[alsa-devel] HD-Audio device unsuppored (Nvidia G2xx-series GPU implemented)

[Takashi Iwai]

At Sat, 19 Dec 2009 13:36:23 +0100,
Stefan Ringel wrote:
> 
> Am 16.12.2009 22:15, schrieb Stefan Ringel:
> > Am 14.12.2009 18:45, schrieb Takashi Iwai:
> >   
> >> At Mon, 14 Dec 2009 12:52:40 +0100,
> >> I wrote:
> >>   
> >>     
> >>>>         
> >>>     
> >>>       
> >>>>>> one codec-id missed (0x10de000d) and codec-name is "G2xx HDMI".
> >>>>>>             
> >>>>       
> >>>>
> >>>> OK, I'll update.
> >>>>         
> >>>     
> >>>       
> >> Below is the updated patch.  This also replaces the PCI ID check using
> >> the generic hd-audio class check.
> >>
> >>
> >> Takashi
> >>
> >>   
> >>     
> > Hi Takashi,
> >
> > I have a theory,
> >
> > 4 codec with 2 channels bundling to one pcm device for 8 channels
> > uncompressed lpcm audio and compressed audio over first codec.
> >
> > Actual configuration is 4 separate pcm device.
> >
> >
> >   
> Hi Takashi,
>  
> I tested your patch and it doesn't work.
> I checked caps from nodes 0x04 and 0x05 and comparing with  
> http://download.intel.com/standards/hdaudio/pdf/hda034-a2.pdf. document.
> The configuration is 4 full HDMI codec not 4 2-channels pcm codec. It
> have CP-flag, ext-channel and in pin-widget that HDMI-flag.

But still it's not sure whether it's really compliant.
Nvidia HDMI codecs have been incompatible with that standard although
they have already supported 8 channels.
And, one wonders why there are 4 HDMI codecs there.

Anyway, try to set up just one codec and don't do anything else.
If it's a standard codec, patch_intelhdmi.c would work.  Create a new
patch code that ignores the instances with codec->addr > 1, so that
you concentrate on only the first codec address.
Or, patch_atihdmi.c can be used by changing the fixed pins...


Takashi