[LTP] [RFC] [PATCH] pselect01: Tune thresholds

[Helge Deller]

> > Hi!
> > > > >   requested_us = tst_timespec_to_us(t->tv) * t->iterations;
> > > > > 
> > > > > If slack_per_run is slack of all iterations, why is it multiplied
> > > > > again with number of iterations?
> > > > 
> > > > That variable is poorly named. It should have been slack_per_call or
> > > > slack_per_iteration. It's slack per one syscall computed using the same
> > > > formula that is used in kernel...
> > > 
> > > If it's per iteration, shouldn't this function be passed time per
> > > iteration?
> > > At the moment, it's time of all iterations:
> > >   requested_us = tst_timespec_to_us(t->tv) * t->iterations;
> > >   threshold = compute_threshold(requested_us, t->iterations);
> > > 
> > > Should it be ... ?
> > >   threshold = compute_threshold(tst_timespec_to_us(t->tv), t->iterations);
> > 
> > Good catch, I will fix that, retest and send v2.
> 
> OK, if we agree, it's supposed to be time of single iteration,
> isn't new function (in some cases) more strict than it used to be?
> 
> Before we had:
>   threshold = requested_us / 100 + 200 * t->iterations;
> which was 1% slack plus small number (200) that scales with iterations.
> 
> Now we have 0.1% slack plus small number (40) that scales
> with iterations plus big number (20000) that scales inverse
> to number number of iterations.
> 
> Say for 1000 us and 1000 iterations we used to have slack
> 10000 + 200*1000 =~ 210000. And now it's
> (40 + 1) * 1000 + 20000 / (1000/200+1) = 41000 + 3333 =~ 44333


Additionally, please keep in mind that on some architectures the
time measured is based on the jiffies clock resolution which depends
on the CONFIG_HZ value and results in a resolutions of e.g. 0.01, 
0.004, or 0.001 seconds which is way above the resolution you get
on the x86 platform (with the TSC clock).

This testcase was failing for me on the parisc platform because of this inaccuracy.

Maybe the resolution of the CLOCK_MONOTONIC clock (which is basically used in this
test) should be checked with clock_getres() and it's result included in the calculation?

Helge