[alsa-devel] OSS app works but aplay does not with SNDRV_PCM_FMTBIT_U32_LE

[Clemens Ladisch]

Steve Strobel wrote:
> At 10:03 AM 9/19/2007, Clemens Ladisch wrote:
> >Steve Strobel wrote:
> > > At 12:46 AM 9/19/2007, Clemens Ladisch wrote:
> > > > or did you create a card-specific .conf file for your controller?
> > >
> > > Not intentionally, but there might be one left over from the AD1836
> > > driver I started with (I have just been modifying that driver,
> > > planning to rename it after I get it working).  Where would I look
> > > for such a file?
> >
> >In /usr/share/alsa/cards/, with the driver name.
> 
> I don't have a file there;  that directory doesn't even exist:
> Should I have a configuration file in one of those places?  If I 
> don't, where do the default values come from?

Well, there is an alsa.conf somewhere, otherwise the utils wouldn't run.

> >Does it work when running aplay with the parameter "-D plughw"?
> 
> No.  It searches through more rules, but still says, "Sample format 
> non available".  I put a log of the console output with and without 
> "-D plughw" at <http://link-comm.com/temp/aplay-log.txt> to save
> bandwidth.

Now it says that access RW_INTERLEAVED doesn't work.
I'd really like to see the source code of the driver, especially the
snd_pcm_hardware structure.

> One thing I hadn't noticed before is the message:
> 
> ALSA lib ../../../src/pcm/pcm_params.c:2152:(snd_pcm_hw_refine_slave) Slave PCM not usable
> refine done - result = -22
> 
> Does a result code of -22 mean anything significant?

Nothing that hasn't been already been told in the error message.
-EINVAL is expected in this case.

> You wrote in an earlier message:> > snd_pcm_hw_refine()
>  > > ACCESS = 00000000ffffffffffffffff -> 0000000000000008
>  > > FORMAT = 0000000000000400 -> 0000000000000000
>  >
>  > aplay tries to set S32_LE, but the driver doesn't accept it.
> 
> I think I understand how the driver uses the snd_pcm_hardware_t 
> structure to tell ALSA what it supports, but I don't understand how 
> ALSA tries to set things in return.  I am surprised that it would try 
> to set it to a data type that the driver never said it could 
> support.

A configuration file can force a specific sample format.  Usually, only
dmix does that.


Regards,
Clemens