Re: [PATCH v2 3/3] ARM: shmobile: marzen: Add Display Unit support

[Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>]

Hello.

On 15-03-2013 5:36, Laurent Pinchart wrote:

>>>>> Only the DU0 VGA output is currently supported. Support for the DU0 LVDS
>>>>> and DU1 LVDS outputs will require information about the panels that will
>>>>> be connected to those outputs.

>>>>> Signed-off-by: Laurent Pinchart
>>>>> <laurent.pinchart+renesas@ideasonboard.com>
>>>>> ---

>>>>>     arch/arm/configs/marzen_defconfig     |  2 ++
>>>>>     arch/arm/mach-shmobile/board-marzen.c | 65
>>>>>     +++++++++++++++++++++++++++++
>>>>>     2 files changed, 67 insertions(+)

>>>>> diff --git a/arch/arm/mach-shmobile/board-marzen.c
>>>>> b/arch/arm/mach-shmobile/board-marzen.c index cdcb799..0020506 100644
>>>>> --- a/arch/arm/mach-shmobile/board-marzen.c
>>>>> +++ b/arch/arm/mach-shmobile/board-marzen.c

>>>> [...]

>>>>> @@ -147,6 +148,38 @@ static struct platform_device hspi_device = {
>>>>>     	.num_resources	= ARRAY_SIZE(hspi_resources),
>>>>>     };
>>>>>
>>>>> +/* DU */
>>>>> +static struct resource rcar_du_resources[] = {
>>>>> +	[0] = {
>>>>> +		.name	= "Display Unit",
>>>>> +		.start	= 0xfff80000,
>>>>> +		.end	= 0xfffb1007,
>>>>> +		.flags	= IORESOURCE_MEM,
>>>>> +	},
>>>>> +	[1] = {
>>>>> +		.start	= gic_spi(31),
>>>>> +		.flags	= IORESOURCE_IRQ,
>>>>> +	},
>>>>> +};
>>>>> +
>>>>> +static struct rcar_du_platform_data rcar_du_pdata = {
>>>>> +	.encoders = {
>>>>> +		[0] = {
>>>>> +			.encoder = RCAR_DU_ENCODER_VGA,
>>>>> +		},
>>>>> +	},
>>>>> +};
>>>>> +
>>>>> +static struct platform_device rcar_du_device = {
>>>>> +	.name		= "rcar-du",
>>>>> +	.num_resources	= ARRAY_SIZE(rcar_du_resources),
>>>>> +	.resource	= rcar_du_resources,
>>>>> +	.dev	= {
>>>>> +		.platform_data = &rcar_du_pdata,
>>>>> +		.coherent_dma_mask = ~0,
>>>>> +	},
>>>>> +};
>>>>> +

>>>> Are we seeing again SoC device declared in the board file? That simply
>>>> doesn't scale...

>>> The goal is obviously to move all that to DT, but there's no DT bindings
>>> for the DU DRM driver yet.

>> I don't see how it justifies dubious non-DT design. Let me tell/remind you
>> about the LTSI-3.4 tree where all this stuff can be backported and which
>> doesn't have DT support, AFAIR.

> This patch is an example only, I should have marked it as such. If I need to
> push something similar to mainline I will keep your comment in mind and see
> how I can move the platform device to setup-r8a7779.c. Platform data, however,
> need to be provided on a per-board basis.

    Yes, it's the usual case of their usage.
    Thanks for heeding to my comment.

WBR, Sergei