Jay Estabrook wrote:
>
> On Fri, May 12, 2000 at 02:06:30PM -0600, Michal Jaegermann wrote:
> >
> > As a result
> > cycle_freq was left at 0 and a calibrating loop was stuck.
>
> Sigh; I'm afraid cycle_freq may be 0 for more than UX, it may be
> that way for all MILOs.
I am afraid that I checked that only on UX. Its boot sequence a bit
different as it has 'ldmilo.exe' loader. OTOH "HWRPB cycle frequency
bogus" is not showing up on EB64+ booting via MILO so at least there
'hwrpb->cycle_freq' cannot differ too much from 'est_cycle_freq'.
> Anyway, I'd like it better to *force* the use of the calculation, if
> equal to 0; try replacing:
>
> cycle_freq = hwrpb->cycle_freq;
>
> with
>
> if (!(cycle_freq = hwrpb->cycle_freq)) cycle_freq = est_cycle_freq;
This will have really the same effect as 'diff' will be zero,
'ppm_error' therefore zero as well, and in both ways we end up with
'est_cycle_freq' in 'cycle_freq'. The only difference is that "seems
inaccurate" message will be not printed. I did what I did mainly
because an old version of this message was showing up on UX; it does not
look that it carries a very meaningful information anyway. :-)
Michal
