[linux-audio-user] Re: quattro

[Patrick Shirkey]

Brian Redfern wrote:
> If I run jackd as root it freezes my machine, but if I run it as as a
> normal user that it hasn't frozen, but as root after about 2 minutes it
> totally freezes up.
> 

This doesn't happen for me.

I use jack + quattro with a celeron 1 Ghz pIII, ASUS microATX mobo, 512 
MB RAM, sis630 VGA.

kernel-2.4.21-ll
usb-ohci.
gcc-2.95.4
enlightenment-0.16.x
alsa-driver-0.9.4 (Jul 31 2003 cvs)
jackd-0.90 tarball which Paul made because I cannot compile cvs for some 
reason.

I can run for hours without dropouts as root with:

jackd -v -R -a -d alsa -d quattro2 -p 1024 -n 3 -zt

(I cannot actually use alsa as a normal user on this machine but I never 
want to anyway. ;)

I wouldn't say the jack driver is useless for me. In fact I just 
realised that even though I get spurious amounts of error messages:

late driver wakeup: nframes to process = 2048.

with -n 4 I can run all four channels without obvious playback errors 
and with only a couple of dropouts for a simple 10 second record test in 
ardour (while also running alsaplayer). However my computer has issues 
with changing virtual desktops while running jackd so this crashed after 
about two minutes. Straight after I changed desktops though.

Running with -n 3 and recorded for 7 minutes before jackd shutdown. The 
sound quality is poorer (very obvious crackling and pops) but I didn't 
switch desktops.

This time I prevented a hard reboot by alt+sysreq+e , alt+sysreq+i. BTW 
I also experience the problem of jackd not accepting ^C in the term with 
this version.

Running with -n 6 the sound quality is much cleaner. I was able to 
listen to the previous 7 minutes then record for another 2 before jackd 
died. Trying again I was able to record for about 9 minutes before jackd 
crashed. Luckily, this time I was able to ^C out :/

This is what the output from jack looks like:

---------
load = 11.0997 max usecs: 2507.000, spare = 20712.000
late driver wakeup: nframes to process = 2048.
late driver wakeup: nframes to process = 2048.
late driver wakeup: nframes to process = 2048.
late driver wakeup: nframes to process = 2048.
late driver wakeup: nframes to process = 2048.
late driver wakeup: nframes to process = 2048.
late driver wakeup: nframes to process = 2048.
late driver wakeup: nframes to process = 2048.
late driver wakeup: nframes to process = 2048.
late driver wakeup: nframes to process = 2048.
late driver wakeup: nframes to process = 2048.
late driver wakeup: nframes to process = 2048.
late driver wakeup: nframes to process = 2048.
late driver wakeup: nframes to process = 2048.
late driver wakeup: nframes to process = 2048.
late driver wakeup: nframes to process = 2048.
late driver wakeup: nframes to process = 2048.
late driver wakeup: nframes to process = 2048.
late driver wakeup: nframes to process = 2048.
load = 10.8494 max usecs: 2461.000, spare = 20758.000
late driver wakeup: nframes to process = 2048.
late driver wakeup: nframes to process = 2048.
late driver wakeup: nframes to process = 2048.
late driver wakeup: nframes to process = 2048.
late driver wakeup: nframes to process = 2048.
late driver wakeup: nframes to process = 2048.
late driver wakeup: nframes to process = 2048.
late driver wakeup: nframes to process = 3072.
delay of 23439.000 usecs exceeds estimated spare time of 20758.000; 
restart ...
delay of 23439.000 usecs exceeds estimated spare time of 20758.000; 
restart ...
delay of 23439.000 usecs exceeds estimated spare time of 20758.000; 
restart ...
ALSA lib pcm_hw.c:494:(snd_pcm_hw_start) SNDRV_PCM_IOCTL_START failed: 
Broken pi
pe
could not start playback (Broken pipe)
jackd watchdog: timeout - killing jackd
Aborted
--------

The load = 10.8494 max usecs: 2461.000, spare = 20758.000 messages are 
almost exactly equally spaced until the last one where we get

late driver wakeup: nframes to process = 3072.

jackd seems to be happy as long as the late driver wakeup doesn't exceed 
3072. Maybe this could be enforced somehow to make a simple fix for USB 
devices. Although I would rather see the complete solution implemented. 
(hint hint nudge nudge).



> On Sat, 2003-11-08 at 16:29, Jack O'Quin wrote:
> 
>>Ryan <ruinaudio at comcast.net> writes:
>>
>>
>>>Jackd also freezes my machine when using my usb interface (emi 2|6) and
>>>passing the -R flag.  I think this is an known issue with jack and usb
>>>devices, for all intents and purposes jack with various usb cards is
>>>useless currently.  Some apps seem to work fine for me too.
>>
>>This may be known to some people.  I recall vague mention of USB
>>problems on various mailing lists.  But, nothing specific.
>>
>>Although this sounds like a very serious problem, I don't see any bug
>>report in <http://jackit.sourceforge.net/mantis>, the JACK Mantis
>>database.  If it makes JACK useless, someone should report it,
>>providing enough information for developers to work on the problem.
>>
>>Regards,
> 
> 
> 
> 


-- 
Patrick Shirkey - Boost Hardware Ltd.
Http://www.boosthardware.com
Http://www.djcj.org - The Linux Audio Users guide
========================================

Being on stage with the band in front of crowds shouting, "Get off! No! 
We want normal music!", I think that was more like acting than anything 
I've ever done.

Goldie, 8 Nov, 2002
The Scotsman