Re: [Qemu-devel] QEMU 0.5.3 release

[art yerkes]

On Sun, 18 Apr 2004 19:58:03 +0200
"J. Mayer" <address@hidden> wrote:

> On Tue, 2004-04-13 at 18:00, Martin Garton wrote:
> > On Sun, 4 Apr 2004, Fabrice Bellard wrote:
> > 
> > >    - initial x86-64 host support (Gwenole Beauchesne)
> > 
> > Does this mean it will soon be possible to compile and run qemu on amd64
> > emulating x86?
> > 
> > Or is there still a lot to do?
> 
> There are some assumptions in softmmu that addresses are 32 bits long. I
> sent a patch to Fabrice to handle those problems:
> I'm now able to launch PPC system emulation on an Athlon64.
> I cannot test user emulation, for now, because it doesn't link. It seems
> that the linker script assumes to be on an opteron, not an Athlon64 (the
> Athlon64 MMU is only 48 bits...) but I didn't tried to fix it for now.

It turns out that the mmu assumption is fairly easy to fix...  Basically,
since qemu uses addend in the tlb struct to address either a host or guest
address, addend must be void *, along with everything in its path.

This patch allows the mmu to succeed, but the actual instructions have the
same problem (32 bit host address assumption).  More work is needed for
full support.  Maybe we and other interested users can help make this work.
On my amd64, I can use qemu compiled 32bit, but there seem to be bugs
initializing the virtual ethernet, which I need.
-- 
IBM had little or no expertise on Intel processors.
 -- paragraph 54, SCO v IBM, second amended complaint
No matter how cynical you become, it's never enough to keep up.
 -- Lily Tomlin

qemu.diff
Description: Binary data