[Coco] format problems

[Gene Heskett]

On Saturday 19 January 2019 17:21:26 L. Curtis Boyle wrote:

> Format takes enough system RAM to build an entire track image at once
> (so it can build the interleaves), and not just the data portion, but
> the header, footer, sync, CRC, etc. The system call for write track
> should be changed to allocate a temporary 8k block from outside the
> system map, since it has to kill IRQ’s during the write track anyways
> (for the standard floppy controller; no halt controllers and/or hard
> drive controllers will do things a little differently).

So that might not/won't fix it for my no-halt disto controller. Whats it 
take for the no-halt? Humm, I thought that was disabled during 
formatting? The fact that its coming into rb1773 at write track, or read 
track, either one should bypass the sector buffer if rb1773 is written 
correctly.  Have we found a bug in rb1773?  Sure nuff smells like it, 
Curtis. Maybe I should try it with another, halting controller? That 
would be a revolting development to quote Jackie.

> As for SHELL, 
> it should be kept at <$1E00 bytes simply so that it doesn’t take more
> thank one 8K block in it’s process space. Since processes always have
> the program mapped at the end of the 64k process space, and the last
> $200 (512) bytes is always vector page RAM and I/O, if you push it
> past that it has to allocate 16k in your process space so that it
> won’t try to overwrite the vector page & I/O, and still leave you up
> to 56k data area for the shell to use, if needed.
>
> Sent from my iPhone
>
> > On Jan 19, 2019, at 4:08 PM, Tormod Volden <lists.tormod at gmail.com> 
wrote:
> >> On Sat, Jan 19, 2019 at 10:40 PM Gene Heskett wrote:
> >> I have managed to get the twin seagate 1G drives to spin up again,
> >> so my coco is running. My latest build of nitros9 has a format
> >> command that $10C7 byte long, same as my latest nitros9 build for a
> >> 6309.
> >
> > ...
> >
> >> At the end of all this, it looks as if I have exactly the same code
> >> as my March 2016 build, possibly with some descriptor fixes by
> >> Chris Hawk.
> >
> > Hi Gene,
> >
> > The level1/cmds/format.asm in the repo hasn't had any code changes
> > since 2005: Module size: $10C7  #4295
> > Module CRC : $0ABD19 (Good)
> >
> >> Format it unchanged. Still needs 11170 bytes of sysram, but but I
> >> only have 10k available. That equals an out of ram crash by about
> >> the 3rd track written to a floppy. Since it writes a track at a
> >> time while formatting, that seems like its asking for more memory
> >> than it needs.
> >
> > Yes, can anyone explain why it needs so much memory?
> >
> >> But adjusting its memory either direction only makes it crash
> >> before even stepping to the next track. So I still can't format
> >> anything.
> >>
> >> Whats the maximum size I can merge to make a shell, this one is
> >> $1DC7 long.
> >
> > Your $1DC7 is 7623 decimal. The cmds/makefile in the repo warns that
> > the shell module must be less than 7681 bytes. Not sure why.
> >
> > Regards,
> > Tormod
> >
> > --
> > Coco mailing list
> > Coco at maltedmedia.com
> > https://pairlist5.pair.net/mailman/listinfo/coco


Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page <http://geneslinuxbox.net:6309/gene>