[PlanetCCRMA] wierd hanging of jack, or some other process using fc3 edge kernel

[Fernando Lopez-Lezcano]

On Mon, 2005-04-04 at 16:25, Shayne O'Connor wrote:
> On Mon, 2005-04-04 at 12:57 -0700, Fernando Lopez-Lezcano wrote:
> > On Mon, 2005-04-04 at 04:56, Shayne O'Connor wrote:
> > > this is pretty vague, but i've been getting sporadic hanging of - i
> > > think - the jack daemon with the latest CCRMA edge kernel for fedora
> > > core 3.
> > > 
> > > what happens is this. either:
> > > 
> > > * i start jack with qjackctl, and then have a session with ardour,
> > > hydrogen, and maybe some VST effects. sometimes, after closing hydrogen
> > > and ardour, i'll press the "stop" button on qjactctl - but it will just
> > > hang at that inbetween "stopping" state, never actually stopping
> > > completely until i reboot.
> > > 
> > > * or, i go to start jack, which will fire up - but with no input/output
> > > channels, and ardour complaining that the current song was recorded at
> > > 48000khz, not 0hz (yes - that's the sample rate that qjackctl shows!).
> > > if i then stop jack, it will stop, but will not start again until i
> > > reboot. i guess there is some interesting messages when it shuts down,
> > > too - when it tries to umount /home, it is busy, along
> > > with /var/lib/jack/tmp. 
> > > 
> > > when i run gnome-system-monitor to try and kill jack, it doesn't show
> > > jackd at all, but rather some blank processes that apparently aren't
> > > using any resources, but are sure preventing my system from functioning
> > > correctly.
> > > 
> > > here's a screenshot of the processes window:
> > > 
> > > http://www.machinehasnoagenda.com/images/unknown_process.gif
> > >
> > > this isn't happening *all* the time, but enough to move me to action :)
> > 
> > Have you looked at /var/log/messages or the output of dmesg? I smells
> > like something is causing a kernel oops, and after that happens all bets
> > are off in terms of how usable the system remains. 
> 
> 
> i've dug through the system logs, but didn't find any explicity mentions
> of a "kernel oops" (i've been wondering - what the heck *is* a "kernel
> oops"?),

Basically it is a "crash" internal to the kernel, it finds some internal
inconsistency that should not happen, reports an error (grep -i oops to
find them) with a stack dump and shuts down a portion of the kernel. In
extreme cases the machine hangs completely - the kernel is so confused
that nothing else is allowed to happen. In other cases just a driver
crashes and then anything having to do with that driver hangs, which may
mean the whole thing hangs or just weird things happen. 

> but i've decided to include some of the errors and warnings i
> did find in there. most of these probly have nothing at all to do with
> the problem i describe, but i'll list them in what i think is their
> order of relevance (if, indeed, they have any - if they don't, i'll post
> some more pertinent ones as soon as it happens again):

Hmmm, in a fast read I don't see anything strange in what you posted. 

Do a "ps -auwx" from a terminal when this happens again (instead of
using the process monitor window). That will give us more information on
what processes are up and what is their state.

-- Fernando