caiaq driver: audio 4 dj and jackd

caiaq driver: audio 4 dj and jackd

[Gabriel M. Beddingfield]

Is it possible to make all 4 ins/outs in the Audio 4 DJ available 
through the first substream device?

A user was having trouble using the NI Audio 4 DJ device with jackd. 
When jack connected, this 4-in/4-out device would appear as 2-in/2-out. 
This was the result of the fact that the channels are split up into 
stereo substreams.  I.e. the first two channels are found in hw:x,0,0 
and the last two channels are in hw:x,0,1.

The work-around was to use the `multi' plug-in to make these 2 PCM's 
appear as a single PCM.

There's several situations where this is a Good Thing, because it makes 
it easy for the user to treat the streams independently.

BUT... in the pro-audio case this is not a good thing at all.

Ideally, substream 0 would support 1, 2, or 4 channels (for both 
playback and capture).  When substream 0 is using more than 2 channels, 
then substream 1 would return EBUSY when in use.

So... would it be possible to do this in the caiaq driver?

-gabriel

Re: caiaq driver: audio 4 dj and jackd

[Daniel Mack]

On Sun, Dec 01, 2013 at 07:02:27PM -0800, Gabriel M. Beddingfield wrote:
> A user was having trouble using the NI Audio 4 DJ device with jackd.
> When jack connected, this 4-in/4-out device would appear as
> 2-in/2-out. This was the result of the fact that the channels are
> split up into stereo substreams.  I.e. the first two channels are
> found in hw:x,0,0 and the last two channels are in hw:x,0,1.

That's right, but I don't see a problem with that. Why was jack having
trouble using the two interfaces? After all, the streams are guaranteed
to be synchronous, as they share the same clock in hardware.

> The work-around was to use the `multi' plug-in to make these 2 PCM's
> appear as a single PCM.

Yes, that's the other possibility.

> Ideally, substream 0 would support 1, 2, or 4 channels (for both
> playback and capture).  When substream 0 is using more than 2
> channels, then substream 1 would return EBUSY when in use.
> 
> So... would it be possible to do this in the caiaq driver?

Technically yes, but changing that would mean breakage for all other
users, and making it configurable would result in yet another moving
part which makes things even more complex.


Thanks,
Daniel

Re: caiaq driver: audio 4 dj and jackd

[Gabriel M. Beddingfield]

On 12/02/2013 01:59 AM, Daniel Mack wrote:
>> split up into stereo substreams.  I.e. the first two channels are
>> found in hw:x,0,0 and the last two channels are in hw:x,0,1.
>
> That's right, but I don't see a problem with that. Why was jack having
> trouble using the two interfaces?

Jack has always opened only one PCM device... which is the Right Thing 
for most recording sound cards (e.g. ice1712, hammerfall, etc.). This is 
the first "pro-sumer" device I've ever seen with multiple substream devices.

> After all, the streams are guaranteed
> to be synchronous, as they share the same clock in hardware.

TBH, the whole substream concept is lost on me.  Where does this 
guarantee come from?  From the substream concept or from intimate 
knowledge of this device?  (I.e. how would JACK know about the clock 
guarantee?)

...and is the the substream concept (in ALSA) trying to solve?

>> Ideally, substream 0 would support 1, 2, or 4 channels (for both
>> playback and capture).  When substream 0 is using more than 2
>> channels, then substream 1 would return EBUSY when in use.
>>
>> So... would it be possible to do this in the caiaq driver?
>
> Technically yes, but changing that would mean breakage for all other
> users, and making it configurable would result in yet another moving
> part which makes things even more complex.

How would this cause breakage?  Users that use 2 stereo substreams could 
still work normally.

But, FULL ACK on the moving parts.  That would indeed make the driver a 
little gross on the inside.

-gabe

Re: caiaq driver: audio 4 dj and jackd

[Daniel Mack]

On 12/02/2013 04:15 PM, Gabriel M. Beddingfield wrote:
> On 12/02/2013 01:59 AM, Daniel Mack wrote:

> Jack has always opened only one PCM device... which is the Right Thing 
> for most recording sound cards (e.g. ice1712, hammerfall, etc.). This is 
> the first "pro-sumer" device I've ever seen with multiple substream devices.

I was also uncertain when I wrote the driver which way is better. But
given that there are ways in ALSA to synchronously start separate
streams, I eventually went the stereo pairs way.

>> After all, the streams are guaranteed
>> to be synchronous, as they share the same clock in hardware.
> 
> TBH, the whole substream concept is lost on me.  Where does this 
> guarantee come from?  From the substream concept or from intimate 
> knowledge of this device? 

Both :)

> (I.e. how would JACK know about the clock 
> guarantee?)

snd_pcm_info_get_sync() should return the same clock id for both
interfaces in that case. But I must admit I haven't tried this in awhile.

>> Technically yes, but changing that would mean breakage for all other
>> users, and making it configurable would result in yet another moving
>> part which makes things even more complex.
> 
> How would this cause breakage?  Users that use 2 stereo substreams could 
> still work normally.

Yes, unless you opened the first subdevice with more than 2 channels.
Then an existing user for the second stream breaks. IOW: the possible
configuration and usage for subdevice 1 changes depending on that of
subdevice 0.

> But, FULL ACK on the moving parts.  That would indeed make the driver a 
> little gross on the inside.

What are your results with the multi plugin? Is that something you can
live with?


Daniel

Re: caiaq driver: audio 4 dj and jackd

[Gabriel M. Beddingfield]

On 12/02/2013 08:00 AM, Daniel Mack wrote:
>> TBH, the whole substream concept is lost on me.  Where does this
>> guarantee come from?  From the substream concept or from intimate
>> knowledge of this device?
>
> Both :)
>
>> (I.e. how would JACK know about the clock
>> guarantee?)
>
> snd_pcm_info_get_sync() should return the same clock id for both
> interfaces in that case. But I must admit I haven't tried this in awhile.

Thanks!  I'll have to play around with that.

>> But, FULL ACK on the moving parts.  That would indeed make the driver a
>> little gross on the inside.
>
> What are your results with the multi plugin? Is that something you can
> live with?

I don't actually have one of these devices and the end-user I was 
helping is happy now, AFAIK.

But solving this required editing asoundrc files (ugh) and casting 
several magic spells.  Would be nice if it "just works."

Anyway, perhaps the solution is to get JACK to recognize the extra 
substreams.

-gabriel

Re: caiaq driver: audio 4 dj and jackd

[Alan Horstmann]

On Monday 02 December 2013 16:14, Gabriel M. Beddingfield wrote:
> On 12/02/2013 08:00 AM, Daniel Mack wrote:
> >> TBH, the whole substream concept is lost on me.  Where does this
> >> guarantee come from?  From the substream concept or from intimate
> >> knowledge of this device?
> >
> > Both :)

<snip>

> > What are your results with the multi plugin? Is that something you can
> > live with?
>
> I don't actually have one of these devices and the end-user I was
> helping is happy now, AFAIK.
>
> But solving this required editing asoundrc files (ugh) and casting
> several magic spells.  Would be nice if it "just works."

The same problem occurs with Portaudio on this device (or similar) - it would 
be great to just be able to open one device with 4 channels, or 2 stereo 
pairs.

I had wondered whether it could be done by some card-specific definitions in 
eg /usr/share/alsa/cards as is done particularly with PCI cards??  Anybody 
any ideas?

Gabriel, could you post the multi pcm you worked out for this, as it may help 
others?

Regards

Alan

Re: caiaq driver: audio 4 dj and jackd

[Gabriel M. Beddingfield]

On 12/02/2013 11:48 AM, Alan Horstmann wrote:
>> But solving this required editing asoundrc files (ugh) and casting
>> several magic spells.  Would be nice if it "just works."
>
> The same problem occurs with Portaudio on this device (or similar) - it would
> be great to just be able to open one device with 4 channels, or 2 stereo
> pairs.
>
> I had wondered whether it could be done by some card-specific definitions in
> eg /usr/share/alsa/cards as is done particularly with PCI cards??  Anybody
> any ideas?

Good idea.

>
> Gabriel, could you post the multi pcm you worked out for this, as it may help
> others?

I can't take credit for figuring it out.  Got it from here:

http://ywwg.com/wordpress/?p=772

-gabe

Re: caiaq driver: audio 4 dj and jackd

[Gabriel M. Beddingfield]

On 12/02/2013 08:00 AM, Daniel Mack wrote:
>>> Technically yes, but changing that would mean breakage for all other
>>> users, and making it configurable would result in yet another moving
>>> part which makes things even more complex.
>>
>> How would this cause breakage?  Users that use 2 stereo substreams could
>> still work normally.
>
> Yes, unless you opened the first subdevice with more than 2 channels.
> Then an existing user for the second stream breaks. IOW: the possible
> configuration and usage for subdevice 1 changes depending on that of
> subdevice 0.

Suppose a 2nd PCM device is added that would allow 1-4 channel 
playback/capture (call it "Multichannel Audio" or something).  This pcm1 
would be mutually exclusive with pcm0 (I.e. while one is open you can't 
use the other).

Existing applications using hw:n,0,0 and hw:n,0,1 see the exact same 
interface/behavior.  The only failure mode is if the multichannel PCM is 
in use.

Applications using multichannel work-flows would use hw:n,1.

While it's not a "just works" solution -- directing users to hw:n,1 is a 
whole lot better than teaching them how to set up a multi plugin.

-gabe

Re: caiaq driver: audio 4 dj and jackd

[Gabriel M. Beddingfield]

Hi Daniel,

On 12/04/2013 08:11 PM, Gabriel M. Beddingfield wrote:
> On 12/02/2013 08:00 AM, Daniel Mack wrote:
> Suppose a 2nd PCM device is added that would allow 1-4 channel
> playback/capture (call it "Multichannel Audio" or something).  This pcm1
> would be mutually exclusive with pcm0 (I.e. while one is open you can't
> use the other).
>
> Existing applications using hw:n,0,0 and hw:n,0,1 see the exact same
> interface/behavior.  The only failure mode is if the multichannel PCM is
> in use.
>
> Applications using multichannel work-flows would use hw:n,1.
>
> While it's not a "just works" solution -- directing users to hw:n,1 is a
> whole lot better than teaching them how to set up a multi plugin.

I'm considering buying one of these Audio 4 DJ devices to make this 
change.  Would you be willing to take a change like this?

Thanks,
Gabe

Re: caiaq driver: audio 4 dj and jackd

[Daniel Mack]

Hi Gabe,

On 12/10/2013 03:27 AM, Gabriel M. Beddingfield wrote:
> On 12/04/2013 08:11 PM, Gabriel M. Beddingfield wrote:
>> On 12/02/2013 08:00 AM, Daniel Mack wrote:
>> Suppose a 2nd PCM device is added that would allow 1-4 channel
>> playback/capture (call it "Multichannel Audio" or something).  This pcm1
>> would be mutually exclusive with pcm0 (I.e. while one is open you can't
>> use the other).
>>
>> Existing applications using hw:n,0,0 and hw:n,0,1 see the exact same
>> interface/behavior.  The only failure mode is if the multichannel PCM is
>> in use.
>>
>> Applications using multichannel work-flows would use hw:n,1.
>>
>> While it's not a "just works" solution -- directing users to hw:n,1 is a
>> whole lot better than teaching them how to set up a multi plugin.
> 
> I'm considering buying one of these Audio 4 DJ devices to make this 
> change.  Would you be willing to take a change like this?

Well, yes. The longer I think about it, the more I believe it would be
worth a try how such a modified driver behaves.

I might do it myself, but it'll take a while until I can start working
on it. If you are willing to help, I'd happily test a patch :)


Daniel