[Qemu-devel] GSOC Qemu Project

[Qemu-devel] GSOC Qemu Project

[aahud]

Hello my name is Aaron Hudson,

I am sending this email, admittedly late, to suggest a new project. I 
would love to see Qemu ported to an operating system without a 
monolithic kernel. My initial ideas for target platforms are Minix3 or a 
Plan9-like (ie Harvey or 9front). My personal preference would be 
Harvey, as I have made small contributions to that project, and am 
simply more familiar.

Another goal might be to look into hypervisors designed for microkernels 
(NOVA, Fiasco.OC, OKL4, etc), and making sure that the userspace 
solution is flexible enough to support those implementations. Regardless 
of whether or not the summer project would include that as a goal.

If you would like more detail or simply like to talk about the idea, I'd 
love to communicate via email, or IRC.

Thank you,

Aaron Hudson

Senior @ New Mexico Institute of Mining and Technology
Computer Science Major
System Administrator for Computer Science Department

Re: [Qemu-devel] GSOC Qemu Project

[Peter Maydell]

On 18 March 2017 at 22:48, aahud <aahud@cs.nmt.edu> wrote:
> I am sending this email, admittedly late, to suggest a new project. I would
> love to see Qemu ported to an operating system without a monolithic kernel.
> My initial ideas for target platforms are Minix3 or a Plan9-like (ie Harvey
> or 9front). My personal preference would be Harvey, as I have made small
> contributions to that project, and am simply more familiar.

So from the project's perspective, I'm afraid this would be
rather going against the direction we're currently headed.
We're actually looking right now at *removing* support for
some host OSes which are unused and unmaintained (eg AIX,
Solaris, Haiku), because we have no way to test them and we
strongly suspect that few or no users are using them. So I'd be
reluctant to add host support for another platform unless:
 * there was clearly a large body of users who wanted it
 * there was going to be ongoing maintenance and support
   for the code for several years (ie more than just the length
   of a GSoC project)
 * there's an easily accessible machine for doing build testing

thanks
-- PMM