Re: [PATCH v3 6/8] dt-bindings: microchip,sparx5-spi-mux: Add Sparx5 SPI mux driver bindings

From: Lars Povlsen
Date: Tue Jul 14 2020 - 04:52:08 EST



Rob Herring writes:

> On Thu, Jul 02, 2020 at 12:13:29PM +0200, Lars Povlsen wrote:
>> The Microchip Sparx5 SPI controller has two bus segments, and use this
>> mux to control the bus interface mapping for any chip selects. This
>> decribes the bindings used to configure the mux driver.
>>
>> Signed-off-by: Lars Povlsen <lars.povlsen@xxxxxxxxxxxxx>
>> ---
>> .../mux/microchip,sparx5-spi-mux.yaml | 71 +++++++++++++++++++
>> 1 file changed, 71 insertions(+)
>> create mode 100644 Documentation/devicetree/bindings/mux/microchip,sparx5-spi-mux.yaml
>>
>> diff --git a/Documentation/devicetree/bindings/mux/microchip,sparx5-spi-mux.yaml b/Documentation/devicetree/bindings/mux/microchip,sparx5-spi-mux.yaml
>> new file mode 100644
>> index 0000000000000..b0ce3b15a69e5
>> --- /dev/null
>> +++ b/Documentation/devicetree/bindings/mux/microchip,sparx5-spi-mux.yaml
>> @@ -0,0 +1,71 @@
>> +# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
>> +%YAML 1.2
>> +---
>> +$id: http://devicetree.org/schemas/mux/microchip,sparx5-spi-mux.yaml#
>> +$schema: http://devicetree.org/meta-schemas/core.yaml#
>> +
>> +title: Microchip Sparx5 SPI mux
>> +
>> +maintainers:
>> + - Lars Povlsen <lars.povlsen@xxxxxxxxxxxxx>
>> +
>> +description: |
>> + The Microchip Sparx5 SPI controller has two bus segments. In order
>> + to switch between the appropriate bus for any given SPI slave
>> + (defined by a chip select), this mux driver is used. The device tree
>> + node for the mux will define the bus mapping for any chip
>> + selects. The default bus mapping for any chip select is "0", such
>> + that only non-default mappings need to be explicitly defined.
>> +
>> +properties:
>> + compatible:
>> + enum:
>> + - microchip,sparx5-spi-mux
>> +
>> + '#address-cells':
>> + const: 1
>> +
>> + '#size-cells':
>> + const: 0
>> +
>> + '#mux-control-cells':
>> + const: 0
>> +
>> +required:
>> + - compatible
>> +
>> +additionalProperties: false
>> +
>> +patternProperties:
>> + "^mux@[0-9a-f]$":
>> + type: object
>> +
>> + properties:
>> + reg:
>> + description:
>> + Chip select to define bus mapping for.
>> + minimum: 0
>> + maximum: 15
>> +
>> + microchip,bus-interface:
>> + description:
>> + The bus interface to use for this chip select.
>> + $ref: /schemas/types.yaml#/definitions/uint32
>> + enum: [0, 1]
>> +
>> + required:
>> + - reg
>> + - microchip,bus-interface
>> +
>> +examples:
>> + - |
>> + mux: mux-controller {
>> + compatible = "microchip,sparx5-spi-mux";
>> + #address-cells = <1>;
>> + #size-cells = <0>;
>> + #mux-control-cells = <0>;
>
> How is this mux accessed? You have no control interface defined.
>
>> + mux@e {
>> + reg = <14>;
>> + microchip,bus-interface = <1>;
>
> This looks odd. I take it that there's 2 muxes for this h/w? If so then
> #mux-control-cells should be 1 and the cell value can be whatever you
> want that is meaningful for the mux controller. Could be 0,1 or perhaps
> 0xe if that's more useful.
>

Rob,

The intended use was for the SPI driver to use mux_control_select(mux,
<cs>) and then have the mux driver translate the <cs> to the right bus
interface, according to its configuration. The SPI driver would have a
"mux-controls" property bound to this mux.

Anyway, I am getting pushed in the direction of using the "mux-mmio" and
"spi-mux" combo, so this driver and bindings are being dropped again.

I am currently awaiting the "rx-sample-delay-ns" issue to be clarified
such that I can refresh the series.

Thank you for your comments!

> Rob

--
Lars Povlsen,
Microchip