Return to

A little teaser of what is to come :)


This is a more general passthrough related question.

Does anyone know what the situation is with laptops, like for example Dell XPS 15/Precision 5520 that have a dedicated GPU? I’m about to get one with an M1200.

How are the outputs mapped to GPUs, what’s the situation with iommu?

There’s a small chance that the Linux group inside Dell / Project Sputnik or the Canonical folks might even be interested in helping out/providing some kind of help given that they ship Quadro laptops. @wendell Jupiter Broadcasting people should have a contact if you think it’s worth reaching out - JB folks could cover this story too.


You know you are doing good when you get more stars per day than the linux kernel.


Is anyone having issues running this on ubuntu? I am on the libvirt part:

domain type=‘kvm’>
<qemu:arg value=’-device’/>
<qemu:arg value=‘ivshmem-doorbell,chardev=ivshmem,vectors=1’/>
<qemu:arg value=’-chardev’/>
<qemu:arg value=‘socket,path=/tmp/ivshmem_socket,id=ivshmem’/>

I am getting the following output:
error: XML document failed to validate against schema: Unable to validate doc against /usr/share/libvirt/schemas/domain.rng
Expecting an element name, got nothing
Invalid sequence in interleave
Invalid sequence in interleave
Element domain failed to validate content

Failed. Try again? [y,n,i,f,?]:

Anyone know what to do?


You appear to be missing brackets on the opening qemu:commandline? Also, there’s now a support thread you should be posting in.


your first line needs to look like this:

<domain type='kvm' xmlns:qemu=''>

Otherwise you can’t add qemu arguments to the XML.


This, can confirm


I think the thread title should be renamed if this is going to be the official thread for feedback/kudos


I think we need an IRC tbh. I feel it’d more convenient for support purposes


Yeah, but then it gets into the whole IRC or Discord issue. IRC is for old timers and Discord is for the new and hip(sters)


IRC, it simply has better integration with things, more support and GitHub will tie directly into it. I will see about getting a channel up on FreeNode today.

Is anyone here interested in maintaining parts of the content on the looking glass website? Between programming and support I am finding I don’t have the time needed to keep the website current, or tweak things in it.

I believe have a logo coming from Crysta for the site, once we have that it would be good to start looking at making the website look the part.


Why not open source the website? You could look into moving the website itself to GitHub too, running off of GitHub Pages:

I’ve seen a few websites set up like this and it seems to work really well, with anyone being able to submit changes for the website just like with any other project.


I would rather go with what we have so far, and add a wiki to it for this kind of thing, for now I need some authors that would like to contribute to the guide, etc. They would be granted permission to work on these.


Latest looking glass running fine here on a rather unmighty rig:
AMD 7650K CPU (4 Cores) with integrated GPU for host
and a simple Nvidia 1050 (not even Ti) running in a win10 pro guest.

What you see here is a chrome running in the guest with a 3d version of wikipedia - running absolutely smooth.

Thanks! So I learned vfio passthrough AND looking glass in 1 day.

Only thing thats not so nice is key combo / hotkey stealing between gnome and win 10 here and mouse input is not on par so you don’t know where you click, even with mouse acceleration disabled inside win10.

But that’s one of the reasons you guys use a tiling wm, don’t you? :wink:
When input bugs are done, this piece of software is damn cool.

I am tempted to hand some money over, when the input thing is correct.


Well done! Glad to see it working well on some lower end stuff.

Key combo stuff is not really resolvable, we are at the mercy of the WM with what keys are bound, the WM gets them before us, and we have no way to tell if the WM got them first and passed them on or not.

The mouse issues, I assume you’re not running with a 1:1 ratio? If the client is scaling the input there will be mouse issues at current, this is harder then you would think to correct. Please also make sure you have the latest version of the client, a few fixes went in over the last 24 hours that address some mouse issues.

Nope, I use it because Gnome3 sucks, and I do 99% of my work in a console, needing to move from the keyboard to the mouse and back is a huge waste of time.

Pretty tall order there, especially since most of the input problems are due to bugs in Qemu, not Looking Glass.

Remember that the focus of Looking Glass is to get the image back to the host, the input stuff was literally added for my own convenience, turned out to be handy so it was kept.


have you tried running the og thief on the system, it only seems proper given the name


Ok I see there’s not much you can do about the mouse input. Keyboard is not that important for that is not perfect inside qemu guest and is of course influenced by gnome. I think it is not really related to the looking glass client.

But for mouse input - yes I am using latest code and maybe I should try fullscreen (-F) and see if that betters mouse input.

Besides that looking glass performs really well! Nice job at that early stage.

Have to investigate some settings for I am still a noob at gpu passthrough.

Interesting note at the end: I am using a beamer in my living room and have connected both gpus with their own hdmi input cable. I just switch the output via remote control. So I don’t really need two screens :wink:


This is me right at this moment. *sighs*


Many people do that, we keep getting asked what is the point of Looking Glass.

  1. I dont have dual digital inputs on my screens, only 1x DVI and 1x VGA
  2. I want to be able to inline/embed the window into my tiling window manager
  3. You can do things with looking glass you can’t do with a physical screen, for example, go and create a custom 4k resolution and have it display on a 1080p monitor, ie, NVIDIA DSR for free on any video card, without NVIDIA…


Oh and to put things into perspective… over the last two days I have been writing the Vulkan renderer for Looking Glass… 1200 lines now and all it does is show a black screen… about to scream, this API is a nightmare to intialize for something as simple as dumping a bitmap to the screen.


Well, it is a lower level API so it needs more calls to initialize. I do feel your pain though. It’s like me trying to Overclock my Xeon.