lkml.org 
[lkml]   [2004]   [Nov]   [27]   [last100]   RSS Feed
Views: [more markup]  [less markup]  [headers]  [forward] 
 
Messages in this thread
/
DateFri, 26 Nov 2004 09:52:39 +0100 (MET)
FromEsben Nielsen <>
SubjectRe: Priority Inheritance Test (Real-Time Preemption)
I will look through your arguments when I get time (probably not
before Sunday :-(). It does fit with the meassurements I have so far.

The reason for running 100000 samples is that for high depth and many
tasks the statistics is quite pure at the end. So I can just as well set
it high and do all the daily life stuff.
Esben

On Fri, 26 Nov 2004, Ingo Molnar wrote:

> 
> * Ingo Molnar <mingo@elte.hu> wrote:
> 
> > 
> > * Ingo Molnar <mingo@elte.hu> wrote:
> > 
> > > the additional +1 msec comes from the fact that 1-deep lock/unlock of
> > > lock1 is an allowed operation too - 2 msec would be the limit if the
> > > only sequence is the 2-deep one.
> > > 
> > > so i think the numbers, at least in the 2-deep case, are quite close
> > > to the theoretical boundary.
> > 
> > in the generic case i think the theoretical boundary should be something
> > like:
> > 
> >    sum(i=1...n)(i) == (n+1) * n / 2
> > 
> > 	n=1	limit=1
> > 	n=2	limit=3
> > 	n=3	limit=6
> > 	n=4	limit=10
> > 
> > this is quite close to what you have measured for n=1,2,3, and i think
> > it's becoming exponentially harder to trigger the worst-case with higher
> > N, so the measured results will likely be lower than that.
> 
> also, you might want to try the simpler N-deep-locking-only variant,
> where the maximum latency should be 'n'. This likely needs some changes
> to the blocker.c code though - i.e. set 'max' always to lock_depth.
> 
> 	Ingo
> 

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

\
 
 \ /
  Last update: 2005-03-22 14:08    [from the cache]
©2003-2008