[alsa-devel] ASOC, SPDIF and AC97
I'm back to working on an ASOC driver for the STAC9766 used in the MPC5200 based Efika.
I'm having trouble with SPDIF. I can make two devices like this in the codec file. But how do I get the SPDIF device hooked up to the right AC97 slot, 10/11? Are any of the ASOC codecs implementing SPDIF?
I also see that the STAC9766 is missing an ac97_patch. From my reading of the datasheet it needs somethings fixed up.
Does the ac97 support belong in the PCI directory? It took me a while to locate it while working on ASOC. With a model more like ASOC codecs, the AC97 patches could be pushed into the codec files.
static struct snd_soc_dai stac9766_dai[] = { { .name = "STAC9766-Analog", .id = 0, /* stream cababilities */ .playback = { .stream_name = "STAC9766 Analog", .channels_min = 1, .channels_max = 6, .rates = STAC9766_RATES, .formats = SNDRV_PCM_FMTBIT_S16, }, .capture = { .stream_name = "STAC9766 Analog", .channels_min = 1, .channels_max = 2, .rates = STAC9766_RATES, .formats = SNDRV_PCM_FMTBIT_S16, }, /* alsa ops */ .prepare = ac97_analog_prepare, }, { .name = "stac9766-S/PDIF", .id = 1, /* stream cababilities */ .playback = { .stream_name = "STAC9766 Digital", .channels_min = 1, .channels_max = 2, .rates = STAC9766_RATES, .formats = SNDRV_PCM_FMTBIT_S16, }, /* alsa ops */ .prepare = ac97_digital_prepare, }};
On Tue, May 27, 2008 at 02:16:54PM -0400, Jon Smirl wrote:
I'm having trouble with SPDIF. I can make two devices like this in the codec file. But how do I get the SPDIF device hooked up to the right AC97 slot, 10/11? Are any of the ASOC codecs implementing SPDIF?
None that I know of implementing SPDIF (or at least, none with driver support for it) but it might be worth taking a look at the PXA2xx AC97 controller support - it implements supports for AC97 timeslots beyond the base playback and record channels.
Briefly, the approach is to provide a DAI in the AC97 controller driver which handles the relevant AC97 timeslots and a DAI in the CODEC driver which does the CODEC side of things then match the two together.
I also see that the STAC9766 is missing an ac97_patch. From my reading of the datasheet it needs somethings fixed up.
Does the ac97 support belong in the PCI directory? It took me a while to locate it while working on ASOC. With a model more like ASOC codecs, the AC97 patches could be pushed into the codec files.
The ASoC AC97 support doesn't use the AC97 patch infrastructure at all - it's completely independent so things are pretty much as you describe, everything is in the CODEC drivers.
On 5/28/08, Mark Brown broonie@opensource.wolfsonmicro.com wrote:
On Tue, May 27, 2008 at 02:16:54PM -0400, Jon Smirl wrote:
I'm having trouble with SPDIF. I can make two devices like this in the codec file. But how do I get the SPDIF device hooked up to the right AC97 slot, 10/11? Are any of the ASOC codecs implementing SPDIF?
None that I know of implementing SPDIF (or at least, none with driver support for it) but it might be worth taking a look at the PXA2xx AC97 controller support - it implements supports for AC97 timeslots beyond the base playback and record channels.
I'll look at PXA2xx. I had assumed that some of the SOC codecs were implementing SPDIF and that wrongly led me into the pci code.
I'm practicing on the Efika while I wait on our own hardware. The Efika hardware can support simultaneous independent playback on analog and spdif. I see now that it is too hard to jump straight to independent playback, instead I'll work on getting them both to play standalone.
For independent playback I need to interleave the data from two buffers and DMA it into the PSC. I believe the Bestcomm DMA controller can do this. Does ALSA hand me the two buffers simultaneously? I haven't located any code yet supporting independent playback.
Briefly, the approach is to provide a DAI in the AC97 controller driver which handles the relevant AC97 timeslots and a DAI in the CODEC driver which does the CODEC side of things then match the two together.
I also see that the STAC9766 is missing an ac97_patch. From my reading of the datasheet it needs somethings fixed up.
Does the ac97 support belong in the PCI directory? It took me a while to locate it while working on ASOC. With a model more like ASOC codecs, the AC97 patches could be pushed into the codec files.
The ASoC AC97 support doesn't use the AC97 patch infrastructure at all - it's completely independent so things are pretty much as you describe, everything is in the CODEC drivers.
I've figured out now that the pci codec code is independent of the asoc code. Are there plans to unify this in the future?
On Wed, May 28, 2008 at 09:40:13AM -0400, Jon Smirl wrote:
can do this. Does ALSA hand me the two buffers simultaneously? I haven't located any code yet supporting independent playback.
It shouldn't be fundamentally different to multiple independent records which the PXA code does support. If you present them as separate PCMs then ALSA won't synchronise them for you.
The ASoC AC97 support doesn't use the AC97 patch infrastructure at all - it's completely independent so things are pretty much as you describe, everything is in the CODEC drivers.
I've figured out now that the pci codec code is independent of the asoc code. Are there plans to unify this in the future?
I'm not aware of anyone working on doing so.
On 5/28/08, Mark Brown broonie@opensource.wolfsonmicro.com wrote:
On Wed, May 28, 2008 at 09:40:13AM -0400, Jon Smirl wrote:
can do this. Does ALSA hand me the two buffers simultaneously? I haven't located any code yet supporting independent playback.
It shouldn't be fundamentally different to multiple independent records which the PXA code does support. If you present them as separate PCMs then ALSA won't synchronize them for you.
Any words of wisdom on how to synchronize?
The ASoC AC97 support doesn't use the AC97 patch infrastructure at all - it's completely independent so things are pretty much as you describe, everything is in the CODEC drivers.
I've figured out now that the pci codec code is independent of the asoc code. Are there plans to unify this in the future?
I'm not aware of anyone working on doing so.
On Wed, May 28, 2008 at 04:15:40PM -0400, Jon Smirl wrote:
On 5/28/08, Mark Brown broonie@opensource.wolfsonmicro.com wrote:
It shouldn't be fundamentally different to multiple independent records which the PXA code does support. If you present them as separate PCMs then ALSA won't synchronize them for you.
Any words of wisdom on how to synchronize?
It's not a situation I've run into myself. I'd try something like having the DMA driver delay the application starting the second stream but I've no idea if that'd work or do what you need.
participants (2)
-
Jon Smirl -
Mark Brown