Re: [PATCH v2 1/6] gpio: i8255: Introduce the i8255 module

From: Bartosz Golaszewski
Date: Wed Jul 13 2022 - 03:38:07 EST


On Tue, Jul 12, 2022 at 5:06 AM William Breathitt Gray
<william.gray@xxxxxxxxxx> wrote:
>
> On Mon, Jul 11, 2022 at 03:02:10PM +0200, Linus Walleij wrote:
> > On Fri, Jul 8, 2022 at 1:16 AM William Breathitt Gray
> > <william.gray@xxxxxxxxxx> wrote:
> >
> > > Exposes consumer functions providing support for Intel 8255 Programmable
> > > Peripheral Interface devices. A CONFIG_GPIO_I8255 Kconfig option is
> > > introduced; modules wanting access to these functions should select this
> > > Kconfig option.
> > >
> > > Tested-by: Fred Eckert <Frede@xxxxxxxxxxxx>
> > > Cc: John Hentges <jhentges@xxxxxxxxxxx>
> > > Cc: Jay Dolan <jay.dolan@xxxxxxxxxxx>
> > > Signed-off-by: William Breathitt Gray <william.gray@xxxxxxxxxx>
> >
> > This chip is like 50 years old, but so am I and I am not obsolete, it's about
> > time that we implement a proper driver for it!
> >
> > But I suppose you are not really using the actual discrete i8255 component?
> > This is certainly used as integrated into some bridge or so? (Should be
> > mentioned in the commit.)
>
> Interestingly, there are some PC/104 devices out there that use actual
> i8255 components (e.g. Diamond Systems Onyx-MM with its 82C55 chips),
> but honestly the majority of devices I come across are simply emulating
> the i8255 interface in an FPGA or similar.
>
> I'll adjust the commit to make it clearer that this is a library for
> i8255-compatible interfaces rather than support for any physical Intel
> 8255 chip in particular.
>
> > > +config GPIO_I8255
> > > + tristate
> >
> > That's a bit terse :D Explain that this is a Intel 8255 PPI chip first developed
> > in the first half of the 1970ies.
>
> Ack.
>
> > > +++ b/include/linux/gpio/i8255.h
> >
> > You need to provide a rationale for the separate .h file in the commit
> > message even if it is clear
> > how it is used in the following patches.
> >
> > Yours,
> > Linus Walleij
>
> I think I'll move this to gpio/driver.h as per Andy Shevchenko's

I don't think this is what Andy meant. I think he suggested moving
this header into drivers/gpio/ because it doesn't make sense for it to
be publicly accessible for anyone else than the GPIO drivers.

Andy: correct me if I'm wrong.

Bart

> suggestion. For now only a few drivers under drivers/gpio/ use this
> library, so it probably doesn't need to be separate just yet.
>