At Thu, 14 Feb 2008 17:59:05 +0100, Rainer Zimmermann wrote:

Hi Takashi and Pavel,

Takashi Iwai wrote:

...

...

The constraint comes from the codec, the chip itself can handle 192kHz, e.g. the 192kHz SPDIF input works fine. Actually more of the ice1724 cards have this limit. If this issue is solved in a general way, it could be implemented to other cards with 192kHz DACs/96kHz ADCs too.

Thanks for a quick information. Then, yes, we should fix this issue in general. The separate lists of rates for playback and capture would be required in the end...

ok. Actually, as Pavel mentioned SPDIF, I guess there should be a third set of rates for SPDIF, to be used when the appropriate channel is switched to SPDIF.

That's a good point.

...

...

...

Anyway, we can limit this in a scenario like below: - disallow the rate over 96kHz if the capture stream is being opened - if the rate is set above 96kHz, the capture stream cannot be opened, returns -EBUSY or so.

Yes, I was thinking of something along those lines. But 2 possible issues:

• maybe a stupid question, but couldn't there be some "backdoor"? E.g., could a

playback stream change its rate to above 96kHz while a capture stream is open?

• In the latter case, I think it should read: "if a playback stream is running

at above 96kHz, the capture stream cannot be opened". Otherwise, e.g. arecord couldn't open the device if another app left the rate at 192kHz.

Yes, the defnial of opening a capture stream over 96kHz is a drawback in this method.

Are there other examples (non-ice17xx) how this is handled?

Either setting a fixed rate via a control or disallow-streams-that- don't-support way.

...

...

...

BTW, is it supposed to be still highly experimental? If the driver works more or less stably with your device (and on the recent kernel), it'd be also good to put this to the upstream, i.e. in alsa-kernel tree instead of alsa-driver tree.

Yes, so far the driver is stable on my system (2.6.22) and on Piotr's, but I would still consider it experimental. It still contains experimental code and some experimental controls, and I guess there should be more testing, in particular the MI/ODI/O and SPDIF part is still untested. Well, your decision...

It's rather your decision at this stage since I have no hardware for testing. I'd just like to mention that merging to the upstream gives more users a chance to try.

As for the vu-meters:

...

Since the meaning of these read-only controls is clear only upon detailed study of the driver and Envy24 datasheet plus I encounter a lot of questins/complaints about them, I suggest to change their type from MIXER to PCM. They would still be readily accessible from amixer or API.

No objection from my part, but could it break any app software? any other comments?

Well, I think it's fine at this moment since we have (still) no envy24control for vt172x. Actually, some other controls are also candidates to hide from the mixer interface. We need a more detailed review in this regard.

Takashi