Hi,
I have a Compaq Deskpro EN with PIII/866 recycled as a thin client.
The problem was that sound on the thin client was played faster than it
should.
I have Fedora 9/x86-64 on the main machine, so the thin client is LTSP5
that comes with F9.
I tried everything (acpi=off, etc) but the strange voice didn't get better.
I discovered this in the dmesg:
intel8x0_measure_ac97_clock: measured 50895 usecs
intel8x0: measured clock 78 rejected
intel8x0: clocking to 48000
I looked at the current kernel sources and found the solution for that.
I had to provide this for the thin client in
/opt/ltsp/i386/etc/modprobe.conf :
options snd-intel8x0 ac97_clock=44100
# lspci -s 00:1f.5
00:1f.5 Multimedia audio controller: Intel Corporation 82801BA/BAM AC'97
Audio Controller (rev 02)
# lspci -ns 00:1f.5
00:1f.5 0401: 8086:2445 (rev 02)
Best regards,
Zolt?n B?sz?rm?nyi
At Sat, 28 Jun 2008 19:42:15 +0200,
Zoltan Boszormenyi wrote:
>
> Hi,
>
> I have a Compaq Deskpro EN with PIII/866 recycled as a thin client.
> The problem was that sound on the thin client was played faster than it
> should.
> I have Fedora 9/x86-64 on the main machine, so the thin client is LTSP5
> that comes with F9.
> I tried everything (acpi=off, etc) but the strange voice didn't get better.
> I discovered this in the dmesg:
Did this problem exist on the earlier kernel?
> intel8x0_measure_ac97_clock: measured 50895 usecs
> intel8x0: measured clock 78 rejected
This value is way too small than normal. It means that either the
clocking or the calculation of the position in
intel8x0_measure_ac97_clock() is wrong. Could you check what is the
value if ichdev->position in that function?
> intel8x0: clocking to 48000
>
> I looked at the current kernel sources and found the solution for that.
> I had to provide this for the thin client in
> /opt/ltsp/i386/etc/modprobe.conf :
>
> options snd-intel8x0 ac97_clock=44100
Usually it's a bit slower than that. See
Documentation/sound/alsa/ALSA-Configuration.txt.
thanks,
Takashi
Takashi Iwai ?rta:
> At Sat, 28 Jun 2008 19:42:15 +0200,
> Zoltan Boszormenyi wrote:
>
>> Hi,
>>
>> I have a Compaq Deskpro EN with PIII/866 recycled as a thin client.
>> The problem was that sound on the thin client was played faster than it
>> should.
>> I have Fedora 9/x86-64 on the main machine, so the thin client is LTSP5
>> that comes with F9.
>> I tried everything (acpi=off, etc) but the strange voice didn't get better.
>> I discovered this in the dmesg:
>>
>
> Did this problem exist on the earlier kernel?
>
IIRC, yes.
When I have set up LTSP5, both the client and the host had a 2.6.24.x
kernel.
Don't ask the exact version, though.
>> intel8x0_measure_ac97_clock: measured 50895 usecs
>> intel8x0: measured clock 78 rejected
>>
>
> This value is way too small than normal. It means that either the
> clocking or the calculation of the position in
> intel8x0_measure_ac97_clock() is wrong. Could you check what is the
> value if ichdev->position in that function?
>
How? Is there a module option for such debug info?
>> intel8x0: clocking to 48000
>>
>> I looked at the current kernel sources and found the solution for that.
>> I had to provide this for the thin client in
>> /opt/ltsp/i386/etc/modprobe.conf :
>>
>> options snd-intel8x0 ac97_clock=44100
>>
>
> Usually it's a bit slower than that. See
> Documentation/sound/alsa/ALSA-Configuration.txt.
>
Thanks.
>
> thanks,
>
> Takashi
>
>
At Sun, 13 Jul 2008 08:39:33 +0200,
Zoltan Boszormenyi wrote:
>
> Takashi Iwai ?rta:
> > At Sat, 28 Jun 2008 19:42:15 +0200,
> > Zoltan Boszormenyi wrote:
> >
> >> Hi,
> >>
> >> I have a Compaq Deskpro EN with PIII/866 recycled as a thin client.
> >> The problem was that sound on the thin client was played faster than it
> >> should.
> >> I have Fedora 9/x86-64 on the main machine, so the thin client is LTSP5
> >> that comes with F9.
> >> I tried everything (acpi=off, etc) but the strange voice didn't get better.
> >> I discovered this in the dmesg:
> >>
> >
> > Did this problem exist on the earlier kernel?
> >
>
> IIRC, yes.
> When I have set up LTSP5, both the client and the host had a 2.6.24.x
> kernel.
> Don't ask the exact version, though.
Any older kernel?
> >> intel8x0_measure_ac97_clock: measured 50895 usecs
> >> intel8x0: measured clock 78 rejected
> >>
> >
> > This value is way too small than normal. It means that either the
> > clocking or the calculation of the position in
> > intel8x0_measure_ac97_clock() is wrong. Could you check what is the
> > value if ichdev->position in that function?
> >
>
> How? Is there a module option for such debug info?
No, just add printk() appropriately.
thanks,
Takashi