Re: [qvm86-devel] Effect of upcoming harware support on Qemu

[Paul Brook]

On Sunday 01 May 2005 23:33, M. Khalid Khan wrote:
> What will be the effect of upcoming virtualization enhancements in
> intel/amd hardware ? Will there still be a need for dynamic translation ?
> As I understand from media reports, the new hardware will allow even the
> privilaged instructions to be virtualized. Will it result in a relatively
> thin software layer to provide full virtualization?

It should allow all guest code to be virtualized, making dynamic translation 
redundant for native guests.

I'm not sure if it's worth extending qvm86 to use this technology. It may be 
better to extend something like Xen to run guest operating systems 
unmodified.

The advantage of a qemu based solution is that you already have all the 
hardware device emulation written. The disadvantage is that that overhead may 
be higher than a pure virtualization based solution.

The best solution may be to design a common interface so that multipele 
emulators/virtualization programs can share hardware emulation code..

Paul