[PlanetCCRMA] jackd xruns problem: solved? (FC2 2.6.7rc4 O7 kernel)

[Fernando Pablo Lopez-Lezcano]

On Sun, 2004-08-15 at 15:26, Steve Harris wrote:
> On Sat, Aug 14, 2004 at 06:14:02 -0700, Fernando Pablo Lopez-Lezcano wrote:
> > Ok, Steve and other Fedora Core 2 bleeding edge users, I have the latest
> > and greatest (as of yesterday, 2.6.8 final was released in between :-)
> > over the fence. Has been tested to boot on all of two (2) machines, a P4
> > latop and an Athlon64 desktop...
> 
> This kernel looks good.
> 
> I had to run everything as root, as I couldn't build network drivers for
> this kernel and I couldn't remeber how to run the realtime lsm, but I got

/sbin/modprobe realcap allcaps=1
(to get the behavior that enables you to run jackstart)

> an AMS, jack-rack (with 6 or 7 plugins) and a JAMin running at 256
> samples, while doing some CPU and disk intensive non-realtime things and
> without a single xrun. It was running for about an hour with me changing
> things in the apps at random moments. I tested briefly at 128, but JAMin
> wasn't happy running at 128, so I couldn't test with much load, and I dont
> run that low for real anyway.

Good news! Very good news!

> I also wasn't using the LD_ASSUME_KERNEL trick (I forgot), I'l try with
> that tommorow.

I'm not using that myself in my tests. After all, we probably want to be
able to run the thing using nptl (long term). 

> I couple of times when I started up jackd I got the storm of "blah, blah
> exceeded, restarting" messages when I connected the first client, but
> oddly, quiting and restarting jackd git rid of them, so maybe theres a
> race condition on some priority setting stuff?

Strange. Anyway, it is encouraging that you are getting better results.
On to downloading 2.6.8.1 and voluntary P0 (new series, Ox is already
old)

-- Fernando