Re: [Nasm-users] Handling timer interrupt

[Frank K. <fbk...@zy...>]

Zulfi Khan wrote:
> I just tried to find the interrupt# in this topic and I got 1ch.

Yeah... 1Ch is supposed to be a "secondary timer interrupt", and it's 
supposed to be "safer" than hooking int 8 itself. However, I couldn't 
get anything to happen, using 1Ch. Maybe it has to be "enabled"(???). I 
switched to hooking int 8... and presto, my clock started working 
(sorta). I really should go back and retry with int 1Ch.

> I thought 8259 is irrelevant because its a separate IC.

It is a separate IC, but it controls which interrupts get sent to the 
CPU, and which don't. At the end of an interrupt routine, we need to 
tell the PIC we're done. Instead of dealing with that, I "chained" the 
original int 8. Grab the original address *before* installing the new 
routine, and jump into it at the end of the new routine. This is a 
common technique to take care of the "housekeeping" that needs to be 
done. Maybe that's not the way you want to do it. Lots to experiment with!

> Its prinnting garbage and F instaed of welcome kernel message.

Yeah... maybe going to an .exe was a mistake. I re-worked both the 
bootsector and the "kernel" code to jump to 7E4h:0, and made cs, ds, and 
es the same - something more like a .com file (but not exactly). This 
seems to be working. I'd like to be able to use separate "segment code" 
and "segment data" - maybe even "segment stack stack". So far, I haven't 
been able to load such a thing successfully, so I stick to "everything 
in one segment". At least I'm seeing the welcome message...

> I would try this again after reading what you have suggested.

The attached "mz2.zip" started out being your code... I altered it quite 
a bit. Made "add_isr" a subroutine, rather than your "install" macro, 
and added a "get_old_isr" to capture the original isr. "mz3.zip" is an 
attempt to integrate this into the more "modular" version. The idea is 
that the routines - add_isr, get_old_isr, write, getkey,... - are 
supposed to be "C callable". I haven't actually tried this - I don't 
"do" C - that's your department! :) I added "hours" - no way this puppy 
is going to run over a hour, but I figured I ought to handle the 
possibility...

I don't know if any of this crap will help you. It "sorta works", but 
isn't "right". The "clock" stays running in a "tight loop", but if I try 
to "getkey" for a delay, the whole thing locks up. So the only "command" 
that works is control-alt-delete. This indicates something seriously 
wrong with the way I've hooked the interrupt (I think). I haven't worked 
out what the problem is.

Takes a lot of rebooting! I downloaded Bochs, but couldn't get it to 
compile - some symbol "not defined in this scope". In a pinch, I'll mess 
with C to get it to compile when it won't, but C++ is beyond me 
completely. Trying to install from the RPM resulted in the usual cascade 
of errors. So I'm stuck with a "real reboot" from a "real floppy"... 
which I generally prefer, but it's time-consuming!

See of you can make anything out of it. Feel free to ask questions, but 
the answer will probably be "I dunno, that was the first thing that came 
into my head." :) Good Luck!

Best,
Frank