atari-bootstrap (aka m68kboot)
Hi all,
[below the boring text there are some requests that you might help with]
I have been comparing the latest source of atari-bootstrap (3.3.4,
Debian unstable) with m68kboot CVS. I wrote they differ simetime ago. I
also said Debian was more up-to-date. I was kind of wrong.
The debian version contains a fix for CT2 and one small simplification
in cookie handling (compared to m68kboot). The source code hierarchy is
flat and contains source code for atari only. It's also easier to build.
The m68kboot CVS contains an important fix in evaluation of user
specified memory size (compared to debian). The source code hierarchy is
much more rich (3 levels of folders) and the source code itself looks
much better modularized. It also contains code for LILO and Amiga
support.
This difference makes it very difficult to compare and with my
close-to-zero knowledge of the boostrapping it's almost impossible to
judge which further differences are fixes (one example for all: m68kboot
loads kernel one page above a start of memory while atari-bootstrap
doesn't skip the first page. what's better and why?)
I have no idea what to use for the d-i bootstrap. I am tempted to go for
the m68kboot as it looks a bit more current but it's ATM probably
completely untested. And there is no time (before Sarge release) to
throw a completely untested code at Atari users and wait what happens.
What's even worse: there doesn't seem to be a compiler and library that
would allow me to build the binary of the bootstrap that would be same
as the one distributed in atari-bootstrap uuencoded. m68k-tos-gcc is not
working anymore. And mine m68k-mint-gcc builds 2xlarger binary that
prints staircase text because of LF/CRLF difference between MiNT and
TOS.
Now what I would need to know:
- does someone have a working m68k-tos-gcc with TOS library suitable for
building the atari-bootstrap?
- if not, is it OK to modify the source code to add CRLF to get rid of
staircase text?
- can I remove the BOOTINFO compatibility #define to disable the 1.0
backward compatibility for kernels 2.0? AFAIK the bootstrap.prg in d-i
was still using the 1.0. Who knows what will change when the 2.1
bootinfo is enabled. I read that kernel then skips the cpu/fpu and hw
tests. Is it better or worse?
I'd have more questions but I'll better stop here.
Petr
Reply to: