Re: Linux 2.6.9 latencies: scheduler bug?

[Sami Farin <7atbggg02@sneakemail.com>]

On Mon, Oct 25, 2004 at 02:58:07PM +0200, Ingo Molnar wrote:
> 
> * Sami Farin <7atbggg02@sneakemail.com> wrote:
> 
> > forget this stupid ip_tables.c patch, latencies have nothing to do with
> > netfilter code, but bad interaction between xmms, rtc_latencytest
> > and iptables.  I now get at max 3.1s (yup, 3100000us) latencies.
> > http://safari.iki.fi/2.6.9-xmms-fun-1.png
> > if you want to reproduce this:
> > 1) run "rtc_latencytest 1024" (can't reproduce with "rtc_latencytest 512")
> > 2) press play in xmms
> > 3) start iptables-script
> 
> to further debug any latencies please apply the -RT patchset

which includes also -mm1...

> and enable PREEMPT_TIMING (and LATENCY_TRACING), that enables us to tell us more

I guess those are not available as individual patches which could be
applied to 2.6.9?

Could someone with LATENCY_TRACING+PREEMPT_TIMING try to reproduce
that xmms/rtc_latencytest/iptables thing...  takes only a minute.

Pretty please?

> about the latencies. You dont have to enable PREEMPT_REALTIME (the most
> experimental feature of the patchset) to measure latencies.
> 
> 	Ingo

--