Re: [RFC PATCH 05/14] ARM: dts: samsung-fimd: add I80 specific properties

[Laurent Pinchart <laurent.pinchart@ideasonboard.com>]

Hi YoungJun,

On Thursday 17 April 2014 14:33:37 YoungJun Cho wrote:
> On 04/17/2014 06:26 AM, Laurent Pinchart wrote:
> > On Tuesday 15 April 2014 14:47:33 YoungJun Cho wrote:
> >> In case of using CPU interface panel, the relevant registers should be
> >> set. So this patch adds relevant dt bindings.
> >> 
> >> Signed-off-by: YoungJun Cho <yj44.cho@samsung.com>
> >> Signed-off-by: Inki Dae <inki.dae@samsung.com>
> >> Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
> >> ---
> >> 
> >>   .../devicetree/bindings/video/samsung-fimd.txt     |    9 +++++++++
> >>   1 file changed, 9 insertions(+)
> >> 
> >> diff --git a/Documentation/devicetree/bindings/video/samsung-fimd.txt
> >> b/Documentation/devicetree/bindings/video/samsung-fimd.txt index
> >> 2dad41b..924c2e1 100644
> >> --- a/Documentation/devicetree/bindings/video/samsung-fimd.txt
> >> +++ b/Documentation/devicetree/bindings/video/samsung-fimd.txt
> >> 
> >> @@ -44,6 +44,15 @@ Optional Properties:
> >>   - display-timings: timing settings for FIMD, as described in document
> >>   [1].
> >>   
> >>   		Can be used in case timings cannot be provided otherwise
> >>   		or to override timings provided by the panel.
> >> 
> >> +- samsung,sysreg-phandle: handle to syscon used to control the system
> 
> Oops... I have to change "samsung,sysreg-phandle" to "samsung,sysreg".
> 
> >> registers
> >> +- vidout-i80-ldi: boolean to support i80 interface instead of rgb one
> >> +- cs-setup: clock cycles for the active period of address signal enable
> >> until
> >> +	chip select is enable in i80 interface
> >> +- wr-setup: clock cycles for the active period of CS signal enable until
> >> +	write signal is enable in i80 interface
> >> +- wr-act: clock cycles for the active period of CS enable in i80
> >> interface
> >> +- wr-hold: clock cycles for the active period of CS disable until write
> >> signal
> >> +	is disable in i80 interface
> > 
> > Shouldn't the interface parameters be considered as a property of the
> > slave device instead ? The bus master side is programmable, and different
> > slaves would have different timing requirements. I think it would make
> > more sense to specify the timings on the slave (panel) side and query them
> > dynamically at runtime. Depending on the slave the timings could be
> > hardcoded in the driver (as they're usually an intrinsic property of the
> > slave) or partially or fully specified in the slave DT node.
> 
> Yes, you're right. These properties are related to panel in a sense.
> 
> But my intention is to use these properties to fimd node in the board
> specific dts file, because it decides the interface with panel and
> these are required by fimd only.

I agree, it's a bit of a gray area. The properties describe the connection 
between the display controller output and the panel. As such they're shared 
between the two components, so deciding on which node to include them in is a 
bit difficult. In this specific case, as the panel dictates the interface 
timings and the display controller is then programmed to comply with those 
timings, I would be tempted to consider timing information as panel 
properties. This obviously complicates driver implementation, as we would need 
to communicate timing information from panel drivers to display controller 
drivers.

> If dynamic query is more logical approach, I should find some ways
> with considering probing order.

I won't push too much for storing interface timing information in the panel DT 
node, but I'm a bit worried that storing it in the display controller DT node 
might cause issues at some point. As DT is an ABI, we need to carefully 
consider the potential issues.

> >>   The device node can contain 'port' child nodes according to the
> >>   bindings
> >> 
> >> defined in [2]. The following are properties specific to those nodes:

-- 
Regards,

Laurent Pinchart