Re: [PATCH 5/5] net: add MTD support to eth_platform_get_mac_address()

[Andrew Lunn]

On Thu, Jul 19, 2018 at 10:14:29AM +0200, Bartosz Golaszewski wrote:
> 2018-07-18 19:03 GMT+02:00 Andrew Lunn <andrew@xxxxxxx>:
> >> >> +#ifdef CONFIG_MTD
> >> >> +     /* NOTE: this should go away as soon as MTD gets nvmem support. */
> >> >> +     if (!addr) {
> >> >> +             struct mtd_info *mtd;
> >> >> +             int rv;
> >> >> +
> >> >> +             mtd = get_mtd_device_nm("MAC-Address");
> >> >
> >> > In order for this to go away, you need to keep backwards
> >> > compatibility. When using nvmem, you look for a cell called
> >> > "mac-address". Here you are looking for "MAC-Address". That is going
> >> > to make backwards compatibility harder. How do you plan to do it?
> >> >
> >> >    Andrew
> >>
> >> I'm trying to adjust to already existing users. The only user of
> >> get_mtd_device_nm() who calls it to read the MAC address registers a
> >> partition called "MAC-Address". We can't change it since it's visible
> >> from user space. In the future we'd just have to have a list of
> >> supported string that we'd use to do the nvmem lookup.
> >
> > Why not have the nvmem cell called "MAC-Address"? When you add nvmem
> > support to MTD, i assume you are going to map each MTD partition to an
> > nvmem cell?
> 
> Because all existing users of nvmem use "mac-address" as the name of
> this cell already. I guess we will need to live with both in this
> particular function.

So i'm not convinced this last patch is making things better. I would
prefer if it was dropped for the moment. Wait until MTD via nvmem is
actually implemented and there is a concrete concept of how a MAC
address would be looked up without having lots of ugly code.

	Andrew