>Second, Fiasco is multiboot comliant. I was forced to add multiboot
>compliant loading again.
As I moved all the binary loading into a plugin, It's ok with me to clutter
it up now. It just didn't seem like a good idea to me to clutter up the
main program with this code.
However, I do want to point out that it is *extremely* trivial to modify a
multiboot-compliant kernel to boot straight from protected mode: hardcode the
multiboot configuration structure (with fixed memory values etc., which is fine
for testing) and add "movl $address, %ebx" (or perhaps it was %eax but IIRC
that was the magic number) as the first instruction of the kernel. This
modification of the guest kernel seems much more trivial than adding multiboot
support to the host..
OTOH, perhaps we'd like to keep the possibility of directly loading
(multiboot) kernels in protected mode, even when we support real mode and a
"real" bootloader... that's handy about plugins, you can pick a different
"BIOS" if you
want :)
-- Ramon
