Re: [alsa-devel] RFC: ice1712 virtual devices

On Sunday 08 November 2009 10:38, you wrote:

At Fri, 30 Oct 2009 09:23:37 +0000,

Alan Horstmann wrote:

...

On Thursday 29 October 2009 17:48, Arno Schuring wrote:

...

This is basically a resend of http://thread.gmane.org/gmane.linux.alsa.devel/59481/focus=59672 , which fixed the front: device of ice1712 cards to accept two-channel input. Currently, the front: device is exposed through the route plugin, which requires all clients to mmap all 10 channels, even though the front device is supposed to be a stereo device.

This patch changes the front: device definition such that it matches the definition of iec958 in the same file. Additionally, I'm tempted to remove the surround* definitions because the chip does not really offer surround-style multichannel: it basically just offers multiple stereo channels, and does not provide any channel mapping beyond stereo.

Finally, I'm also experimenting with the dshare plugin to allow applications to access the iec958: and front: devices simultaneously. Can anyone point me to a working example for this? From reading the alsa-lib documentation, it is not clear to me how I should nest the different plugins.

Many thanks, Arno Schuring

--

diff --git a/src/conf/cards/ICE1712.conf b/src/conf/cards/ICE1712.conf index 01e50d2..d7acb81 100644 --- a/src/conf/cards/ICE1712.conf +++ b/src/conf/cards/ICE1712.conf @@ -32,12 +32,28 @@ ICE1712.pcm.front.0 { @args.CARD { type string }

•   type route
  

•   ttable.0.0 1
  

•   ttable.1.1 1
  

•   slave.pcm {
  

•           type hw
  

•           card $CARD
  

•   type asym
  

•   playback.pcm {
  

•           type route
  

•           ttable.0.0 1
  

•           ttable.1.1 1
  

•           slave.pcm {
  

•                   type hw
  

•                   card $CARD
  

•           }
  

•           slave.format S32_LE
  

•           slave.channels 10
  

•   }
  

•   capture.pcm {
  

•           type route
  

•           ttable.0.0 1
  

•           ttable.1.1 1
  

•           slave.pcm {
  

•                   type hw
  

•                   card $CARD
  

•           }
  

•           slave.format S32_LE
  

•           slave.channels 12
    }
  

}

Bear in mind that the ice1712 has been used on a number of very different audio products aimed at different markets, from pro audio recording (DSP2000, Delta1010) to desktop multi-media (DMX6fire, Aureon7.1) and so the 'best' solution in each can be different.

In the discussion you quoted, I suggested leaving the sample conversion out of the above definition of 'front', as unwanted conversion is always bad. If needed, use plug:front.

The other issue is that capture from 'front' does not make sense on all products. The DMX6fire has a differnt routing, with CD, line & phone inputs. A configuration for this unit would not suit the others. So the most general approach is needed.

Certainly adding the asym and slave parts to 'front' definition would be helpful in all cases IMO, as I proposed then, but preferably not format conversion.

I agree that the capture from "front" PCM isn't considered as valid. The "front", "rear", "center_lfe" definitions are rather for multi-channel playbacks. The capture on these channels aren't useful in most cases.

Takashi,

Arno's original post was just to the list, so I added your cc. His response was also only to the list, but has a patch at the bottom to do just playback asym with channels convertion, (which looks reasonable to me), so might be worth looking back at.

BTW, for example in the case of ice1712, is there a way for different sound cards which use the same driver to have different default config files? The DMX6fire in particular would benefit from specific definitions for the particular mapping of its 6 analogue inputs.

Alan