[alsa-devel] ASoC DAI on Atmel AT32AP7000

[Christos Nikolaou]

On Sun, Aug 30, 2009 at 1:02 PM, Mark
Brown<broonie at opensource.wolfsonmicro.com> wrote:
> On Sat, Aug 29, 2009 at 09:41:59PM +0300, Christos Nikolaou wrote:
>
>> Up to 2.6.28 the ASoC framework was working on AT32AP7000 Atmel platform.
>> Now I try to recompile it on 2.6.31-rc8 and I get
>
>> Advance Linux Sound Architecture Driver Version 1.0.20.
>> No device for DAI atmel-ssc0
>> No device for DAI atmel-ssc1
>> No device for DAI atmel-ssc2
>
> That's fine, these are just warning messages and have no actual effect
> on operation.
>

So that means these are generated but the codec driver will eventually
work right?

>> whereas the ssc devices are properly (as in 2.6.28) initialized
>> ssc ssc.0: Atmel SSC device at 0xffe1c00 (irq 10)
>> ssc ssc.1: Atmel SSC device at 0xffe2000 (irq 11)
>> ssc ssc.2: Atmel SSC device at 0xffe2400 (irq 12)
>
> This is outside the control of ASoC - it's generated by the core SSC
> support in drivers/misc/atmel-ssc.c.  Probably you're missing code to
> register the SSC device in your board file, you should be calling
> add_device_ssc() somewhere in your arch/arm code and ssc_request() in
> your machine file.
>

These statements are already there and the output shown is the drivers
being intilized properly, this output is on the same log as the No
device for DAI errors, I was displaying it here to show that although
the devices are initialized ASoC reports an error

>> It looks that this message comes from snd_soc_register_dai in soc-core.c
>
>> if (!dai->dev)
>>          printk(KERN_WARNING "No device for DAI %s\n", dai->name);
>
>> it seems for some reason it thinks there is no device in dai->dev
>
> Yes, the reason for this being that the Atmel driver currently doesn't
> make any effort to initialise the device.  At the minute that's purely a
> cosmetic error.  Before it becomes an actual problem the drivers will be
> updated to fill in a device.
>

So in a feature version, the Atmel platform to ASoC glue drivers like
playpaq_wm8510.c will initialize the required drivers (probably
filling the ssc pointer I guess) but as for the time being that
non-initialization does not create any problems right?


Christos