[PATCH v2 1/2] drm: bridge: samsung-dsim: Initialize bridge on attach

[Marek Vasut]

On 7/12/24 9:16 AM, Michael Walle wrote:
> Hi Marek,

Hi,

>>>> Thank you for testing and keeping up with this. I will wait for more
>>>> feedback if there is any (Frieder? Lucas? Michael?). If there are no
>>>> objections, then I can merge it in a week or two ?
>>>
>>> I'll try to use your approach on the tc358775. Hopefully, I'll find
>>> some time this week.
>>
>> So ... I wonder ... shall I apply these patches or not ?
> 
> As mentioned on IRC, I tried it to port it for the mediatek DSI
> host, but I gave up and got doubts that this is the way to go. I
> think this is too invasive (in a sense that it changes behavior)

I would argue it makes the behavior well defined. If that breaks some 
drivers that depended on the undefined behavior before, those should be 
fixed too.

> and not that easy to implement on other drivers.

How so ? At least the DSIM and STM32 DW DSI host can switch lanes to 
LP11 state. Is the mediatek host not capable of that ?

> Given that this requirement is far more common across DSI bridges,
> I'd favor a more general solution which isn't a workaround.

I think we only had a look at the TI DSI83 / ICN6211 / Toshiba TC358767 
bridges, but we did not look at many panels, did we ? Do panels require 
lanes in non-LP11 state on start up ?

Was there any progress on the generic LP11 solution, I think you did 
mention something was in progress ? How would that even look like ?