Re: [REGRESSION][BISECTED] Audio volume issues since 4178d78cd7a8

[Christoffer Sandberg <cs@xxxxxxxxx>]

On 2.10.2024 23:28, Jerry Luo wrote:
> On 10/2/24 10:00 AM, Takashi Iwai wrote:
> > On Wed, 02 Oct 2024 10:21:22 +0200,
> > Christoffer Sandberg wrote:
> > > On 30.9.2024 09:44, Takashi Iwai wrote:
> > > > On Mon, 23 Sep 2024 21:37:42 +0200,
> > > > Jerry Luo wrote:
> > > > > Hi Takashi,
> > > > >
> > > > > On Mon, 16 Sep 2024 19:22:05 +0200,
> > > > >
> > > > > Takashi Iwai wrote:
> > > > >
> > > > >      Could you give alsa-info.sh output from both working and
> > > > > non-working
> > > > >      cases?  Run the script with --no-upload option and attach the
> > > > > outputs.
> > > > >
> > > > >      thanks,
> > > > >
> > > > >      Takashi
> > > > >
> > > > > Issue now reappear, output from alsa-info.sh are attached. If they
> > > > > are still
> > > > > needed.
> > > > Thanks.  The obvious difference seems to be the assignment of two 
> > > > DACs
> > > > 0x10 and 0x11 for headphone and speaker outputs.
> > > >
> > > > Christoffer, how are those on your machines?
> > > I attached alsa-info from the Sirius Gen2 device.
> > >
> > > Comparing the working/nonworking of Jerry, yeah, the assignment of
> > > 0x10 and 0x11 looks switched around. I don't see what difference this
> > > would make. Also, node 0x22 has "bass speaker" controls in the
> > > non-working version.
> > >
> > > Comparing the Sirius Gen2 alsa-info with Jerrys, to me it looks like
> > > the non-working version corresponds to our working version.
> > >
> > > I would expect the non-working version to happen all the time though
> > > with regards to the "bass speaker" controls. Why would this only
> > > happen sometimes?
> > Thanks!  The assignment of DACs depend on the pins and topology, so it
> > can be a bit sensitive.
> >
> > Now looking more closely at both outputs, I wonder how the commit
> > breaks pang14.  Maybe it has a PCI SSID 2782:12c5 (or 12c3) while the
> > codec SSID is 2782:12b3?  If so, the patch below should fix.

Interesting, you're right, PCI SSID c3/c5 and codec SSID c3/c5 for the 
Siriuses.

I had a look around. In patch_realtek there are some cases where codec 
SSID match is needed as well. Would it be better/safer to directly do 
this immediately or keep it as an exception where it breaks or for known 
sensitive models/brands?

> > Could you guys try it and verify whether it fixes for Pangolin and
> > doesn't break Sirius?
> >
> >
> > Takashi
> It does seems to fix the issue on Pangolin. It might worth mention
> that the headphone output will have the same issue when the speaker is
> not working. Now they are all good. Thanks!
>
>
> Jerry

Tested on our devices, both speakers still active.

That's an ok from my side as well, thanks!

Christoffer