[PlanetCCRMA] jackd xruns problem: solved?

Fernando Pablo Lopez-Lezcano nando@ccrma.Stanford.EDU
Sat Aug 14 12:50:02 2004


On Sat, 2004-08-14 at 08:26, Steve Harris wrote:
> On Fri, Aug 13, 2004 at 01:17:52 -0700, Fernando Pablo Lopez-Lezcano wrote:
> > > Then I tried to compile a 2.6.8-rc4 with CK2 and.. no more xruns!
> > > (with # jackd --realtime -Rv -d alsa -p 128 -n 3 -r 44100
> > >  and same applications running).
> > > 
> > > I must admit, I didn'to do hard testing by now, but
> > > I think it could be encouraging!
> > 
> > There has been a lot of progress in the last few weeks. I'm currently
> > running 2.6.8rc4 plus Ingo Molnar's volutary scheduling O5 and it is
> > looking very very good. The gurus are hard at work tracking hard stuff
> > that deals with mlockall...[*]
> 
> How serious are the mloclall problems? Any chance of throwing an RPM or
> kernel source over the fence?

Yes, I could do that. In fact I should have already done that.
Apparently (if I understand things right) a mlockall on an unrelated
process can trigger an xrun in jack. Mlockall is rare outside jack, I
think, but it is still a fundamental problem. 

>  > [*] ... so no luck yet with getting a proper patch for the radeon video
> > cards :-(
> 
> Which kind of radeon is it? The FireGL based chips (fglrx), or the others?
> I've got a fglrx based chip in my laptop and its a right pain generally,
> if it hindered reatime performance that would make it even more anoying.

I have a radeon mobility 7500 in my laptop and a bunch of other radeons
on desktops (up to 9200, I think). I don't know what chipset they are
based on, how do you find out?

I do have a very old patch that solves the problem but it is not a
proper patch because it reschules with a lock on. I'll insist on a fix
when I see lack of activity on the part of the gurus :-)

-- Fernando