Re: [patch] latency tracer, 2.6.15-rc7

[Lee Revell]

On Fri, 2005-12-30 at 16:59 -0800, Mark Knecht wrote:
> I've noted for awhile that on my AMD64 machine that has xrun issues
> that at least annecdotally it has always seemed that the network
> interface was somehow involved. I wonder if this may turn out to be
> true? 

Yes, it probably is.  Since at least 2.6.14 almost all of the 1ms+
latencies left in the kernel are due to long running network softirqs -
thanks to lots of work by Ingo and others there are almost no long-held
spinlocks left.  So I would certainly expect audio underruns to
correspond to heavy network activity.

I believe that softirqs usually run on the processor that they were
raised on so if you have an SMP system you could test this by locking
all interrupt handling to one processor and running JACK on the other
and see if your xruns decrease.

Lee

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/