[PATCH-v2] Repair vsmsetup on GX2

[Nils]

The attached patch fixes vsmsetup for the AMD GX2 .
It was broken at least since rev 4611.
This patch is build and boot tested, it now gives a :
VSA2 VR signature verified message.

Signed-off-by: Nils Jacobs

v2:This updated patch also changes Rambase back to 0x4000
in the Kconfig files of the Rumba, Frontrunner, Btest
and Rev_a boards.
This time Kconfig build tested :OK.

Thanks,Nils

[ron minnich]

Acked-by: Ronald G. Minnich

[Stefan Reinauer]

On 1/14/10 8:08 PM, ron minnich wrote:
> Acked-by: Ronald G. Minnich

I'd like to suggest solving this in a different way if possible.

Roughly two objections:

Moving coreboot down to 0x4000 is not safe, and we should attempt to fix
it living above 1MB. There are two different ways for doing this:
 - Adding an entry to the GDT that allows a real mode segment above 1MB
(Rudolf Marek did that once for resume)
 - Copying a stub code to the low 4KB (i.e. 0x600 is safe) for the real
mode calls. util/x86emu/x86.c does that.

Also, adding yet another copy of a GDT seems quite counter productive
since we have about half a dozen already...

Stefan

[ron minnich]

On Thu, Jan 14, 2010 at 11:44 AM, Stefan Reinauer  wrote:
> On 1/14/10 8:08 PM, ron minnich wrote:
>
> Acked-by: Ronald G. Minnich
>
> I'd like to suggest solving this in a different way if possible.
>
> Roughly two objections:
>
> Moving coreboot down to 0x4000 is not safe, and we should attempt to fix it
> living above 1MB. There are two different ways for doing this:
>  - Adding an entry to the GDT that allows a real mode segment above 1MB
> (Rudolf Marek did that once for resume)
>  - Copying a stub code to the low 4KB (i.e. 0x600 is safe) for the real mode
> calls. util/x86emu/x86.c does that.
>
> Also, adding yet another copy of a GDT seems quite counter productive since
> we have about half a dozen already...

Agree with your objections. I am thinking we can get it working and
then do one more go round to fix the problems you point out?

ron

[Stefan Reinauer]

On 1/14/10 11:33 PM, ron minnich wrote:
> On Thu, Jan 14, 2010 at 11:44 AM, Stefan Reinauer  wrote:
> > On 1/14/10 8:08 PM, ron minnich wrote:
> >
> > Acked-by: Ronald G. Minnich
> >
> > I'd like to suggest solving this in a different way if possible.
> >
> > Roughly two objections:
> >
> > Moving coreboot down to 0x4000 is not safe, and we should attempt to fix it
> > living above 1MB. There are two different ways for doing this:
> > Adding an entry to the GDT that allows a real mode segment above 1MB
> >
> > (Rudolf Marek did that once for resume)
> > Copying a stub code to the low 4KB (i.e. 0x600 is safe) for the real mode
> >
> > calls. util/x86emu/x86.c does that.
> >
> > Also, adding yet another copy of a GDT seems quite counter productive since
> > we have about half a dozen already...
>
> Agree with your objections. I am thinking we can get it working and
> then do one more go round to fix the problems you point out?

oh absolutely.