Re: [PATCH v4 0/3] TV driver for Samsung S5P platform (media part)

[Tomasz Stanislawski <t.stanislaws@samsung.com>]

JiUn Yu wrote:
>   
>> 5. Mixer & Video Processor driver. It is called 's5p-mixer' because of
>> historical reasons. It was decided combine VP and MXR drivers into one because
>> of shared interrupt and very similar interface via V4L2 nodes. The driver is a
>> realization of many-to-many relation between multiple input layers and multiple
>> outputs. All shared resources are kept in struct mxr_device. It provides
>> utilities for management and synchronization of access to resources and
>> reference counting. The outputs are obtained from HDMI/SDO private data.  One
>> layer is a single video node. Simple inheritance is applied because there only
>> little difference between layer's types. Every layer type implements set of
>> ops.  There are different ops for Mixer layers and other for VP layer.
>>     
>
> I agreed with subdev of hdmi, hdmiphy, sdo and videoDAC. It is very flexible in case of adding new interface or removing current interface.
> But 's5p-mixer' driver is not flexible. So, If new scaler is added instead of VP or mixer is someting changed,
> I think current architecture of tvout driver can't support.
> How about separating vp and mixer driver?
>   
Hi Yu,
The designed architecture TV driver was not prepared for removal of VP, 
or MIxer input nodes.
I decided only to allow flexible outputs because I knew 2 boards with 
different output configuration (Goni - only SDO, Universal - HDMI and SDO).
I need more information about VP substitute before changing design of 
the whole driver. Some of extensions could be applied by adding extra 
layer type to existing layer_vp and layer_grp ones.

Best regards
Tomasz Stanislawski
> N떑꿩�r툤y鉉싕b쾊Ф푤v�^�)頻{.n�+돴쪐{콡�bj)�鍊w*\x1fjgП�\x1e텎쉸듶줷/곴�z받뻿�2듷솳鈺�&�)傘첺뛴��\x1e췍쳺�h�\x0f�j:+v돣둾�明