Re: [PATCH 0/2] rewrite mtk-quadspi spi-nor driver with spi-mem

From: Chuanhong Guo
Date: Tue Feb 25 2020 - 20:31:58 EST


Hi!

On Wed, Feb 26, 2020 at 1:31 AM Rob Herring <robh@xxxxxxxxxx> wrote:
>
> On Thu, Feb 20, 2020 at 07:58:06AM +0800, Chuanhong Guo wrote:
> > Hi!
> >
> > On Tue, Feb 18, 2020 at 8:55 PM Mark Brown <broonie@xxxxxxxxxx> wrote:
> > > This is an ABI break so you shouldn't be doing this, if the existing
> > > binding works it should continue to work.
> >
> > The missing spi-max-frequency is the only part preventing old
> > device tree to work with this driver.
> > If the goal is to make existing dt binding work, I could patch dt using
> > of_add_property in v2. I saw similar device tree patching for legacy
> > bindings in pinctrl-single driver.

I just noticed that of_add_property isn't a exported symbol, which means that
device tree patching isn't possible unless driver is builtin.

>
> You should should really only need 'spi-max-frequency' if the max freq
> is less than the minimum of the host and device max freq.

But current spi framework forces that a "spi-max-frequency" property
is present. [0]
Should we patch spi framework then?

[0] https://elixir.bootlin.com/linux/latest/source/drivers/spi/spi.c#L1951
--
Regards,
Chuanhong Guo