On Mon, Apr 20, 2009 at 10:23:30AM +0300, Peter Ujfalusi wrote:
I have been thinking of implementing this in core in a similar manner as I have implemented in the twl4030 codec, but I'm not sure that the assumption of having a maximum of two streams (one for playback and one for capture) holds for all codec/machine pairs. If it does, I can improve the core's constraint handling with the approach taken in the twl4030 codec. With my limited tests it passed all cases that I could come up...
The single streams are assumed in the ASoC core so there's not much concern there. Without working on the ALSA core itself there's not much you can do to fully plug all the holes, though.
SNDRV_PCM_HW_PARAM_SAMPLE_BITS,SNDRV_PCM_HW_PARAM_CHANNELS,
Are you sure that these need to match exactly and don't set maxima?
These settings are for the audio interface (HiFi interface), which means they are changing the interface format for both playback and capture. On top of that in 2 channel mode the interface has to be in I2S mode, in 4 channel mode it has to be in DSP_A mode. So I think they have to be exact match in order to not to break the existing stream.
Oh, nice. With some care it should be possible to allow at least a maximum bit rate setting (since lower bit rates are going to be wire compatible if enough clocks are generated).