Annoying whitenoise sound coming from snd_hda enabled chipset

[Yoshihiro Ota]

On Thu, 26 Feb 2009 22:05:29 +0200
Alexander Motin <mav at FreeBSD.org> wrote:

> Yoshihiro Ota wrote:
> > On Tue, 24 Feb 2009 21:57:27 +0200
> > Alexander Motin <mav at FreeBSD.org> wrote:
> >> Yoshihiro Ota wrote:
> >>> In my case, with hdac or pcm device, 8-CURRENT fails to boot very requentry.
> >>> It fails to prove a device and doesn't proceed farthar than that point.
> >>> Now it only boots once in 5 or 10 reboots.
> >>>
> >>> When it boots, it prints lots of following messages.
> >>>
> >>> hdac0: HDA Codec #0: Conexant CX20549 (Venice)
> >>> hdac0: unable to allocate widgets!
> >>> hdac0: unable to allocate widgets!
> >>> hdac0: unable to allocate widgets!
> >>> hdac0: unable to allocate widgets!
> >>> hdac0: unable to allocate widgets!
> >>> hdac0: hdac_command_send_internal: TIMEOUT numcmd=1, sent=1, received=0
> >>> hdac0: hdac_command_send_internal: TIMEOUT numcmd=1, sent=1, received=0
> >>> hdac0: hdac_command_send_internal: TIMEOUT numcmd=1, sent=1, received=0
> >>> hdac0: hdac_command_send_internal: TIMEOUT numcmd=1, sent=1, received=0
> >>> hdac0: hdac_command_send_internal: TIMEOUT numcmd=1, sent=1, received=0
> >>>
> >>> I think this started happening about a week ago or so, sometime between
> >>> 13 and 16th.
> >> The only significant change last time was enabling MSI by default. But I 
> >> don't think it should lead to such errors. I have tried even operation 
> >> completely without interrupts working and there is not such errors. Can 
> >> you send me complete verbose dmesg with the problem and `pciconf -lvc` 
> >> output?
> >>
> >> Also you may try to disable MSI by setting 'hint.hdac.0.msi=0' hint.
> > 
> > Once I disabled snd_hda_load="YES" in /boot/loader.conf, it comes up
> > all times.  However, when I run "kldload snd_hda", the system stops responding,
> > i.e. crashes.
> > 
> > So, now I added 'hint.hdac.0.msi="0"' in the hint; then, system comes up fine.
> > I tried "kldload snd_hda" and got the following output.
> > System didn't crash after adding the hint so that I added snd_hda_load="YES"
> > back to /boot/loader.conf.
> 
> I have committed a patch to the CURRENT that should disable MSI for your 
> HDA controller by default. Test it please.

That's r189086, isn't it?

So, for testing, I will remove hint.hdac.0.msi and add 'snd_hda_load="YES"' back.
Does that sound valid test?

I will report it tomorrow.

Thanks,
Hiro