> > + vdevs=[
> > + VirtualDevice(
> > + "net_virtio_user0,mac=00:01:02:03:04:05,path=/tmp/vhost-net,server=1"
> > + )
> > + ],
> > + ) as virtio:
> > + pass
> > + # end session and reconnect
> how could this is launched twice? Could use some explanation here.
Essentially the way it works is by launching a vhost-user testpmd
session, which waits for a virtio-user testpmd session to connect to:
EAL: Detected CPU lcores: 32
EAL: Detected NUMA nodes: 2
EAL: Detected static linkage of DPDK
EAL: Multi-process socket /var/run/dpdk/virtio_27338_20251003172250/mp_socket
EAL: Selected IOVA mode 'VA'
EAL: VFIO support initialized
VIRTIO_DRIVER: vhost_user_start_server(): (/tmp/vhost-net) waiting for
client connection...
Then, when the virtio-user session is launched, it connects to the
vhost-user socket /tmp/vhost-net:
VHOST_CONFIG: (/tmp/vhost-net) connected
...
VHOST_CONFIG: (/tmp/vhost-net) new device status(0x0000000f):
VHOST_CONFIG: (/tmp/vhost-net) -RESET: 0
VHOST_CONFIG: (/tmp/vhost-net) -ACKNOWLEDGE: 1
VHOST_CONFIG: (/tmp/vhost-net) -DRIVER: 1
VHOST_CONFIG: (/tmp/vhost-net) -FEATURES_OK: 1
VHOST_CONFIG: (/tmp/vhost-net) -DRIVER_OK: 1
VHOST_CONFIG: (/tmp/vhost-net) -DEVICE_NEED_RESET: 0
VHOST_CONFIG: (/tmp/vhost-net) -FAILED: 0
By launching the virtio-user session through the context manager, then
closing it and starting a new session with a new context manager, the
original virtio-user session is closed, disconnecting the client. The
test case is verifying that the vhost-user socket can support multiple
sequential client connections.
> > + with TestPmdShell(
> > + prefix="virtio",
> > + no_pci=True,
> > + memory_channels=4,
> > + vdevs=[
> > + VirtualDevice(
> > + "net_virtio_user0,mac=00:01:02:03:04:05,path=/tmp/vhost-net,server=1"
> > + )
> Wouldn't it make more sense to store the vdev in a variable and re-use
> it?
Definitely, I'll fix that
<snip>
> > + self.sut_node = self._ctx.sut_node
> > + if not isinstance(self._ctx.sut_node.main_session, LinuxSession):
> surely you could just use the `self.sut_node` you've just created :D
Good catch thanks
I'll send out a new version within the next week addressing these
points. Thanks for the review!
More information about the dev
mailing list