[Qemu-devel] Re: Intel VMX support?

[Qemu-devel] Re: Intel VMX support?

[Arun Sharma]

Nate Lawson wrote:
> Have you considered adding support for this instruction set?
> http://cache-www.intel.com/cd/00/00/19/76/197666_197666.pdf

It's not clear if you're trying to execute instructions such as VMREAD
inside qemu or use qemu's device models + VT-x (used to be called VMX)
for executing instructions.

If it's the latter, the code is in xen-unstable now. Can boot Linux,
FreeBSD, MSDOS etc.

We took a 0.6.1 snapshot and were able to run with very few modifications
to the "hw" directory.

There were some discussions on where qemu should run on xen-devel (Look
for the thread "Device model architecture")

The main changes we did were:

- accelerate i8254 by moving parts into the hypervisor
- We had some hacks to run without the BIOS, but now that we have BIOS
  support, these can go away.

In general, we do have a desire to keep the tools/ioemu/hw directory in
Xen insync with the upstream qemu development.

Comments/suggestions?

        -Arun

[Qemu-devel] Re: Intel VMX support?

[Nate Lawson]

Arun Sharma wrote:
> Nate Lawson wrote:
> 
>>Have you considered adding support for this instruction set?
>>http://cache-www.intel.com/cd/00/00/19/76/197666_197666.pdf
> 
> It's not clear if you're trying to execute instructions such as VMREAD
> inside qemu or use qemu's device models + VT-x (used to be called VMX)
> for executing instructions.
> 
> If it's the latter, the code is in xen-unstable now. Can boot Linux,
> FreeBSD, MSDOS etc.
> 
> We took a 0.6.1 snapshot and were able to run with very few modifications
> to the "hw" directory.
> 
> There were some discussions on where qemu should run on xen-devel (Look
> for the thread "Device model architecture")
> 
> The main changes we did were:
> 
> - accelerate i8254 by moving parts into the hypervisor
> - We had some hacks to run without the BIOS, but now that we have BIOS
>   support, these can go away.
> 
> In general, we do have a desire to keep the tools/ioemu/hw directory in
> Xen insync with the upstream qemu development.
> 
> Comments/suggestions?

No, just that I thought qemu could handle this on its own, separate from 
Xen.

-- 
Nate