[linux-audio-dev] xruns, VP kernel, interrupts

Florian Schmidt mista.tapas at gmx.net
Sat Oct 2 11:44:30 UTC 2004


On Sat, 02 Oct 2004 04:29:50 -0400
Lee Revell <rlrevell at joe-job.com> wrote:

> On Sat, 2004-10-02 at 02:36, John Hedditch wrote:
> > OK. The 'Unexpected pointer ...' messages seem to be related to IRQ
> > threading, so turning them off fixed that. I still get xruns when
> > opening or closing hydrogen or ardour. Also, ZynAddSubFX generates
> > huge amounts of them.
> > 
> 
> Disable XRUN debugging.  It is not reliable.  The wiki needs to be
> updated.  Does jackd still report xruns?

Hmm, why do you think it's not reliable? It might not be too precise but
it still is a useful tool for debugging the alsa drivers themselfes i
can imagine. I for example still experience sporadic xruns independent
of the period size. Sometimes none for 12hs, then 2 per h for a while.
Since the longest non preemptible critical section reported is only
250us, the preempt timing code is not useful in debugging these. But the
xrun traces might give the driver dev some clue about what's going on.

But independent of that, the Wiki really needs to be updated. The xrun
diagnosis needs to be clearly seperated from the preempt timing stuff.

I will get to this tomorrow.

> 
> Also did you read the WIKI regarding the NPTL problem?

Good point.

flo



More information about the Linux-audio-dev mailing list