On Fri, Mar 17, 2023 at 05:30:38PM +0100, Álvaro Fernández Rojas wrote:
> El vie, 17 mar 2023 a las 17:27, Andrew Lunn (<andrew@xxxxxxx>) escribió:
> >
> > > > The proposed solution is too radical for a problem that was not properly
> > > > characterized yet, so this patch set has my temporary NACK.
> > >
> > > Forgive me, but why do you consider this solution too radical?
> >
> > I have to agree with Vladimir here. The problem is not the driver, but
> > when the driver is instantiated. It seems radical to remove a driver
> > just because it loads at the wrong time. Ideally you want the driver
> > to figure out now is not a good time and return -EPROBE_DEFER, because
> > a resource it requires it not available.
>
> Ok, I'm open to suggestions.
> Any ideas on how exactly to figure out when it's a good time to probe
> or return -EPROBE_DEFER instead?
Vladimir already said:
> > > We need to know what resource belonging to the switch is it that the
> > > MDIO mux needs.
Please answer that question. Once we know what the resource is, we can
look at how to export it to the mux in a way that is safe.
Andrew
