[LAU] qjackctl-0.3.x and 49% cpu

[Rui Nuno Capela]

John Anderson wrote:
> On Tue, 2007-11-27 at 14:33 +0000, Rui Nuno Capela wrote:
>> On Tue, November 27, 2007 12:10, John Anderson wrote:
>>> I'm finding that qjackctl-0.3.0, 0.3.1 and 0.3.1a all use up 49% cpu. My
>>> box is a CoreDuo, so that's probably 100% of one cpu.
>>>
>>> qjackctl-0.2.22 doesn't do this, so I've downgraded for now.
>>>
>>> If it's a bug, what can I do to help?
>>>
>> symptoms are that right after jackd is started the first time, qjackctl
>> starts eating 98% of a cpu, isn't it?
> 
> It also happens when I have an existing jackd running, and I start up
> qjackctl.
> 
>> it's a known issue on some Qt4 incarnations, specially the debian based
>> ones, iirc. it has been reported on at least one case that updating to a
>> different libqt4 makes the issue go away.
> 
> I've just upgraded from qt-4.3.1 to qt-4.3.2, and the problem is still
> there.
> 

it's not quite a nominal qt4 version issue here, but seems to be due on
some circumstantial libqt4 build conditions (compile time flags?). for
example, i had this one report on debian testing (lenny) that the latest
libqt4-4.3.2 got it through alright, but most of the previous
iterations, including an early one also tagged as libqt4-4.3.2, was
indeed infected.


>> unfortunately, i never caught that behavior reproduced on any of my
>> opensuse 10.x boxes, nor in windows :o)
> 
> Makes it kinda hard to debug ;-|
> 

yes, people handed some strace outputs and it just boiled on some poll()
calls in libqt4, allegedly under QProcess class implementation. can't
say or dig any deeper from that, are just suspicions, not facts

cheers
-- 
rncbc aka Rui Nuno Capela
rncbc at rncbc.org