On Tue, May 14, 2024 at 12:45 PM Michal Prívozník <mprivozn(a)redhat.com> wrote:
On 5/13/24 15:55, Rayhan Faizel wrote:
> Previously, the network device hotplug logic would try to ensure only CCW or
> PCI addresses. With recent support for the usb-net model, this patch will
> ensure USB addresses for usb-net network devices.
>
> [Changes in v2]
> - Additional information in commit message
> - Relevant gitlab issue link (mostly already resolved but hotplugging is
> mentioned in one of the comments)
This changelog belongs ...
>
> Resolves:
https://gitlab.com/libvirt/libvirt/-/issues/14
> Signed-off-by: Rayhan Faizel <rayhan.faizel(a)gmail.com>
> ---
... here. Right after these three dashes. The changelong isn't necessary
helpful when reading git log, it's needed only for review.
> src/qemu/qemu_hotplug.c | 7 +++++--
> 1 file changed, 5 insertions(+), 2 deletions(-)
>
> diff --git a/src/qemu/qemu_hotplug.c b/src/qemu/qemu_hotplug.c
> index 774962b0df..3b39941780 100644
> --- a/src/qemu/qemu_hotplug.c
> +++ b/src/qemu/qemu_hotplug.c
> @@ -1159,8 +1159,11 @@ qemuDomainAttachNetDevice(virQEMUDriver *driver,
> goto cleanup;
> }
>
> - if (qemuDomainIsS390CCW(vm->def) &&
> - net->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI) {
> + if (net->model == VIR_DOMAIN_NET_MODEL_USB_NET) {
> + if (virDomainUSBAddressEnsure(priv->usbaddrs, &net->info) <
0)
> + goto cleanup;
> + } else if (qemuDomainIsS390CCW(vm->def) &&
> + net->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI) {
> net->info.type = VIR_DOMAIN_DEVICE_ADDRESS_TYPE_CCW;
> if (!(ccwaddrs = virDomainCCWAddressSetCreateFromDomain(vm->def)))
> goto cleanup;
Right. I was wondering what about the case when a new domain is defined,
whether we do fill in address in that case and we do. It's just that our
test case doesn't cover that case. I mean, there's
tests/qemuxmlconfdata/net-usb.xml and also
tests/qemuxmlconfdata/net-usb.x86_64-latest.xml but because the former
doesn't have any USB address, it causes qemuDomainAssignUSBAddresses()
to return early without assigning any addresses.
This is something I've also noticed. The auto-assigned USB address
does not show up in the x86-64-latest.xml file but appears later in
the args file. Some other examples like usb-audio, usb-mtp, usb disk,
etc. in the test cases seem to exhibit the same behaviour so I didn't
think too much of it. Though from virt-manager's POV, the USB address
does still appear correctly.
At any rate, your patch is correct.
Reviewed-by: Michal Privoznik <mprivozn(a)redhat.com>
Sorry for the double mail, forgot to reply to all.
--
Rayhan Faizel