Re: [PATCH v3 01/46] net: add qemu_{configure, create}_nic_device(), qemu_find_nic_info()

[Thomas Huth]

On 26/01/2024 15.16, David Woodhouse wrote:
> On Fri, 2024-01-26 at 12:10 +0100, Thomas Huth wrote:
> > > +/* "Please create a device, if you have a configuration for it" */
> > > +DeviceState *qemu_create_nic_device(const char *typename, bool match_default,
> > > +                                    const char *alias)
> > > +{
> > > +    NICInfo *nd = qemu_find_nic_info(typename, match_default, alias);
> > > +    DeviceState *dev;
> > > +
> > > +    if (!nd) {
> > > +        return NULL;
> > > +    }
> >
> > The qemu_check_nic_model() function that was used in some code that you
> > turned into qemu_create_nic_device() used to set:
> >
> >       if (!nd->model)
> >           nd->model = g_strdup(default_model);
> >
> > (in the qemu_find_nic_model() function that has been called by
> > qemu_check_nic_model())
> >
> > Should we do that also here to make sure that nd->model is not NULL afterwards?
>
> Good question, but I don't think we care. The qdev_set_nic_properties()
> function certainly doesn't propagate nd->model to anywhere.
>
> I renamed nd->model to nd->modelname in a patch shown below, just to be
> 100% sure I'm not missing any other code paths which might consume it.

Ok, thanks for checking! Maybe mention it in the patch description in v4, so 
that we've got it recorded somewhere that nd->model might be left at NULL 
afterwards, but that there are no further consumers, so it should be fine?

 Thomas