[linux-audio-dev] Jack Callbacks

Dave Robillard drobilla at connect.carleton.ca
Wed Feb 25 01:45:52 UTC 2004


On Sun, 2004-02-01 at 08:34, Paul Davis wrote:
> >On Sat, Jan 31, 2004 at 01:51:48PM -0500, Dave Robillard wrote:
> >> > --- jackd/engine.c      13 Jan 2004 15:11:27 -0000      1.80
> >> > +++ jackd/engine.c      31 Jan 2004 01:07:26 -0000
> >> > @@ -1503,6 +1503,7 @@
> >> >         for (node = client->ports; node; node = jack_slist_next (node)) {
> >> >                 port = (jack_port_internal_t *) node->data;
> >> >                 jack_port_clear_connections (engine, port);
> >> > +               jack_port_registration_notify (engine, port->shared->id, F
> >ALSE);
> >> >                 jack_port_release (engine, port);
> > >         }
> >> 
> >> Yep, that calls it.  Maybe it wasn't there before because apps are
> >> expected to deregister their ports before exiting?  (None I've tested
> >> actually do).
> >> 
> >> I don't see any problems having jack do it for clients though.. if the
> >> client actually does deregister its ports before disconnecting, they
> >> won't be around anymore (ie won't get double freed).
> >
> >The only thing I added was the notification to the other clients that the
> >ports were being deregistered. It always happend before.
> >
> >We need a comment from the Benevolent Dictator as to wether this is a good
> >idea.
> 
> approved. it was just a thinko to not put this in before, rather than
> a design feature. however, it does need to be tested to ensure it
> doesn't cause "system wide" effects at client shutdown. if a couple
> more people could test out steve's one line patch and report
> themselves free of any problems, we can apply it.
> 
> --p

What was the verdict on this?  Will it be in the next release of jack?

I can do whatever further testing is necessary, at least on my (limited)
hardware...

-Dave




More information about the Linux-audio-dev mailing list