Re: [PATCH] virtio-net : Add check for VIRTIO_NET_F_MAC

[Cindy Lu]

On Wed, Sep 29, 2021 at 9:36 PM Michael S. Tsirkin <mst@redhat.com> wrote:
>
> On Wed, Sep 29, 2021 at 08:08:40PM +0800, Cindy Lu wrote:
> > On Wed, Sep 29, 2021 at 6:07 PM Michael Tokarev <mjt@tls.msk.ru> wrote:
> > >
> > > 29.09.2021 09:52, Cindy Lu wrote:
> > > > For vdpa device, if the host support VIRTIO_NET_F_MAC
> > > > we need to read the mac address from hardware, so need
> > > > to check this bit, the logic is
> > > > 1 if the host support VIRTIO_NET_F_MAC and the mac address
> > > >     is correct, qemu will use the mac address in hardware
> > > > 2.if the host not support , qemu will use the mac from cmdline
> > >
> > > So if hw supports NET_F_MAC, cmdline-provided parameter will
> > > silently be ignored?
> > >
> > yes, this is based on the virtio spec, you can check this document in
> > 5.1.5 Device Initialization
> > https://docs.oasis-open.org/virtio/virtio/v1.1/csprd01/virtio-v1.1-csprd01.html
>
> Maybe use the hw mac if mac is not provided? If provided
> make sure the command line matches the hardware, and fail
> otherwise?
>

so here come to the final question. which mac address has the higher priority?
I think the NET_F_MAC bit means the hw mac address > command-line address.
if the hw drivers want to change this. they can simply remove this bit.


> > Also, this check it only working for vdpa device
> > > s/host not support/host does not support this feature/
> > Thanks , will fix this
> > >
> > > > 3.if the cmdline not provide mac address, qemu will use radam mac
> > > > address
> > >
> > > s/not/does not/
> > > s/radam/random/
> > >
> > thanks, will fix this
> > > Thanks,
> > >
> > > /mjt
> > >
>