Regression in 2.6.19 and 2.6.20 for snd_hda_intel

Regression in 2.6.19 and 2.6.20 for snd_hda_intel

[Larry Finger]

I have a new laptop - an HP Pavilion dv2125nr with an AMD Turion 64 X2 Mobile CPU. With any kernel
later than 2.6.18, the sound does not work. The audio interface is reported by lspci as "Audio
device: nVidia Corporation MCP51 High Definition Audio (rev a2)", and uses the snd_hda_intel module.

Using git bisect, I determined that there are two problems. If commit a7da6ce564a8092..., which is
entitled "hda-codec-Add independent headphone volume control" is present, I get no sound at all.
When this commit is reverted, I get sound, but playing a sound file results in about an 0.5 sec
fragment being replayed over and over forever. If commit 7376d013fc6d3a45..., which is entitled
"Simple patch to enable Message Signalled Interrupts for the HDA Intel audio controller" is
reverted, the sound works fine.

Please tell me what other information is required.

Larry

Re: Regression in 2.6.19 and 2.6.20 for snd_hda_intel

[Daniel Drake]

Larry Finger wrote:
> When this commit is reverted, I get sound, but playing a sound file results in about an 0.5 sec
> fragment being replayed over and over forever. If commit 7376d013fc6d3a45..., which is entitled
> "Simple patch to enable Message Signalled Interrupts for the HDA Intel audio controller" is
> reverted, the sound works fine.

MSI is now disabled by default on this driver, so I'm pretty sure you 
only have the first bug talking in terms of the most recent kernels.

Daniel

Re: Regression in 2.6.19 and 2.6.20 for snd_hda_intel

[Larry Finger]

Daniel Drake wrote:
> Larry Finger wrote:
>> When this commit is reverted, I get sound, but playing a sound file
>> results in about an 0.5 sec
>> fragment being replayed over and over forever. If commit
>> 7376d013fc6d3a45..., which is entitled
>> "Simple patch to enable Message Signalled Interrupts for the HDA Intel
>> audio controller" is
>> reverted, the sound works fine.
> 
> MSI is now disabled by default on this driver, so I'm pretty sure you
> only have the first bug talking in terms of the most recent kernels.

You are correct. Only the "hda_codec-Add independent headphone volume control" needs to be reverted.

Thanks,

Larry

Re: Regression in 2.6.19 and 2.6.20 for snd_hda_intel

[Daniel Drake]

Larry Finger wrote:
> You are correct. Only the "hda_codec-Add independent headphone volume control" needs to be reverted.

The best course of action is probably to file a report here:
https://bugtrack.alsa-project.org/alsa-bug

Daniel

Re: [Alsa-devel] Regression in 2.6.19 and 2.6.20 for snd_hda_intel

[Larry Finger]

Tobin Davis wrote:
> Which alsa patch was this?  I'm not seeing anything in the hg logs for
> this.  Or is this something from the kernel side?

It seems to have come from suse. The full commit message is:

commit a7da6ce564a80952d9c0b210deca5a8cd3474a31
Author: Takashi Iwai <tiwai@suse.de>
Date:   Wed Sep 6 14:03:14 2006 +0200

    [ALSA] hda-codec - Add independent headphone volume control

    This patch addes the support of the independent 'Headphone' volume
    control to the generic codec parser.  Some codecs (e.g. Conexant)
    have separate connections to the headphone and the independent amp
    adjustment is needed.

    Signed-off-by: Takashi Iwai <tiwai@suse.de>
    Signed-off-by: Jaroslav Kysela <perex@suse.cz>

:100644 100644 dedfc5b... 97e9af1... M  sound/pci/hda/hda_generic.c

Larry

Re: [Alsa-devel] Regression in 2.6.19 and 2.6.20 for snd_hda_intel

[Stephen Hemminger]

On Sat, 30 Dec 2006 22:50:41 -0600
Larry Finger <larry.finger@lwfinger.net> wrote:

> Tobin Davis wrote:
> > Which alsa patch was this?  I'm not seeing anything in the hg logs for
> > this.  Or is this something from the kernel side?
> 
> It seems to have come from suse. The full commit message is:
> 
> commit a7da6ce564a80952d9c0b210deca5a8cd3474a31
> Author: Takashi Iwai <tiwai@suse.de>
> Date:   Wed Sep 6 14:03:14 2006 +0200
> 
>     [ALSA] hda-codec - Add independent headphone volume control
> 
>     This patch addes the support of the independent 'Headphone' volume
>     control to the generic codec parser.  Some codecs (e.g. Conexant)
>     have separate connections to the headphone and the independent amp
>     adjustment is needed.
> 
>     Signed-off-by: Takashi Iwai <tiwai@suse.de>
>     Signed-off-by: Jaroslav Kysela <perex@suse.cz>
> 
> :100644 100644 dedfc5b... 97e9af1... M  sound/pci/hda/hda_generic.c
> 
> Larry
> 
> 

Just make sure you don't break the headphone jack on some systems.
The headphone jack on the Panasonic T5 didn't work until recent kernels
(probably this was the thing that made it work).

-- 
Stephen Hemminger <shemminger@osdl.org>

Re: [Alsa-devel] Regression in 2.6.19 and 2.6.20 for snd_hda_intel

[Takashi Iwai]

At Sat, 30 Dec 2006 22:50:41 -0600,
Larry Finger wrote:
> 
> Tobin Davis wrote:
> > Which alsa patch was this?  I'm not seeing anything in the hg logs for
> > this.  Or is this something from the kernel side?
> 
> It seems to have come from suse. The full commit message is:
> 
> commit a7da6ce564a80952d9c0b210deca5a8cd3474a31
> Author: Takashi Iwai <tiwai@suse.de>
> Date:   Wed Sep 6 14:03:14 2006 +0200
> 
>     [ALSA] hda-codec - Add independent headphone volume control
> 
>     This patch addes the support of the independent 'Headphone' volume
>     control to the generic codec parser.  Some codecs (e.g. Conexant)
>     have separate connections to the headphone and the independent amp
>     adjustment is needed.
> 
>     Signed-off-by: Takashi Iwai <tiwai@suse.de>
>     Signed-off-by: Jaroslav Kysela <perex@suse.cz>
> 
> :100644 100644 dedfc5b... 97e9af1... M  sound/pci/hda/hda_generic.c

Make sure that you unmuted both Master and Headphone volumes (if you
have).
With this patch, you may have now two individual controls while the
older version has only one.


Takashi