Re: [SCHED] wrong priority calc - SIMPLE test case

[Paolo Ornati <ornati@fastwebnet.it>]

On Mon, 02 Jan 2006 10:15:43 +0100
Mike Galbraith <efault@gmx.de> wrote:

> >This is the bad situation I hate: some cpu-eaters that eat all the CPU
> >time BUT have a really good priority only because they sleeps a bit.
> 
> Yup, your proggy fools the interactivity estimator quite well.  This 
> problem was addressed a long time ago, and thought to be more or less 
> cured.  Guess not.

In my original real-life test case (transcode) I found that the problem
started with the removing of "interactive_credit":

http://groups.google.com/group/fa.linux.kernel/browse_thread/thread/6aa5c93c379ae9e1/a9a83db6446edaf7?lnk=st&q=insubject%3Asched+author%3APaolo+author%3AOrnati&rnum=1&hl=en#a9a83db6446edaf7

This is not actually true... in fact that change only unhidden the
problem for that particular test-case.

With my little proggy I'm now able to reproduce the problem even with
"interactive_credit" applied (for example with a 2.6.10 kernel).

Said this, and since "nicksched" doesn't have this problem at all, it
is an ingosched (and others as well) problem.

-- 
	Paolo Ornati
	Linux 2.6.15-rc7-plugsched on x86_64