Re: [PATCH] Add SoundCard driver for OKI SEMICONDUCTOR ML7213 IOH

[Tomoya MORINAGA <tomoya-linux@dsn.lapis-semi.com>]

Hi Iwai,

(2011/10/21 23:16), Takashi Iwai wrote:
> At Mon, 17 Oct 2011 13:28:39 +0900,
> Tomoya MORINAGA wrote:
>>
>> Hi Iwai,
>>
>> We have just started porting to ASoC structure.
>
> Please add ASoC maintainers to Cc if you have questions about ASoC.
>
>> I have a question.
>>
>> As you reviewed before, currently, our driver consists of 2 parts,
>> Soundcard driver and I2S driver.
>>
>> Soundcard consists of 2 parts
>>     - ALSA interface / control part
>>     - CODEC control part
>>
>> I2S driver consists of 4 parts
>>    - HAL
>>    - DMA control / interrupt control
>>    - Soundcard interface part
>>    - PCI interface function
>>
>> According to "soc" Documentation,
>> We must divide to 3 parts, platform driver, machine driver and codec driver.
>>
>> So, I divided like the following parts.
>>
>> platform driver
>>    - ALSA interface / control part
>>    - HAL
>>    - DMA control / interrupt control
>>    - Soundcard interface part
>>    - PCI interface function
>>
>> machine driver
>>    - (none)
>>
>> codec driver
>>     - CODEC control part
>>
>> Is the above dividing true ?
>
> Not really.  ASoC is designed to be much more modular.  Take a look at
> the codes in sound/soc/*.  You'll grasp how the components are split.
> (The documents in Documentation/sound/alsa/soc/ are slightly
>   obsoleted...)

Of course, I read Documentation/sound/alsa/soc/ and sound/soc/*.
However, it's not easy for me.

I have just 2 questions.

1. PCI interface function.
Any current ASoC drivers don't have PCI interface function.
So I don't know where the function should be in machine driver or 
platform driver.

2. Register Access
Can platform driver access register ?
According to the soc document, platform driver must not access hardware,
however, some drivers looks accessing their hardware.

>
> The conversion of PCM part is usually straightforward, as found in
> *-pcm.c.  The DAI setup depends on the hardware implementation.
> That's for machine driver.  But it's hard to tell in more details
> until I see the actual code snippet you are working on...

I understand.
I'm going to send our modified driver soon.

Thanks,

-- 
tomoya
ROHM Co., Ltd.