FYI, and it in case that it can help, I have succesfully installed Qubes
4.0 rc3 in KVM with nested virtualization and iommu enabled.
I had to hack a little bit the qemu intel_iommu.c, but now I have fully
fonctional Qubes 4 (with all the VM in hvm mode).
The VM are not flying but they do work.
BePe - From BKO
On 01/27/2018 01:31 AM, Andrew David Wong wrote:
-----BEGIN PGP SIGNED MESSAGE-----
On 2018-01-26 09:37, Marek Marczykowski-Górecki wrote:
Manual testing of installation is a time consuming job, so I was looking
for some tool to automate it. And found one:
This works by launching the system under test in KVM and then interact
with it as user would do (mouse clicks, keyboard etc). Then compare
screenshots (or parts of it).
The nice thing is when using new new enough KVM, qemu and hardware with
VT-x + EPT, Qubes installation works there! In theory KVM supports even
nested virtualization, so HVM should works inside such setup, but in
practice Xen and/or qemu crashes badly... But PV works fine, which is
enough for installation testing.
I've uploaded some basic tests here:
The problem is we don't have a place to host it in long run. While I can
launch it from time to time on one of my testing laptops, it doesn't
scale well. And access to it is restricted, which limit who can work on
this (and on issues found by tests).
Does anyone have some place/recommendation where to look? Hardware we
- CPU with VT-x + EPT, so KVM will work with reasonable performance (
means a physical machine, not virtual one)
- 6-8 GB RAM (4GB for KVM + something for OpenQA itself)
- 50 GB of disk (preferably more)
Does anyone have a spare machine running somewhere, and willing to
Alternatively, it looks like we can get one for ~70 EUR/month. Andrew,
Michael, could we use opencollective money for this?
The OpenCollective money is supposed to be used to pay developers to
work on the open-source version of Qubes. While this would not be a
direct payment to a developer, I can see how it would qualify, since
it's directly saving you and other devs time and work that would
otherwise have to be spent on testing. I think it would adhere to the
spirit of the rule, which is that the money should be spent on
improving the open-source version of Qubes. (Arguably, it would be
even more efficient than paying a developer directly, since Qubes
would probably benefit a lot more from the testing bought by that ~70
EUR/month than it would from the tiny amount of developer time ~70
EUR/month would pay for.)
However, I'd be surprised if there were no one in the Qubes community
willing to share a suitable machine for this purpose. My guess is that
there aren't as many readers on qubes-project as there are on
qubes-devel and qubes-users, so I'm CCing those other lists.
Andrew David Wong (Axon)
Community Manager, Qubes OS
-----BEGIN PGP SIGNATURE-----
-----END PGP SIGNATURE-----
You received this message because you are subscribed to the Google Groups
To unsubscribe from this group and stop receiving emails from it, send an email
To post to this group, send email to firstname.lastname@example.org.
To view this discussion on the web visit
For more options, visit https://groups.google.com/d/optout.