Re: [PATCH 00/21] spi: s3c64xx: winter cleanup and gs101 support

[Sam Protsenko <semen.protsenko@xxxxxxxxxx>]

On Thu, Jan 25, 2024 at 4:25 PM Andi Shyti <andi.shyti@xxxxxxxxxx> wrote:
>
> Hi Tudor,
>
> > >> The patch set cleans a bit the driver and adds support for gs101 SPI.
> > >>
> > >> Apart of the SPI patches, I added support for iowrite{8,16}_32 accessors
> > >> in asm-generic/io.h. This will allow devices that require 32 bits
> > >> register accesses to write data in chunks of 8 or 16 bits (a typical use
> > >> case is SPI, where clients can request transfers in words of 8 bits for
> > >> example). GS101 only allows 32bit register accesses otherwise it raisses
> > >> a Serror Interrupt and hangs the system, thus the accessors are needed
> > >> here. If the accessors are fine, I expect they'll be queued either to
> > >> the SPI tree or to the ASM header files tree, but by providing an
> > >> immutable tag, so that the other tree can merge them too.
> > >>
> > >> The SPI patches were tested with the spi-loopback-test on the gs101
> > >> controller.
> > >
> > > The reformatting in this series will conflict with the SPI changes in:
> > >
> > >    https://lore.kernel.org/r/20240120012948.8836-1-semen.protsenko@xxxxxxxxxx
> > >
> > > Can you please pull those into this series or otherwise coordinate?
> >
> > ah, I haven't noticed Sam's updates. I'll rebase on top of his set and
> > adapt if necessary. I'll review that set in a sec.
>
> it's a long series, please give it a few days before resending
> it.
>

Also, I recommend splitting it up in a way I suggested before:

  1. First add gs101 support with minimal amount of patches (without
.fifosize introduction, etc)
  2. Then do all those cleanups and reworks on top of that

The reason why I think it's better than doing that vice-versa is that
I feel (2) can take a lot of time/review rounds to get polished and
accepted. So this way you can make sure gs101 support gets applied
sooner. It'll also make it easier to do the backporting work later, if
that's ever needed.

> Thanks,
> Andi