Created attachment 79464 [details] Screenshot when system is installed (not complete garbage but still broken) When you take a screenshot (reproducable with both Boxes and virsh screenshot) for a domain that is running F19 (Beta TC4), the screenshot is * perfectly fine when guest is in text/console mode. * complete garbage during graphical anaconda running. * no garbage but still broken (don't know how to describe it so attaching a screenshot) when X is running after installation. Versions: On host/client: libvirt: libvirt-daemon-kvm 1.0.5-3.fc19 (but i can also reproduce against latest git master: a842df78eac28ec42b76797f76866b84a4126d0f). spice-server: 0.12.2-5.fc19 (but i can also reproduce against latest git master: c09faf2382c47aa5ea1cdf30b4a75f410da74623) In guest: xorg-x11-drv-qxl: 0.1.1-0.4.20130312gita474a71.fc19 kernel: 3.9.0-301.fc19 Problem does not occur against VM running an older Fedora nor if you setup 'vga' as video device in the domain rather than 'qxl'. Hence the reason I think the QXL driver in the guest is the most likely culprit.
adding David in CC
Any progress on this. Fedora 19 is almost out and this bug makes Boxes look pretty bad.
(In reply to comment #2) > Any progress on this. Fedora 19 is almost out and this bug makes Boxes look > pretty bad. The main reason this bug is very important is that once Fedora 19 is released and its ISOs are finalized, this bug will never get fixed in the ISO and in a UI like Boxes such cosmetics are super important.
totally a spice server bug. commit 81fb6f1504fb9ef71f2382f44af34756668296e8 Author: Alon Levy <alevy@redhat.com> Date: Fri Feb 24 23:19:31 2012 +0200 qxl: make qxl_render_update async RHBZ# 747011 This even says in it: With this patch sdl+spice works fine. But spice by itself doesn't produce the expected screendumps unless repeated a few times, due to ppm_save being called before update_area (rendering done in spice server thread) having a chance to complete. Fixed by next patch, but see commit message for problem introduced by it. The next patch was never committed at least to the qemu tree I'm looking at. Its what I expected we do an async update area, but save the ppm without blocking for completion of the update.
I'm guessing this was fixed? If not, please mark it as obsolete as f19 isn't supported anymore anyway.
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.