[PATCH v2 3/3] dt-bindings: gpio: Add Nuvoton NPCM750 serial I/O expansion interface(SGPIO)

Jim Liu jim.t90615 at gmail.com
Tue Nov 15 20:21:39 AEDT 2022


Hi Linus and Krzysztof

Thanks for your understanding and your suggestion.
I will follow your suggestion to modify the yaml file.
-> nuvoton,input-ngpios = <...>
-> nuvoton,output-ngpios = <...>

And I don't think the node name needs to use gpio.
because it's not a general gpio, so I reference aspeed dts and use sgpio.
Could I use the sgpio node name or could you provide some suggestions?


If you have any questions or are confused please let me know.
Your comments are most welcome.

Best regards,
Jim
On Mon, Nov 14, 2022 at 6:13 PM Linus Walleij <linus.walleij at linaro.org> wrote:
>
> On Mon, Nov 14, 2022 at 9:38 AM Jim Liu <jim.t90615 at gmail.com> wrote:
>
> > Our sgpio module has 64 pins output and 64 pins input.
> > Soc have 8 reg to control 64 output pins
> > and  8 reg to control 64 input pins.
> > so the pin is only for gpi or gpo.
> >
> > The common property ngpio can be out or in.
> > so i need to create d_out and d_in to control it.
> > customers can set the number of output or input pins to use.
> > the driver will open the ports to use.
> > ex: if  i set d_out=9   and d_in=20
> > driver will open two output ports and three input ports.
> >
> > Another method  is the driver default opens all ports , in this
> > situation the driver doesn't need d_out and d_in.
>
> Finally I get it!
>
> Some of the above should go into the binding document so that
> others understand it too.
>
> Have you considered splitting this into 2 instances with 2 DT nodes:
> one with up to 64 output-only pins and one with up to 64 input-only pins?
> That means more nodes in the DT and more compatibles. If all
> the registers are in the same place maybe this is not a good
> idea.
>
> If you feel you need to keep the two properties, create something custom
> for your hardware because this is not generally useful, e.g.
>
> nuvoton,input-ngpios = <...>
> nuvoton,output-ngpios = <...>
>
> By this nomenclature it also becomes more evident what is going on.
>
> Yours,
> Linus Walleij


More information about the openbmc mailing list