This is an archived copy of the Xen.org mailing list, which we have preserved to ensure that existing links to archives are not broken. The live archive, which contains the latest emails, can be found at http://lists.xen.org/
Home Products Support Community News


Re: [Xen-users] What does viridian=1 do?

On 08/21/2009 02:15 PM, James Harper wrote:

Am 21.08.2009 04:23, schrieb James Harper:
So for faster IO you need GPLPV drivers.
I still would vote for further investigation and development on
enlighten IO. It would bring us out of the hell with driver signing,
would also make a lot of stuff easier from the users stand.
The work of James Harper is pretty good but for 2008 x64 more than
far away from ready for production.

Yes, definitely that would be good.

Feel free to start working on it :) I believe the needed viridian
can be seen at least from linux hyper-v driver (linux-ic). Also iirc ms
released some docs about the interfaces.

You are talking about reverse engineering a backend driver to match the
frontend driver in Linux. That would certainly be an interesting project,
I wonder how Microsoft would feel about it :)


Uhm, James you have the best knowledge about this topic (the other way
around), would you say this is a real gap?

I'm not at all familiar with HyperV aside from a little bit of knowledge
about what viridian=1 does, and only then because viridian=1 crashed gplpv
due to a bug in the way I had implemented my cpuid calls.

The point I was making is that Microsoft have provided the open source
drivers to better allow other operating systems to integrate with their
product. If you reverse engineer that to make their operating systems work
better with xen then they might get a little bit upset... or they might
not... I'm not a Microsoft lawyer :)

Guessing how the otherside should behave is not for the faint of heart I suppose. Ask the guys from samba on how they struggle to make samba a Domain controller ... Concerning upsetting Microsoft I don't see the problem that's what Samba team is doing and there is a lot of project in the wild doing the same with MS products and other products. Problem starts if you use patented software methods (and well at least in US mainly for the moment as for UE software patents are not valid).

I have 0 knowledge of HyperV technology but maybe the drivers for a guest Windows into a Windows host are not by default installed in the guest but proposed by the host to the guest in a similar way of how it works with virtualbox or vmware (but it seems that starting from w2k8 this drivers are already installed ... is it true ? is it true also for Windows 7 ?).

I found some documentation about the functional spec of HyperV
There may well be some technical limitations that prevent a HyperV
compatible backend layer being added to Xen... I don't know enough about
either to say.

If it could be done, then a whole lot of things would 'just work', and as
you say it could solve a lot of driver issues.

Or is the GPLPV-driver on the
way to fill it anyway soon, especially are there plans to get the
drivers signed? As I see it there wouldn't be any other proper solutions
for this topic.

I'm trying to fix a few bugs in the shutdown/suspend/resume/etc paths at the
moment and it's proving a long and frustrating exercise, and I haven't had a
lot of time to work on it lately.

Xen-users mailing list