[Coco] More high-density success

Chuck Youse cyouse at serialtechnologies.com
Sat Jul 5 19:01:34 EDT 2008


Well, folks, I think I can officially call the hardware a success.
I just booted (partially) NitrOS-9 off a high-density floppy disk - of
course it freaks out because the wrong boot module is in the boot track,
but that's the next step on my list .. !  Then a polled mode I/O driver
(for starters - the no-halt will get tricky, and I want to put the
hardware through its paces first).
 
I undercovered a rather nasty issue with the 82077 and its handling of
256-byte sectors.  Not a show stopper, but it will make detecting
underrun/overrun conditions impossible, which is unfortunate because
they're possible if interrupts are disabled for too long - I have a
feeling some of the kernel innards of NitrOS-9 will require some lovin'
in order to cooperate with this controller - essentially dedicating FIRQ
to a pseudo-dma scheme and only disabling it when the stack is invalid.

Basically the problem is that this controller requires that the CPU
transfer data in or out of it within 320uS of its request.  Tight, but
not impossible.  If anyone wants the details of my scheme, pipe up and
I'll bore you with the minutiae..

I've been watching the posts go back and forth, my apologies for not
addressing any of them specifically, I've just been busy hacking..

C.





More information about the Coco mailing list