[PlanetCCRMA] added: ALPHA version of Planet Core components for FC2

[Fernando Pablo Lopez-Lezcano]

On Sun, 2004-06-20 at 11:30, Steve Harris wrote:
> On Sun, Jun 20, 2004 at 11:04:01 -0700, Fernando Pablo Lopez-Lezcano wrote:
> > > When I started it from qjackctl I got the printf storm xrun feedback
> > > problem.  Running it from the command line it was a bit better, but I
> > > still had a problem with the "maximum time of something xxx.yyyy,
> > > restarting" messages. My suscpicion is that theres more to it than that
> > > though.  I guess I should get on with that message patch incase to see if
> > > it helps.
> > 
> > Aw rats, I have to release a patched version of jack for fc2, there are
> > some initializations missing in the realtime thread creation code that
> > should not make a difference, but apparently do (ie: I explicitly set
> > what should be the default). I think you are seeing the same problems I
> > was seeing until I added that. 
> 
> I've just made the patch, it did help, but it didnt seem to really fix the
> problem, I've attched it just incase you want to try it, or I was doing
> something wrong.
> 
> I ran it as root. The printf storms occured less frequently (basicly only
> when it was starting up), but there was still a stream rather than a
> couple of messages, so I think the problem is deeper.
> 
> My 100% guess, no knowledge, basic gut reaction is that the dead reconing
> code is off, and theres actually nothing wrong, but the resets are causing
> the storm.

Ok, a new version of jack is there, ready for you to upgrade to :-)

Does not have your patch, it includes a patch that properly initializes
some stuff before the realtime thread is created. For some reason under
fc2 that seems to make a difference. If the patch is not included (if I
remember correctly) the realtime threads are not created SCHED_FIFO at
all, you can check that with chrt before upgrading. But pthread_create
does not complain and all the code assumes the thread has been created
just fine. 

-- Fernando