On 5/15/06, Pavel Cahyna <pavel@netbsd.org> wrote:
> On Mon, May 15, 2006 at 11:51:24AM +0200, Julio M. Merino Vidal wrote:
> > On 5/15/06, Pavel Cahyna <pavel@netbsd.org> wrote:
> > >yes, Just Work - that's why I want do do it before NetBSD 4.0. I'll
> > >commit it probably today.
> > >
> > >But the "Just Work" means "except the symbol table". Please, would you
> > >have time and knowledge to work on that?
> >
> > It looks like that the symbol table support when loading NetBSD from
> > GRUB has been broken for some time already and is not related to your
> > change to the linker script.  At least xtraeme@ has told me so for a
> > month.
> >
> > I know where the problem is, but not yet why it happens; need some
> > time to fix it ;-)
>
> Hmm, you told in
> http://thread.gmane.org/gmane.os.netbsd.devel.kernel/18148/focus=3D18843
> that reserving some extra space by the Multiboot header is necessary for
> the symbol table to work. Were ypu wrong?

No, and it fact it used to work.  But I do not know yet why it does
not work any more even with the patch to GRUB...

> Anyway, it would be good to get rid of MULTIBOOT_SYMTAB_SPACE...

Indeed.  The ideal thing (as martin@ suggested a while ago) could be
to add a "ksyms" parser for the data structure passed in by multiboot.
 This requires some work but is the way to go.

Cheers,

--=20
Julio M. Merino Vidal <jmmv84@gmail.com>
The Julipedia - http://julipedia.blogspot.com/