[...]
> In short, you're proposing a massive project, and hasn't come up with
> any justification other than the fact that "the other guys do it", which
> is no justification at all.
I think you're mistaken about the size of the project. Move the 16
bit setup code to setup_arch and convert for execution in protected
mode. I don't say this is easy, but it won't take that long for
someone who understands the platform inside out. The code to set up a
v86 box and call the bios is already in apm.c. Now throw away that
stupid parameter block and put everything on the kernel command line.
Again, not necessarily easy, but short.
Now vmlinux is a complete image bootable by any loader that
understands ELF. There are at least three of these already (grub;
FreeBSD loader; Solaris/x86 /boot). Backward compatibility headers
already exist; they're what was left after you whacked the
hardware-banging code out of setup.S.
What's the win? It's easier to write a bootloader that can load
multiple different OS's, because the "other guys" are _all_ using ELF
images. Look at it that way, and it's not another format to support;
it's two formats (zImage, bzImage) you no longer have to support. It
is also easier to boot from environments other than bog standard PC.
EFI and netboot come to mind.
zw
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/