Re: None

[Laurent Gualdi]

On Wednesday 21 May 2003 21:08, Farid Hajji wrote:
> > I tried that, and FWICT, Pistachio needs the VME processor capapbility
> > (virtual mode extensions, see
> > http://x86.ddj.com/articles/vme1/vme_overview.htm), which bochs
> > doesn't provide.

I've checked the source of bochs 2.0.2 and virtual 8086 mode
seems to be implemented.

By default after a ./configure, you can see in configure.h of bochs source

#define BX_SUPPORT_V8086_MODE 1 

It seems also that all the code needed to emulate this mode
is present.

> IIRC, someone already suggested a tempfix for bochs regarding VME.
> I don't remember the details without looking back in the archives,
> but was it not simply having bochs falsely _pretend_ that it
> supported VME? This would work as long as VME isn't _used_
> in Pistachio... Of course, adding true VME support to bochs
> would be better. Any volunteers?

Here is a patch that set the VME bit to 1 when BX_SUPPORT_V8086_MODE = 1.
In this patch, the VME bit is only set for pentium processor 
or greater ( cpu-level >= 5 ) .
This patch include also the DE bit hack that set the DE bit to 1
to make the L4ka/hazelnut work with bochs.
 
After applying the patch, you can use this command line to compile bochs with 
the DE bit hack:

./configure --enable-cpu-level=5 --enable-4meg-pages --enable-de-bit-hack  
--enable-disasm --enable-x86-debugger

I didn't test the small VME hack yet with L4ka/Pistachio... maybe someone
can tell me if it is working ?

> > I wonder why Pistachio needs VME, however. Maybe someone can give a
> > hint?
>
> I'm as clueless as you here... ;)

me too ;-))

--
Laurent

patch-de-vme-bochs2.0.2.diff
Description: Text Data