Via 82xx drivers

Via 82xx drivers

[John Matthews]

[-- Attachment #1: Type: text/plain, Size: 2423 bytes --]

Hi,

I'm trying to develop a complete system / distribution using the via "mini-itx" platform.  The idea is that we take in sensor positions from analog sensors and output sounds based on the position of these sensors. This setup is used for music therapy, with disabled children producing music using, for example, ultrasonic position sensors (amongst other types of sensor). I've written some code to accept input from the alsa sequencer and act as a midi router, and I'm using fluidsynth as an output synthesiser.

The mini-itx board is a very small form factor fully integrated pc motherboard  with a via c3 processor.  The sound controller on this board is part of a via 8235 southbridge.

The snd-via82xx alsa module works fine on this board, apart from the fact that it is very, very noisy - unusably so.  There is a background drone and whenever the processor does something you get "clunky" noises, which i'm pretty sure aren't xruns.  I have tested it using fluidsynth and very large buffer sizes, to eliminate the possibility of xruns.  I have ensured everything in the alsamixer is turned down.

Particularly, when a note is turned off in fluidsynth, shortly after you get a pop. 

I have tried using all of the via82xx driver's options, ie. dxs_support=0,1,2,3, ac_97_quirk=0,1,2,3,4 to no avail.  I have also tried passing no-hlt to the kernel, which helps just a little.

Interestingly the 2.4 kernel OSS drivers do the same.  VIAs own OSS drivers do not, and are very quiet, but their latency is quite high.  The source is available to all of these drivers. The windows drivers are fairly quiet as well.

I am pretty sure this is a problem with the alsa drivers using this particular chip, as I have a few friends who are experienced with alsa attempting to use the same board with similar problems.  Also, I have used the same setup (fluidsynth / alsa) on a via 8233a based board and it worked fine.

It could be as simple as an unnaccesible mixer channel being left unmuted.

Has anyone else experienced this problem?  Does anyone have any tips or advice?

I would really like to fix this, as I am a professional programmer, though I have no experience of linux device drivers, and only a little experience in linux generally.  Is there a guide to how alsa drivers work anywhere?  Any information which would help me sort this would be gratefully recieved,

Thanks,

John Matthews.

[-- Attachment #2: Type: text/html, Size: 4021 bytes --]

Via 82xx Drivers

[John Matthews]

[-- Attachment #1: Type: text/plain, Size: 1536 bytes --]

Greetings,

I am writing to describe a problem I have involving the snd-via82xx audio drivers on 2.6 kernels.  I have updated to the latest alsa-drivers to no avail.

Everything works fine, apart from one small issue.  Certain programs have strange clicks at certain points.  The main example is with fluidsynth http://www.nongnu.org/fluid/   When a note is turned off there is a very loud click which can be louder than the note was - the same type of click that happens when you turn a channel on or off.

This doesn't happen with any other sound card that I have tried (tried a few others including snd-ice1712 on an m-audio delta 410).  It also doesn't happen with the via binary oss style drivers,  "viaudiocombo" which are available from via's website.

The clicks do happen (suprisingly) in exactly the same places with fluidsynth running into jack into snd-via82xx.

I have tried this on various different boards I have around, although they are all types of VIA epia mini-itx boards.

I have also tried dxs_support=1, dxs_support=2,dxs_support=3, dxs_support=4 and this doesn't help.

It may well be the same problem which was noted here.  https://bugtrack.alsa-project.org/alsa-bug/view.php?id=1667

Its meant that I have to use the viaudiocombo drivers for this purpose, and I'd rather use the nice up to date alsa ones!

Please, I would be grateful for any help or thoughts - I am an experienced coder, though I've never worked with device drivers before,

Thanks,

John Matthews,
University of York

[-- Attachment #2: Type: text/html, Size: 3156 bytes --]

Re: Via 82xx Drivers

[Takashi Iwai]

At Wed, 24 May 2006 13:27:50 +0100,
John Matthews wrote:
> 
> Greetings,
>  
> I am writing to describe a problem I have involving the snd-via82xx audio drivers on
> 2.6 kernels.  I have updated to the latest alsa-drivers to no avail.
>  
> Everything works fine, apart from one small issue.  Certain programs have strange
> clicks at certain points.  The main example is with fluidsynth http://www.nongnu.org/
> fluid/   When a note is turned off there is a very loud click which can be louder than
> the note was - the same type of click that happens when you turn a channel on or off.
>  
> This doesn't happen with any other sound card that I have tried (tried a few others
> including snd-ice1712 on an m-audio delta 410).  It also doesn't happen with the via
> binary oss style drivers,  "viaudiocombo" which are available from via's website.
>  
> The clicks do happen (suprisingly) in exactly the same places with fluidsynth running
> into jack into snd-via82xx.
>  
> I have tried this on various different boards I have around, although they are all
> types of VIA epia mini-itx boards.
>  
> I have also tried dxs_support=1, dxs_support=2,dxs_support=3, dxs_support=4 and this
> doesn't help.

Did you try the latest version of ALSA driver?
Also, make sure that dxs_support=5 option, e.g. see
/sys/module/snd_via82xx/parameters/*.

If the problem still persists with the latest driver, try to play with
the secondary PCM device (hw:0,1) instead of the default one.
(Assumed that your device is no VIA8233"A".  8233/8235/8237/8251 have
 DXS channels while 8233A not.)


Takashi

Re: Via 82xx Drivers

[John Matthews]

Thanks for the reply.

dxs_support=5 doesn't help either.  Neither does using the other hardware
device.

I've discovered that the click appears as the card starts to output absolute
silence.  Adding a constant low level sinusoid which is so small it is
inaudible to the output stops the clicking.

Does the chip get put into some sort of sleep mode or are the pcm channels 
muted when
absolute silence occurs?

Cheers,

John.

----- Original Message ----- 
From: "Takashi Iwai" <tiwai@suse.de>
To: "John Matthews" <jm169@ohm.york.ac.uk>
Cc: <alsa-devel@lists.sourceforge.net>
Sent: Friday, May 26, 2006 12:15 PM
Subject: Re: [Alsa-devel] Via 82xx Drivers


> At Wed, 24 May 2006 13:27:50 +0100,
> John Matthews wrote:
>>
>> Greetings,
>>
>> I am writing to describe a problem I have involving the snd-via82xx audio 
>> drivers on
>> 2.6 kernels.  I have updated to the latest alsa-drivers to no avail.
>>
>> Everything works fine, apart from one small issue.  Certain programs have 
>> strange
>> clicks at certain points.  The main example is with fluidsynth 
>> http://www.nongnu.org/
>> fluid/   When a note is turned off there is a very loud click which can 
>> be louder than
>> the note was - the same type of click that happens when you turn a 
>> channel on or off.
>>
>> This doesn't happen with any other sound card that I have tried (tried a 
>> few others
>> including snd-ice1712 on an m-audio delta 410).  It also doesn't happen 
>> with the via
>> binary oss style drivers,  "viaudiocombo" which are available from via's 
>> website.
>>
>> The clicks do happen (suprisingly) in exactly the same places with 
>> fluidsynth running
>> into jack into snd-via82xx.
>>
>> I have tried this on various different boards I have around, although 
>> they are all
>> types of VIA epia mini-itx boards.
>>
>> I have also tried dxs_support=1, dxs_support=2,dxs_support=3, 
>> dxs_support=4 and this
>> doesn't help.
>
> Did you try the latest version of ALSA driver?
> Also, make sure that dxs_support=5 option, e.g. see
> /sys/module/snd_via82xx/parameters/*.
>
> If the problem still persists with the latest driver, try to play with
> the secondary PCM device (hw:0,1) instead of the default one.
> (Assumed that your device is no VIA8233"A".  8233/8235/8237/8251 have
> DXS channels while 8233A not.)
>
>
> Takashi
>