[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[QUESTION] Multiboot again
From: |
Ruslan Nikolaev |
Subject: |
[QUESTION] Multiboot again |
Date: |
Thu, 18 Aug 2005 15:46:04 -0700 (PDT) |
Yes, in general I agree with you. I think that prevents loading kernel will be better... but it's not a very principal question.
I HAVE ANOTHER QUESTION:
Kernel usual running in negative virtual address space. For example GRUB Legacy masks entry_point and all memory addresses of program segments with 0xFFFFFF when loading FreeBSD ELF. In contrast for multiboot ELF it loads directly.
I have a question: how can I run kernel in negative address space and use multiboot ELF. May be it is a very stupid question... but I really needs to know about it :)
"Yoshinori K. Okuji" <address@hidden> wrote:
On Thursday 18 August 2005 00:27, Ruslan Nikolaev wrote:
> I have a one question. GRUB requires have a kernel as an executable file
> (at least for ELF). This is not very good from user point. For example user
> can try to load such kernel at operating system and get something like
> "Segmentation fault".
I don't understand what is wrong with this. It is normal to get an error when
the user executes a binary file which is not supposed to run on an alien
environment.
> I think that having kernel as a non-executable file (for example ET_REL for
> ELF) is better. However we need another "entry" point (special symbol). It
> is possible to have something like "multiboot_start". ...And doesn't have
> "_start" or "start" symbols in kernel at all.
I don't agree. From the view of boot loaders, a kernel is truly an
executable.
So it is natural to use ET_EXEC. Applying ET_REL is very strange, because a
kernel is not a relocatable object (who relocates the kernel?).
Okuji
_______________________________________________
Grub-devel mailing list
address@hidden
http://lists.gnu.org/mailman/listinfo/grub-devel
Start your day with Yahoo! - make it your home page