hints around rcar_lvds.c :)

hints around rcar_lvds.c :)

[Tommaso Merciai]

Hi Laurent, All,

Sorry for bothering.
Looking for some feedback :)

I have a similar rcar_lvds.c IP's to handle but in my case:
I have lvds0 and lvds1 that are sharing some common regs (lvds_cmn).

 ----------------------
|    -------------     |
|   |lvds_cmn_regs|    |
|    -------------     |
|                      |
|    -----------       |
|   | lvds0_regs |     |-----> ch0
|    ------------      |
|                      |
|    -----------       |
|   | lvds1_regs |     |-----> ch1
|    ------------      |
 ----------------------


So I'm checking 2 drm dts/driver architecture:

1st architecture:
 - Using a single lvds driver to handle both lvds0 and lvds1.

		 ----------------------
		|                      |
		|                      |
		|                      |
du_lvds0 ------>|                      |----> ch0_lvds
		|      lvds_bridge     |
		|                      |
		|                      |
du_lvds1 ------>|                      |----> ch1_lvds
		|                      |
		 ----------------------


Issue:

Problem here is the 1 single link 2ch mode.
lvds0 and lvds1 can drive 2 display with 2 differents fb (fb0 and fb1).

Having a single drm_bridge to drive both channel give me the following issue:

In single link 2ch mode when for the first time the du encoder drm_attach()
the lvds bridge to the encoder(du) all goes fine and fb0 is created correctly.

Then again the du encoder is trying again to drm_attach() the lvds bridge
but this return -EBUSY obviously because is already attached.

Then I think this is not the way to follow because I need 2 drm_bridges
from the same drm drive, and I think this is not correct.
----------

2nd architecture:
 - Follow rcar_lvds.c way using 2 nodes for lvds0 and lvds1:

		 ------------
du_lvds0 -----> |lvds0_bridge|----> ch0_lvds
		 ------------

		 ------------
du_lvds1 -----> |lvds1_bridge|----> ch1_lvds
		 ------------

Issue:
I thinks this is an optimal approach but in my case here
the problem is that lvds0 and lvds1 share a set of common registers
some common clocks and common reset:

My plan is to manipulate those common regs (lvds_cmn) using
compatible = "simple-mfd", "syscon"; as follow:

lvds_cmn: lvds-cmn {
	compatible = "simple-mfd", "syscon";
	reg = <common_regs>;

	lvds0: lvds0-encoder {

		ports {
			#address-cells = <1>;
			#size-cells = <0>;
			clocks = <&common_clk>, <&dotclok0>, <&phyclock0>;
			resets = <&common_rst>;

			port@0 {
				reg = <0>;
				lvds0_in: endpoint {
					remote-endpoint = <&du_out_lvds0>;
				};
			};

			port@1 {
				reg = <1>;
				lvds_ch0: endpoint {
				};
			};
		};
	};

	lvds1: lvds1-encoder {

		ports {
			#address-cells = <1>;
			#size-cells = <0>;
			clocks = <&common_clk>, <&dotclok1>, <&phyclock1>;
                        resets = <&common_rst>;

			port@0 {
				reg = <0>;
				lvds1_in: endpoint {
					remote-endpoint = <&du_out_lvds1>;
				};
			};

			port@1 {
				reg = <1>;
				lvds_ch1: endpoint {
				};
			};
		};
	};
};
----------

I'm asking to find the best way to represent those IP's.
What do you think?
Any hints/tips would be nice.
Thanks in advance.

Thanks & Regards,
Tommaso

RE: hints around rcar_lvds.c :)

[Biju Das]

+ renesas-soc

> -----Original Message-----
> From: dri-devel <dri-devel-bounces@lists.freedesktop.org> On Behalf Of Tommaso Merciai
> Sent: 26 November 2024 10:15
> To: laurent.pinchart <laurent.pinchart@ideasonboard.com>
> Cc: Kieram Bingham <kieran.bingham+renesas@ideasonboard.com>; David Airlie <airlied@gmail.com>; Simona
> Vetter <simona@ffwll.ch>; dri-devel@lists.freedesktop.org; linux-kernel@vger.kernel.org
> Subject: hints around rcar_lvds.c :)
> 
> Hi Laurent, All,
> 
> Sorry for bothering.
> Looking for some feedback :)
> 
> I have a similar rcar_lvds.c IP's to handle but in my case:
> I have lvds0 and lvds1 that are sharing some common regs (lvds_cmn).
> 
>  ----------------------
> |    -------------     |
> |   |lvds_cmn_regs|    |
> |    -------------     |
> |                      |
> |    -----------       |
> |   | lvds0_regs |     |-----> ch0
> |    ------------      |
> |                      |
> |    -----------       |
> |   | lvds1_regs |     |-----> ch1
> |    ------------      |
>  ----------------------
> 
> 
> So I'm checking 2 drm dts/driver architecture:
> 
> 1st architecture:
>  - Using a single lvds driver to handle both lvds0 and lvds1.
> 
> 		 ----------------------
> 		|                      |
> 		|                      |
> 		|                      |
> du_lvds0 ------>|                      |----> ch0_lvds
> 		|      lvds_bridge     |
> 		|                      |
> 		|                      |
> du_lvds1 ------>|                      |----> ch1_lvds
> 		|                      |
> 		 ----------------------
> 
> 
> Issue:
> 
> Problem here is the 1 single link 2ch mode.
> lvds0 and lvds1 can drive 2 display with 2 differents fb (fb0 and fb1).
> 
> Having a single drm_bridge to drive both channel give me the following issue:
> 
> In single link 2ch mode when for the first time the du encoder drm_attach() the lvds bridge to the
> encoder(du) all goes fine and fb0 is created correctly.
> 
> Then again the du encoder is trying again to drm_attach() the lvds bridge but this return -EBUSY
> obviously because is already attached.
> 
> Then I think this is not the way to follow because I need 2 drm_bridges from the same drm drive, and I
> think this is not correct.
> ----------
> 
> 2nd architecture:
>  - Follow rcar_lvds.c way using 2 nodes for lvds0 and lvds1:
> 
> 		 ------------
> du_lvds0 -----> |lvds0_bridge|----> ch0_lvds
> 		 ------------
> 
> 		 ------------
> du_lvds1 -----> |lvds1_bridge|----> ch1_lvds
> 		 ------------
> 
> Issue:
> I thinks this is an optimal approach but in my case here the problem is that lvds0 and lvds1 share a
> set of common registers some common clocks and common reset:
> 
> My plan is to manipulate those common regs (lvds_cmn) using compatible = "simple-mfd", "syscon"; as
> follow:
> 
> lvds_cmn: lvds-cmn {
> 	compatible = "simple-mfd", "syscon";
> 	reg = <common_regs>;
> 
> 	lvds0: lvds0-encoder {
> 
> 		ports {
> 			#address-cells = <1>;
> 			#size-cells = <0>;
> 			clocks = <&common_clk>, <&dotclok0>, <&phyclock0>;
> 			resets = <&common_rst>;
> 
> 			port@0 {
> 				reg = <0>;
> 				lvds0_in: endpoint {
> 					remote-endpoint = <&du_out_lvds0>;
> 				};
> 			};
> 
> 			port@1 {
> 				reg = <1>;
> 				lvds_ch0: endpoint {
> 				};
> 			};
> 		};
> 	};
> 
> 	lvds1: lvds1-encoder {
> 
> 		ports {
> 			#address-cells = <1>;
> 			#size-cells = <0>;
> 			clocks = <&common_clk>, <&dotclok1>, <&phyclock1>;
>                         resets = <&common_rst>;
> 
> 			port@0 {
> 				reg = <0>;
> 				lvds1_in: endpoint {
> 					remote-endpoint = <&du_out_lvds1>;
> 				};
> 			};
> 
> 			port@1 {
> 				reg = <1>;
> 				lvds_ch1: endpoint {
> 				};
> 			};
> 		};
> 	};
> };
> ----------
> 
> I'm asking to find the best way to represent those IP's.
> What do you think?
> Any hints/tips would be nice.
> Thanks in advance.
> 
> Thanks & Regards,
> Tommaso