Re: [PATCH 2/4] dt-bindings: Document the Raspberry Pi Touchscreen nodes.

[Archit Taneja <architt@codeaurora.org>]

On 05/18/2017 08:25 PM, Laurent Pinchart wrote:
> Hi Archit,
>
> On Thursday 18 May 2017 13:56:19 Archit Taneja wrote:
>> On 05/17/2017 12:16 AM, Eric Anholt wrote:
>
> [snip]
>
>>> In terms of physical connections:
>>>    [15-pin "DSI" connector on 2835]
>>>
>>>     | I2C               | DSI
>>>
>>>    / \        SPI       |
>>>
>>> [TS]  [Atmel]------[TC358762]
>>>
>>>        \                |
>>>
>>>         \PWM            |
>>>
>>>          \              | DPI
>>>
>>> [some backlight]------[some unknown panel]
>>>
>>> The binding I'm trying to create is to expose what's necessary for a
>>> driver that talks I2C to the Atmel, which then controls the PWM and does
>>> the command sequence over SPI to the Toshiba that sets up its end of the
>>> DSI link.
>>
>> The bridge (Atmel + TC358762 combination) here looks like it's primarily
>> an i2c device (i.e, the control bus is i2c). Therefore, the drm-bridge
>> driver here should be an i2c driver instead of a mipi_dsi_driver.
>
> Glad to see we agree, that's what I've proposed in a separate answer :-) I'd
> go one step further though, there should be no DRM bridge, just a DRM panel.

If the PCB containing the controller chips and the panel are part of a single
casing, and the set up won't work with another panel, then yeah, I agree. If the
bridge chips are on a separate adapter board, and there is a possibility to connect
other panels, then maybe a separate DRM bridge and a DRM panel might be a safer bet.

Thanks,
Archit

>
>> We have the facility to create a mipi DSI device without the need to have
>> a corresponding node in DT. The ADV7533 and TC358767 drivers are examples
>> of that.
>>
>> The following is what the binding could look like, it's same as what Rob
>> also mentioned previously in the thread.
>>
>> Thanks,
>> Archit
>>
>> dsi1: dsi@7e700000 {
>> 	#address-cells = <1>;
>> 	#size-cells = <0>;
>> 	<...>
>>
>> 	/* The SoC's DSI input/output port */
>> 	ports {
>> 		#address-cells = <1>;
>> 		#size-cells = <0>;
>>
>> 		/* port@0 if needed */
>>
>> 		port@1 {
>> 			dsi_out_port: endpoint {
>> 				reg = <1>;
>> 				remote-endpoint = <&bridge_dsi_port>;
>> 			};
>> 		};
>> 	};
>> };
>>
>> i2c_dsi: i2c {
>> 	compatible = "i2c-gpio";
>> 	#address-cells = <1>;
>> 	#size-cells = <0>;
>> 	gpios = <&gpio 28 0
>> 		 &gpio 29 0>;
>>
>> 	/* the Atmel + TC35872 bridge */
>> 	pitouchscreen_bridge: bridge@45 {
>
> This should thus be lcd@45.
>
>> 		compatible = "raspberrypi,touchscreen-bridge";
>
> And this raspberrypi,7inch-touchscreen-panel. Shame we haven't standardized
> the vendor name prefix to rpi :-/
>
>> 		reg = <0x45>;
>>
>> 		ports {
>> 			#address-cells = <1>;
>> 			#size-cells = <0>;
>>
>> 			port@0 {
>> 				reg = <0>;
>> 				bridge_dsi_port: endpoint {
>
> This should be named panel_dsi_port.
>
>> 					remote-endpoint = <&dsi_out_port>;
>> 				};
>> 			};
>> 			port@1 {
>> 				reg = <1>;
>> 				bridge_dpi_port: endpoint {
>> 					remote-endpoint =
> <&pitouchscreen_panel_port>;
>> 				};
>> 			};
>
> The second port is thus not needed.
>
>> 		};
>
> So we can simplify this to
>
> 		port {
> 			panel_dsi_port: endpoint {
> 				remote-endpoint = <&dsi_out_port>;
> 			};
> 		};
>
> (no need for a ports node when there's a single port)
>
>> 	};
>> };
>>
>> lcd {
>> 	compatible = "raspberrypi,7inch-touchscreen-panel";
>> 	ports {
>> 		#address-cells = <1>;
>> 		#size-cells = <0>;
>> 		port@0 {
>> 			reg = <0>;
>> 			pitouchscreen_panel_port: endpoint {
>> 				remote-endpoint = <&bridge_dpi_port>;
>> 			};
>> 		};
>> 	};
>> };
>
> And this node can go away.
>

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel