Re: [Protux-devel] some diplomacy...

[Martin Herren]

On Sun, 02 Feb 2003 03:10:39 -0200
Luciano Giordana <address@hidden> wrote:

> oh my God, somebody understands my pain...
> 
> I am experiences EXACLTY the same...
> 
> so I decided.
> 
> until I find some good soul to explain what is a "DEVICE" for ALSA, and what
> is PCM, subdevice, subunit dir, configuration space, and all other 
> non-standard ideas they introduced in audio world..., I will make this : 
> MustuxAudioDeviceMapper will look for stereo buses, which is any THING 
> (device, pcm, whatever) that can play and capture in stereo mode,
> 
> when it is found, libmustux will , itself, handle the subdevicing. If user 
> wants to record only from right channel, MustuxAudioDevice mapper will STILL 
> record in stereo, but will make only the right channel data visible to user. 
> Even if different tracks gets different channels, it will be transparent to 
> the application. if MustuxAudioDeviceMapper cannot find stereo buses, it will 
> allow only mono bus operation, so each track will be able to receive/send 
> audio data only from/to 1 bus each
> 
> Maybe it will be slower, but it will work similarly to VEGAS. And this will 
> be 
> like this until somebody decides to write ALSA docs for normal people, and 
> not olny for ALSA developers (you can see I am really angry with ALSA docs)

yes i think that's the only way for the moment.

In the alsa doxygen-doc there is a example code for generating a sinus, and you 
can choose a lot of options, like access method (interleaved or 
non-interleaved), device/subdevice. I played with it, and never could get 
anything but a stereo signal, even when i set the number of channels to 1.
Under /proc/asound/Live i've got pcm0c, pcm0p ,pcm1c, pcm2c and pcm3p, that 
means 2 playback devices.

When i playback the sinus using device plughw:0,0 (or plughw:0) i hear the 
sinus. With plughw:0,3 it seems to work too, but i don't hear anything (it's 
perhaps the digital out).
Using hw:0,0 is also ok, but hw:0,3 gives an error (Rate doesn't match).

using plughw:0,1 or plughw:0,2 generates an error (no such device), which is ok 
are pcm1 and pcm2 are only capture.

so there the mapping between /proc pcm device and alsa-device seems to be:
pcmX -> (plug)hw:Y,X
where Y is the number of the card.

But the different pcmXp/c devices under /proc seems not to pe different ways of 
accessing the card, but really different buses on the card.

So my wild guess for my SB-Live:

pcm0c: line in
pcm0p: line out
pcm1c: mic in
pcm2c: CD in
pcm3p: digital out

i'll try to make some test to check that (perhaps pcm1c and 2c are inverted)

But everything seems to say that splitting a stereo signal into two monos 
should be done in the application. But perhaps accessing the bus as 
non-interleaved will make it easier (faster) to split it ?
 
> ps - for example : I am using those snd_hw_test_rate(...) functions, to test 
> if a device/bus/pcm/anything (from now on I will call is the ALSA 
> CRAPICES...), can support some rate. I tryed 1GHz on a Sound Blaster 16 !!!! 
> and it returns OK !!!! Isnt it crappy ?

As far as i understood alsa yet, is that there are 2 ways of accessing a card. 
Either directly, or through a plugin. That's probably the difference between a 
hw:y,x and a plughw:y,x device. If you access it through a plugin, you can use 
any rate, and the plugin will convert it to a rate that is supported by the 
card. If you make direct access, you have to check if the rate is really 
supported.

So did you access your SB16 through a plugin or directly ?
If you accessed it through a plugin, it should be normal that you get OK, alsa 
will just downsample the data before really sending it to the card.

If you made a direct access, perhaps we could write a driver to use all those 
old sb16's as wireless network cards... (if they can do 1 GHz, they probably 
can do 2.4 GHz as well)
just joking.

/Martin

-- 
Open your Windows - Free your Mind - Enjoy
http://gnuwin.epfl.ch

Martin Herren                             +41 (0)79 746 57 83
OpenPGP Public key @ http://www.on-the-web.ch/sputnik/gpg.asc

pgpAFelTgxs5t.pgp
Description: PGP signature