Re: [PATCH/RFC v2 3/7] spi: core: Add support for registering SPI slave controllers

From: Mark Brown
Date: Mon Dec 19 2016 - 08:29:12 EST


On Mon, Dec 19, 2016 at 11:02:07AM +0100, Geert Uytterhoeven wrote:
> On Thu, Dec 15, 2016 at 6:53 PM, Mark Brown <broonie@xxxxxxxxxx> wrote:
> > On Mon, Sep 12, 2016 at 10:50:42PM +0200, Geert Uytterhoeven wrote:
> >> TBD:
> >> - s/spi_master/spi_controller/ where appropriate,
> >> - Provide wrappers (e.g. "#define spi_master spi_controller" until all
> >> SPI drivers have been converted),
> >> - Do we want a separate spi_register_slave() instead?

> > This basically looks fine to me - there's these TBDs so I might be
> > missing things and we probably need some GPIO chip select handling but

> Given the hard real-time requirements of SPI slave, supporting GPIO chip
> select may not be feasible.

It's not unknown for SPI devices to have some minimum time requirement
from chip select to first clock so it's workable and probably something
people will want at some point. It can always be done later though.

> As I managed to fix the issue with spi_slave_abort() on MSIOF, I think
> the remaining
> obstacle is the DT binding. Do you have any feedback or other
> suggestions in that area?

> IMHO having the ability to bind to an SPI slave handler either from DT or
> by using the sysfs virtual file is useful to have.

I think it'd be useful to have DT support but really I think the DT
maintainers are going to have more opinions on this than me.

Attachment: signature.asc
Description: PGP signature