Re: [PATCH v1] spi: designware: add support for bits-per-word DT binding

[Conor Dooley <conor@kernel.org>]

[-- Attachment #1: Type: text/plain, Size: 1201 bytes --]

On Wed, Mar 04, 2026 at 08:38:17AM -0600, Tom Rini wrote:
> On Wed, Mar 04, 2026 at 10:41:57AM +0800, Ng, Boon Khai wrote:
> > 
> > > 
> > > Thanks. And can we reasonably follow what Linux does here instead?
> > > 
> > 
> > We don't have ACPI in our devices, instead, i think that can be replaced
> > with DTS.
> 
> I thought Conor said it could be auto-detected still, without ACPI, in
> Linux?

The controller capabilities are auto-detected. This has nothing to do
with ACPI AFAICT, I don't know why that was brought up. What it detects
is not an exact value though, the detected capability is the max of a
range. On Linux, the spi controller driver advertises the range it
supports. The spi device struct carries the bits_per_word that it wants
to use, and the controller driver looks into that struct to figure out
what supported bits_per_word setting to use in the transfer function.
It's not as simple as just reading the register in the controller and
setting the U-Boot drivers bits_per_word to the value, but I don't see
why U-Boot couldn't do what Linux does here.

If you search in spi-dw-core for bits_per_word, you should be able to
get a grasp of how it works.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 228 bytes --]