[LAU] jackd process can't be killed

[Ken Restivo]

On Wed, Dec 31, 2008 at 12:49:26AM -0800, Fernando Lopez-Lezcano wrote:
> On Tue, 2008-12-30 at 21:44 -0800, Ken Restivo wrote:
> > On Tue, Dec 30, 2008 at 04:20:11PM -0500, R Dicaire wrote:
> > > On Tue, Dec 30, 2008 at 3:42 PM, Adam Sampson <ats at offog.org> wrote:
> > > 
> > > > This looks like the 2.6.26 MIDI sequencer problem -- MIDI doesn't work,
> > > > and after jackd has opened the sequencer device for the first time,
> > > > it'll hang trying to close it. Future attempts to open the sequencer
> > > > device then fail because it's in use.
> > > 
> > > Ahhhhhhhh, I was never aware there was a midi issue...jackd could be
> > > killed when I didn't use -X, excellent.
> > > Thank you VERY much.
> > > 
> > > Now here's another situation. I've built alsa and jack. I installed
> > > them into custom locations. In my users .asoundrc
> > > I have a pcm device that calls the jack plugin. How would I get calls
> > > to this plugin to use my custom built
> > > /usr/local/alsa/lib/alsa-lib/libasound_module_pcm_jack.so instead of
> > > the system one in /usr/lib/alsa-lib ?
> >
> > MIDI doesn't work in 2.6.26???!
> > 
> > Is that for real? Is it really a kernel problem or an ALSA problem? 
> > Is there a workaround?
> 
> The problem I see is with the rt patch applied. And only when
> interfacing with external hardware midi interfaces. A partial workaround
> made midi input work, but the internal timer has problems. 
> 
> I have not checked midi under 2.6.26-rt12 yet...

I have a 2.6.26.8-rt12 sitting here built, ready to try booting off of it.

What exactly is the MIDI problem, and how could I duplicate it here? I'd be happy to test with this kernel (on a x86 Atom) and see if it is solved.

-ken