Re: Haswell audio no longer working with new Catpt driver

[Greg KH <gregkh@linuxfoundation.org>]

On Sat, Jan 02, 2021 at 12:50:44PM +0100, Christian Labisch wrote:
> On Fri, 2021-01-01 at 18:24 +0100, Takashi Iwai wrote:
> > On Fri, 01 Jan 2021 14:40:39 +0100,
> > Christian Labisch wrote:
> > > 
> > > $ cat /sys/module/snd_hda_intel/parameters/power_save
> > > 0
> > 
> > Hm, then the best would be to run git bisect for spotting out the
> > breaking commit.  There has been no change in VIA codec driver at all
> > between v5.9 and v5.10, so the rest possibility is either in HD-audio
> > codec helper code or controller code (or both) -- if it's about the
> > changes in the sound driver.
> > 
> > Or you can try the oneliner below as a test shot; it might keep the
> > widget node power D0, which is currently the only possible appearance
> > of the difference between working and non-working cases.
> > 
> > BTW, please avoid top-posting.  It's confusing.
> > 
> > 
> > thanks,
> > 
> > Takashi
> > 
> > -- 8< --
> > --- a/sound/pci/hda/patch_via.c
> > +++ b/sound/pci/hda/patch_via.c
> > @@ -115,7 +115,7 @@ static struct via_spec *via_new_spec(struct hda_codec *codec)
> >         spec->gen.keep_eapd_on = 1;
> >         spec->gen.pcm_playback_hook = via_playback_pcm_hook;
> >         spec->gen.add_stereo_mix_input = HDA_HINT_STEREO_MIX_AUTO;
> > -       codec->power_save_node = 1;
> > +       // codec->power_save_node = 1;
> >         spec->gen.power_down_unused = 1;
> >         codec->patch_ops = via_patch_ops;
> >         return spec;
> 
> Hi Takashi,
> 
> Thank you ! I think we have tried what we could to work around this issue.
> There is obviously something wrong with the kernel which has to get fixed.

Yes, and that is what we are trying to figure out here.  Does the above
change work?

> Hi Greg and Lars,
> 
> From my understanding there are fundamental changes regarding the audio drivers.
> The sound should (and has to) work out-of-the-box without any user interactions.

Of course, but we need to figure out the change that caused the problem
here.  If someone who can reproduce this can run 'git bisect' that is
the fastest way to make this happen.

> As you can see in the post below, other users are facing exactly the same issue.
> It would be great if this could be solved as soon as possible, maybe in 5.10.5 ?
> 
> https://forum.manjaro.org/t/issue-after-kernel-5-10-update-speakers-not-working-on-laptop/45496

5.10.5 will be out next week, and that's not how stable kernels work.
The fix needs to be in Linus's tree first before we can take it in a
stable kernel release.  We need to figure out the fix first before it
can go anywhere...

thanks,

greg k-h