Re: [PATCH] spi: Kconfig: spi-synquacer: Added ARM and ARM64 dependence

[Masahisa Kojima <masahisa.kojima@xxxxxxxxxx>]

Hi Mark,

There was a patch to address same compile error.
https://www.mail-archive.com/linux-kernel@xxxxxxxxxxxxxxx/msg1982194.html

Does it preferable solution to use iowrite32_rep() series?

On Mon, 17 Jun 2019 at 20:46, Mark Brown <broonie@xxxxxxxxxx> wrote:
>
> On Mon, Jun 17, 2019 at 12:21:47PM +0200, Ard Biesheuvel wrote:
> > On Mon, 17 Jun 2019 at 12:17, Mark Brown <broonie@xxxxxxxxxx> wrote:
>
> > > Are you sure it's those functions (which only appear to be defined on
> > > arc according to a quick grep) and are you sure that there's no other
> > > Kconfig symbol specifically for those being defined which can be used
> > > rather than depending on specific architectures?
>
> > I'm not sure I see the point of this. Building this driver for alpha
> > and parisc has no use in practice, and does not provide any additional
> > build coverage given that the driver can be enabled on any ARMA/ARM64
> > build.
>
> It's helpful for compile test coverage, which in turn is useful for
> people doing subsystem or kernel wide changes.  A dependency on ARM64 ||
> COMPILE_TEST would make sense but wouldn't stop things getting built on
> the older archtiectures so you'd still want some dependency for that.
>
> > > > This patch also specifies the default compile type as module.
>
> > > This should be a separate patch and we don't generally change the
> > > default unless there's a reason to do so.'
>
> > That was my suggestion - just 'default m' is generally not accepted,
> > but 'default m' for a driver if you enabled the ARCH_xxxx in question
> > is reasonable, no?
>
> No, there's no special treatment for arch specific drivers here.