Re: [PATCH v3 1/3] dt-bindings: spi: renesas,rzv2h-rspi: allow multiple DMAs

[Conor Dooley <conor@kernel.org>]

[-- Attachment #1: Type: text/plain, Size: 2035 bytes --]

On Thu, Jan 29, 2026 at 05:55:21PM +0000, Cosmin-Gabriel Tanislav wrote:
> > From: Conor Dooley <conor@kernel.org>
> > Sent: Thursday, January 29, 2026 7:45 PM
> > 
> > On Wed, Jan 28, 2026 at 11:51:30PM +0200, Cosmin Tanislav wrote:
> > > All supported SoCs have multiple DMA controllers that can be used with
> > > the RSPI peripheral. The current bindings only allow a single pair of RX
> > > and TX DMAs.
> > >
> > > The DMA core allows specifying multiple DMAs with the same name, and it
> > > will pick the first available one.
> > >
> > > There is an exception in the base dt-schema rules specifically for
> > > allowing this behavior (dtschema/schemas/dma/dma.yaml).
> > >
> > > dma-names:
> > >   anyOf:
> > >     - uniqueItems: true
> > >     - items:
> > >         # Hack around Renesas bindings which repeat entries to support
> > >         # multiple possible DMA providers
> > >         enum: [rx, tx]
> > >
> > > Allow multiple DMAs to have the same name and only restrict the possible
> > > names of the DMA channels, not their count.
> > >
> > > For RZ/T2H and RZ/N2H SoCs, limit the number of DMA channels to 6, as
> > > they have 3 DMA controllers.
> > 
> > What's the rationale behind not setting minItems to 6 here
> > 
> > >
> > > For RZ/V2H and RZ/V2N SoCs, limit the number of DMA channels to 10, as
> > > they have 5 DMA controllers.
> > 
> > and to 10 here? Do any of the spi controllers on these SoCs not have the
> > ability to use all of the available dma controllers?
> > 
> 
> I left minItems to 2 in case it is necessary to wire up SPI to only a
> subset of the DMA controllers, maybe for performance reasons in a
> board-specific dts?
> 
> I know that dts is only supposed to describe the hardware itself, but for
> now this would be the only way to pre-select which DMA controller is used
> for a specific IP.
> 
> Let me know your thoughts.

Yeah, I can buy that argument.
Acked-by: Conor Dooley <conor.dooley@microchip.com>
pw-bot: not-applicable

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