Re: [Qemu-devel] (Before) RFC for new features

[Jim C. Brown]

On Fri, Jul 09, 2004 at 06:57:21PM +0100, Natalia Portillo wrote:
> About the BIOS question we should have a modular BIOS.
> 
> Like AWARD, AMI, Phoenix, etc, BIOS, that have support for a great variety
> of chipsets and CPUs, and are compiled with required modules for each
> motherboard. 
> 

A modular BIOS designed for qemu is ok (e.g. VGABios vs Bochs BIOS vs
some other hand-written BIOS) but supporting a commercial BIOS written
for actual hardware is another matter. This is very difficult as such a BIOS
will expect different hardware in different places. Not to mention adding 
support
for different types of motherboards/ROM chips. Its easier to just fix the
current qemu BIOS.

P.S.

Someone is working on that iirc.... posted to the list a while back. They
loaded qemu directly from a bootloader (so there was no host OS, only a guest
OS) and did some hacking so qemu could access the host BIOS directly. So it
is possible, just very hard.

> > I don't think that the chipset particularly cares what 
> > extended instructions the CPU can execute... However, BIOS 
> > support we are definitely lacking. IIRC, the BIOS we are 
> > using doesn't do any PCI setup
> > - so we rely on some code in pc.c and pci.c to make the 
> > chipset look configured and to set up the BARs.
> > 
> > --
> > Antony T Curtis <address@hidden>

-- 
Infinite complexity begets infinite beauty.
Infinite precision begets infinite perfection.