Re: regression in 2.6.30-rc6 -- via82xx sound very quiet

[Richard Kennedy <richard@rsk.demon.co.uk>]

On Mon, 2009-05-18 at 09:56 +0200, Takashi Iwai wrote:
> At Mon, 18 May 2009 08:52:34 +0100,
> Richard Kennedy wrote:
> > 
> > On Mon, 2009-05-18 at 08:04 +0200, Takashi Iwai wrote:
> > > At Sun, 17 May 2009 21:38:03 +0100,
> > > Richard Kennedy wrote:
> > > > 
> > > > On Sun, 2009-05-17 at 19:16 +0200, Takashi Iwai wrote:
> > > > > At Sun, 17 May 2009 17:47:54 +0100,
> > > > > Richard Kennedy wrote:
> > > > > > 
> > > > > > In the current git 2.6.30-rc6 the sound from my motherboard has become
> > > > > > extremely quiet, I have to turn my speakers up all the way to hear
> > > > > > anything at all.
> > > > > > I'm not too sure when the problem first started but I've just tested
> > > > > > 2.6.30-rc4 and it works correctly.  
> > > > > 
> > > > > Try to correct "DXS Playback Volume" mixer elements.
> > > > > The dB range of these volumes were fixed recently to the correct
> > > > > values.  If the old mixer value is restored using the dB value,
> > > > > it would be restored to a much lower value, which explains your
> > > > > symptom.
> > > > > 
> > > > > Simply set these volumes to the max (i.e. 0dB) and never change unless
> > > > > you really need them.
> > > > > 
> > > > > 
> > > > > thanks,
> > > > > 
> > > > > Takashi
> > > > 
> > > > Hi Takashi,
> > > > 
> > > > thanks for that. Yes that works.
> > > 
> > > Thanks for checking.
> > > 
> > > > I didn't even know I had that control, so I guess this could be a common
> > > > problem.
> > > 
> > > Yes, unfortunately, it would hit to many via82xx ac97 chips.
> > > As the original value was wrong, it can't be corrected so easily from
> > > the driver side.  Manual corrections seem needed, but just only once.
> > > 
> > > 
> > > Takashi
> > 
> > Yes it's easy to do.
> > But it would be nice if we could give the users a hint as to what's gone
> > wrong. Would it be ok to add a log message at startup ? 
> > 
> > something like "your DXS volume control may need reseting".
> 
> Unfortunately this can't be detected by the system because the old
> stored value is still a valid value, just less than expected.
> 
> But yes, we might need to inform this somewhere in a better place.
> 
> 
> Takashi

What about adding something like this ? 

I know it's a bit ugly and it will have to be removed at some point, but
it would at least give the users a clue.

regards
Richard


diff --git a/sound/pci/via82xx.c b/sound/pci/via82xx.c
index 1ef58c5..e856f0a 100644
--- a/sound/pci/via82xx.c
+++ b/sound/pci/via82xx.c
@@ -2465,8 +2465,10 @@ static int __devinit snd_via82xx_probe(struct pci_dev *pci,
 			 */
 			if (dxs_support == VIA_DXS_DISABLE)
 				chip_type = TYPE_VIA8233A;
-			else
+			else {
 				chip_type = TYPE_VIA8233;
+				printk (KERN_WARNING "via82xx - 2.6.30 improved DXS support. User may need to reset DXS control\n");
+			}
 		}
 		if (chip_type == TYPE_VIA8233A)
 			strcpy(card->driver, "VIA8233A");