[linux-audio-user] Re: [Jackit-devel] [Fwd: Re: Linux audio latency]

[Florian Schmidt]

On Sun, 03 Oct 2004 12:55:18 +0300
Sampo Savolainen <v2 at iki.fi> wrote:

> On Sun, 2004-10-03 at 08:25, Florian Schmidt wrote:
> > Starting a new jack client in itself won't produce xruns. Also,
> > cleanly shutting down a jack client will not in itself produce xruns
> > [iirc]. But it seems that many, if not most, jack apps, get some
> > part of the RT operation stuff not right. Especially start up and
> > shutdown.
> 
> Actually, that's not entirely correct. I've been talking to Paul about
> this. The xruns created by exiting clients is due to jackd having to
> rework it's internals. When doing this reordering, jackd has to lock
> the data it's modifying and sometimes (very often) there will be an
> xrun because the audio thread is waiting for the lock to be released.
> 

Ah, and btw: Shouldn't the audio thread just try the lock and if it
fails to claim it produce a period of silence? What spot in the code is
this?

flo