James A. Crippen
Fri Apr 12 20:23:01 2002
Paul Fuqua <firstname.lastname@example.org> writes:
> Once upon a time, that's what one did: put a load band and a microcode
> band on a disk, flag them as "default" with a magic tool, and the boot
> process would load them up and go. (In my first weeks as a microcoder,
> I would often build microcodes that didn't work; in that case I had to
> lug the disk brick to another lab where I could hook it to a handy CADR
> and move the "default" flag to something that worked, then lug it back.)
> I think that would still work, but the "modern" way was to first boot a
> special microcode that would offer a menu of bands to boot, including
> load bands, microcodes, and diagnostics. Or was it two-stage: initial
> boot and self-test, which loaded the menu boot? Fuzzy memory.
Two-stage, I think. ISTR reading that in SSDN in the early chapters,
or perhaps somewhere else...
IIRC there's a very detailed description of the booting process in
SSDN chapter 2 or 3 or n where n is a small natural number for
suitably small values of small.
The menu boot is fairly primitive, and nothing like the complexity of
Symbolics's FEP. But it got the job done anyway.
I would like to see a Symbolics-style FEP for E3 in the future that
allows fiddling with the internals of the CPU and modifying memory.
But that can wait.
James A. Crippen <email@example.com> ,-./-. Anchorage, Alaska,
Lambda Unlimited: Recursion 'R' Us | |/ | USA, 61.20939N, -149.767W
Y = \f.(\x.f(xx)) (\x.f(xx)) | |\ | Earth, Sol System,
Y(F) = F(Y(F)) \_,-_/ Milky Way.