Re: load_end == bss_end == -1

[Yoshinori K. Okuji]

At Tue, 14 May 2002 11:06:28 +0300,
Yuri Zaporogets wrote:
> IMO, multiboot kluge is one big hack itself :) . Modifying two lines
> in boot.c won't harm anything, and it's not a violation of current
> Multiboot standard (behavior of the boot loader under such conditions
> isn't specified anywhere). 'mbchk' program runs all tests successfully,
> so we don't need to touch it at all.

Hmm... So we increase hacks? Oh, this world must be hell.

Okay, if you stick to the idea so much, I don't object. I'm not very
concerned about the a.out kludge anyway.

But it is essential to update the Multiboot Specification precedently,
because I want to avoid any GRUB-specific extension to the spec, if
possible. Would you like to work on this?

On another thing: I don't think it is a good idea to use -1 as invalid
values. It is 0xffffffff actually, so it must be valid
certainly. Therefore, I think it would be better to use 0 as invalid
values instead, because:

1. If you wanna load an OS image, (load_end_addr == 0) is
   nonsense. That makes sense only when (load_addr == 0), but, even in
   this case, (load_end_addr - load_addr == 0) means that no code is
   loaded, so this case is also nonsense anyway.
2. (bss_end_addr == 0) is nonsense, because, as shown above,
   load_end_addr must be greater than zero.

What do you think?

Okuji