still audio hiccups

[Martin Dieringer]

On Sun, 13 May 2007, Ariff Abdullah wrote:

> On Sun, 13 May 2007 14:28:25 +0200 (CEST)
> Martin Dieringer <martin.dieringer at gmx.de> wrote:
>> On Fri, 11 May 2007, Ariff Abdullah wrote:
>>
>>> On Wed, 9 May 2007 03:40:35 +0200 (CEST)
>>> Martin Dieringer <martin.dieringer at gmx.de> wrote:
>>>> On Tue, 8 May 2007, Martin Dieringer wrote:
>>>>
>>>>> On Thu, 3 May 2007, Ariff Abdullah wrote:
>>>>>
>>>>>> On Wed, 2 May 2007 03:40:08 +0200 (CEST)
>>>>>> Martin Dieringer <martin.dieringer at gmx.de> wrote:
>>>>>>>
>>>>>>>
>>>>>>> I thought it was over when I killed powerd. But now it
>>> suddenly >>>> makes "hiccup" every few seconds again.
>>>>>>> This appears on the internal sound card (intel ich4) as well
>>> as >> the >> USB (m-audio transit)
>>>>>>> With mpd, xmms, waveplay...
>>>>>>> No load, processor or disk stress.
>>>>>>>
>>>>>>> - Thinkpad T42p
>>>>>>> - FreeBSD 6.2-STABLE #5: Thu Mar 29 16:23:53 CEST 2007
>>>>>>>
>>>>>>
>>>>>> Please show the output of pciconf -lv | grep ^pcm
>>>>>>
>>>>>
>>>>>
>>>>> pcm0 at pci0:31:5: class=0x040100 card=0x05371014 chip=0x24c58086
>>>>> rev=0x01  hdr=0x00
>>>>>    vendor     = 'Intel Corporation'
>>>>>    device     = '82801DB/DBL/DBM (ICH4/ICH4-L/ICH4-M) AC'97
>>> Audio >> Controller'
>>>>>    class      = multimedia
>>>>>    subclass   = audio
>>>>>
>>>>> # cat /dev/sndstat FreeBSD Audio Driver (newpcm)
>>>>> Installed devices:
>>>>> pcm0: <Intel ICH4 (82801DB)> at io 0xc0000c00, 0xc0000800 irq 11
>>>>> bufsz 16384 kld snd_ich (1p/1r/0v channels duplex default)
>>>>> pcm1: <USB Audio> at ? kld snd_uaudio (1p/1r/0v channels duplex)
>>>>>        mode 1:(output) 2ch, 24/24bit, pcm, 48001-96000Hz
>>>>>        mode 3:(output) 2ch, 16/16bit, pcm, 8000-48000Hz
>>>>>        mode 1:(input) 2ch, 24/24bit, pcm, 48001-96000Hz
>>>>>        mode 2:(input) 2ch, 24/24bit, pcm, 8000-48000Hz
>>>>>        mode 3:(input) 2ch, 16/16bit, pcm, 8000-48000Hz
>>>>
>>>> I switched to ACPI now (instead of APM), but the hiccups
>>> remained. >
>>>> They start after either the system is running for a while (now 6
>>>> hours) or music playback of some time (now about 2 hours
>>> altogether) > and get worse.
>>>>
>>>> Shortly after a reboot everything seems ok.
>>>>
>>>> When I run "vmstat -w 1" they get worse, but I cannot relate them
>>>> to any events..
>>>>
>>>
>>> pcm0 or pcm1, or both? Try enabling vchans, or grab binary modules
>>> from http://people.freebsd.org/~ariff/BINARY_MODULES/ . Your clock
>>> issues could be the prime suspect here.
>>
>>
>> setting vchans=1 does resampling of 44.1 to 48kHz, which is
>> unacceptable.
>>
>> I use a hacked snd_uaudio driver for the m-audio transit, so I
>> suspect I cannot use your BINARY_MODULES...
>>
>> I have no more clock issues since I moved to ACPI btw.
>>
>
> Well, as I said... __try__ .
>
> I still don't know whether your issues is with pcm0 (snd_ich) _or_
> pcm1 (snd_uaudio, hacked? What kind of hack?)

  ...MULTIPLE_ENDPOINTS has to be #defined in uaudio.c to make the 
Transit work.

>_or_ _both_ .

_both_!


> If you insist on not using (or even _trying_) the binary modules,
> look for the patchset in the above directory. Disable acpi_throttle
> (read, "acpi_throttle", nothing to with powerd or est or speedstep)
> if that matters.
>

the patchset there doesn't apply to -STABLE now...


m.