[PATCH v2 7/7] arm64: dts: allwinner: a64: enable ANX6345 bridge on Teres-I

[Andrzej Hajda]

On 07.06.2019 11:40, Torsten Duwe wrote:
> On Fri, Jun 07, 2019 at 08:28:02AM +0200, Maxime Ripard wrote:
>> On Thu, Jun 06, 2019 at 03:59:27PM +0200, Harald Geyer wrote:
>>> If think valid compatible properties would be:
>>> compatible = "innolux,n116bge", "simple-panel";
>>> compatible = "edp-connector", "simple-panel";
>> A connector isn't a panel.
>>
>>> compatible = "innolux,n116bge", "edp-connector", "simple-panel";
>> And the innolux,n116bge is certainly not a connector either.
>>
>>> compatible = "edp-connector", "innolux,n116bge", "simple-panel";
>>>
>>> I can't make up my mind which one I prefere. However neither of these
>>> variants requires actually implmenting an edp-connector driver.
>> No-one asked to do an edp-connector driver. You should use it in your
>> DT, but if you want to have some code in your driver that parses the
>> DT directly, I'm totally fine with that.
> I must admit I fail to understand what that extra node would be good for.
> Logically, the eDP far side is connected to the well-known n116bge.
> Inside the laptop case it might as well be a flat ribbon cable or
> soldered directly.
> In good intention, that's all I wanted to express in the DT. I don't
> know whether the relevant mechanical dimensions of the panel and the
> connector are standardised, so whether one could in theory assemble it
> with a different panel than the one it came with.
>
> OTOH, as I checked during the discussion with anarsoul, the panel's
> supply voltage is permanently connected to the main 3.3V rail.
> We already agreed that the eDP output port must not neccessarily be
> specified, this setup is a good example why: because the panel is
> always powered, the anx6345 can always pull valid EDID data from it
> so at this stage there's no need for any OS driver to reach beyond
> the bridge. IIRC even the backlight got switched off for the blank
> screen without.
>
> All I wanted to say is that "there's usually an n116bge behind it";
> but this is mostly redundant.
>
> So, shall we just drop the output port specification (along with the
> panel node) in order to get one step further?


I am not sure if I understand whole discussion here, but I also do not
understand whole edp-connector thing.

According to VESA[1] eDP is "Internal display interface" - there is no
external connector for eDP, the way it is connected is integrator's
decision, but it is fixed - ie end user do not plug/unplug it.

If I remember correctly in some boards eDP is connected to some DP
connector (odroid xu3 if I remember correctly), but this is non-standard
hack, and for this case in bindings there should be rather dp-connector
not edp-connector.

[1]:
https://www.vesa.org/wp-content/uploads/2010/12/DisplayPort-DevCon-Presentation-eDP-Dec-2010-v3.pdf


Regards

Andrzej


>
>> I guess you should describe why do you think it's "clear", because I'm
>> not sure this is obvious for everyone here. eDP allows to discover
>> which device is on the other side and its supported timings, just like
>> HDMI for example (or regular DP, for that matter). Would you think
>> it's clearly preferable to ship a DT with the DP/HDMI monitor
>> connected on the other side exposed as a panel as well?
> Well, as I wrote: "in good intention". That's the panel that comes with
> the kit but it is very well detected automatically, just like you describe.
>
> So, just leave it out?
>
> 	Torsten
>
>
>