Sound broken (first couple of seconds play looping) on Sony Vaio VPX11S1E (HDA, ALC262)

[Takashi Iwai]

On Thu, 29 Sep 2022 11:05:47 +0200,
Hans de Goede wrote:
> 
> Hi,
> 
> On 9/29/22 08:10, Takashi Iwai wrote:
> > On Wed, 28 Sep 2022 18:33:55 +0200,
> > Hans de Goede wrote:
> >>
> >> Hi,
> >>
> >> On 9/11/22 09:05, Takashi Iwai wrote:
> >>> On Sun, 11 Sep 2022 08:15:59 +0200,
> >>> Takashi Iwai wrote:
> >>>>
> >>>> On Sat, 10 Sep 2022 19:52:36 +0200,
> >>>> Hans de Goede wrote:
> >>>>>
> >>>>> Hi All,
> >>>>>
> >>>>> To test some kernel work (backlight refactor) which I have been doing
> >>>>> I have booted Debian Testing with a 6.0 kernel on a Sony Vaio VPX11S1E
> >>>>> (which is ancient).
> >>>>>
> >>>>> I noticed that when tab-completing something in a terminal the
> >>>>> terminal bell sound would keep repeating and playing another longer
> >>>>> sound sample gets stuck with the first couple of seconds of that
> >>>>> sample looping.
> >>>>>
> >>>>> This is under GNOME3 with pulseaudio as sound server.
> >>>>>
> >>>>> I accidentally found an interesting workaround if I run:
> >>>>>
> >>>>> aplay -Dplughw:CARD=MID,DEV=0 /usr/share/sounds/alsa/Front_Left.wav
> >>>>>
> >>>>> once, then that works properly and after that the problem is gone...
> >>>>>
> >>>>> Note that this laptop does have only 1 speaker AFAICT.
> >>>>>
> >>>>> alsa-info output below.
> >>>>>
> >>>>> I would be happy to test any patches / module-options which might
> >>>>> fix this.
> >>>>
> >>>> Could you check the very latest Linus tree (6.0-rc5)?
> >>>> There has been a regression fix for HD-audio, and possibly this might
> >>>> be your case.
> >>>
> >>> And if this doesn't change the behavior, try snoop=0 option for
> >>> snd-hda-intel module.
> >>
> >> Thanks, I've just tried with 6.0-rc6 both with and without snoop=0
> >> and neither helps I'm afraid.
> > 
> > Then one another thing could be position_fix option of snd-hda-intel.
> > Otherwise we need bisection -- supposing it worked well in the past.
> 
> Thanks, position_fix=1 does the trick (I did not try any of
> the other possible values).
> 
> I see there is a snd_pci_quirk position_fix_list, so I guess
> I should prepare a patch adding a quirk for this?

I guess this is rather controller issue, so I'd patch for Poulsbo like
below.  Let me know if it works, then I'll cook a proper patch.

> Or are some other position_fix values preferred and should
> I try those first?

LPIB should suffice, so no further test about the option needed.


thanks,

Takashi

-- 8< --
--- a/sound/pci/hda/hda_intel.c
+++ b/sound/pci/hda/hda_intel.c
@@ -2547,7 +2547,8 @@ static const struct pci_device_id azx_ids[] = {
 	  .driver_data = AZX_DRIVER_SCH | AZX_DCAPS_INTEL_PCH_NOPM },
 	/* Poulsbo */
 	{ PCI_DEVICE(0x8086, 0x811b),
-	  .driver_data = AZX_DRIVER_SCH | AZX_DCAPS_INTEL_PCH_BASE },
+	  .driver_data = AZX_DRIVER_SCH | AZX_DCAPS_INTEL_PCH_BASE |
+	  AZX_DCAPS_POSFIX_LPIB },
 	/* Oaktrail */
 	{ PCI_DEVICE(0x8086, 0x080a),
 	  .driver_data = AZX_DRIVER_SCH | AZX_DCAPS_INTEL_PCH_BASE },