Re: nvmm

[Maxime Villard <max%m00nbsd.net@localhost>]

Le 10/10/2018 à 18:08, Rhialto a écrit :
> On Mon 08 Oct 2018 at 12:02:14 +0200, Maxime Villard wrote:
> > Here is support for hardware-accelerated virtualization on x86 [1]. A generic
>
> Since I'm currently working at an ISP which also provides an OpenStack
> cloud: when can I use this with Qemu? :-)

I wrote a patch for Qemu the other day, but it's not complete.

My biggest problem is that Qemu doesn't seem to have an instruction decoder
for hard accel.

There is one case where we need to disassemble a guest instruction: when the
guest touches the LAPIC page.

It seems that we have to provide our own small decoder for that, to get
source/destination/size from an instruction.

I know we also need a gva->gpa translator, independently from Qemu, so we'll
have to add it too.

Meanwhile, I've made some adjustments and fixes in NVMM, to ease the Qemu
code.