Re: Multiple video elements lead to erratic mouse behavior through spice
On Wed, Jun 10, 2020 at 01:58:13PM +0200, Christian Ehrhardt wrote:
Hi,
with a guest set up as in the following example mouse behavior breaks:
<graphics type='spice' port='5901' autoport='no'
listen='0.0.0.0'>
<listen type='address' address='0.0.0.0'/>
<gl enable='no'/>
</graphics>
<video>
<model type='qxl' ram='65536' vram='16384'
vgamem='16384' heads='1'
primary='yes'/>
<address type='pci' domain='0x0000' bus='0x00'
slot='0x01'
function='0x0'/>
</video>
<video>
<model type='qxl' ram='65536' vram='16384'
vgamem='16384' heads='1'/>
<address type='pci' domain='0x0000' bus='0x08'
slot='0x01'
function='0x0'/>
</video>
Please note that this is for illustration purposes as this everyone can
easily retry.
I originally had QXL + GPU as mediated device, where the behavior is the
same.
If I know connect to that via spice, spice is smart and opens both displays:
$ virt-viewer --connect qemu+ssh://ubuntu@10.246.115.255/system guestname
But while with one display mouse integration (to move in/out smoothly) and
positioning of the mouse pointer is fine.
It is broken with this double display setup.
What I see is:
- no mouse integration anymore (falls back to capture the mouse)
- often the mouse can not move upwards anymore
- sometimes "left" also doesn't work
Let me know if someone wants a video in case the description is still too
odd.
I have tried spice-client-gtk and virt-viewer, but both behave the same way.
If I take away one of the displays it works well again.
So I'm wondering how virtual-multi-monitor is supposed to work, are there
known bugs or tweaks that would apply here?
Did you enable the SPICE agent ? IIUC, the SPICE agent provides a
paravirtualized mouse device, and that may be what's required to have
this work properly with multi-monitor. The agent is also what lets
you dynamically turn on/off each head and do display resizing.
BTW, there are two ways todo multi-monitor with SPICE. One as youve
shown here, and the oither using a single QXL card, but with heads=NNN
instead of heads=1. One approach works best with Windows, and one
approach works best with Linux, but I can't remember which way around.
I expect they'll probably still have the same mouse behaviour though.
P.S. I tried VNC but that just has "other" issues. It only
shows 1 of 2
displays and on the one I see over amplifies the mouse movement by the
amount the other display adds to ther overall desktop size.
VNC simply doesn't support multi-monitor at all. In theory we could
block this in libvirt, but then if QEMU fixed it, libvirt would be
needlessly blocking it.
Regards,
Daniel
--
|: https://berrange.com -o- https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org -o- https://fstop138.berrange.com :|
|: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|