GNOME Bugzilla – Bug 745104
remote-display: remove spice detection
Last modified: 2016-03-09 14:38:51 UTC
Spice can be used with virgl for accelerated rendering. The llvm-pipe check is enough for disabling animations when using the QXL driver.
Created attachment 297793 [details] [review] remote-display: remove spice detection
Can you point me at documentation for using virgl with spice ? Does this work today if I run a vm in boxes ?
(In reply to Matthias Clasen from comment #2) > Can you point me at documentation for using virgl with spice ? Does this > work today if I run a vm in boxes ? Well, it's not easy today, you need a lot of development branches, at some specific revision. But yes, there is some proof-of-concept work to integrate virgl with spice locally.
then this bug should probably be targeted post-3.16 ?
Review of attachment 297793 [details] [review]: To be committed for GNOME 3.18 (or whenever VirGL actually starts shipping to end-users).
virgl is soon to be ready (merged in qemu & mesa and partially in kernel), I am looking at Spice integration again. This patch is still valid. An alternative for Spice/2d would be to check if the video driver is QXL instead. I don't know the best way to do that though.
Let me know when it's actually all upstream, and we'll commit it, hopefully for 3.20.
(In reply to Bastien Nocera from comment #7) > Let me know when it's actually all upstream, and we'll commit it, hopefully > for 3.20. I wish we would do that before, it's quite annoying for testing. It's the first thing you notice. If one can come up with a solution involving QXL driver check, could we have it before?
(In reply to Marc-Andre Lureau from comment #8) > (In reply to Bastien Nocera from comment #7) > > Let me know when it's actually all upstream, and we'll commit it, hopefully > > for 3.20. > > I wish we would do that before, it's quite annoying for testing. Not sure I understand why you wouldn't be able to have a test gnome-settings-daemon for that, but hey. > It's the > first thing you notice. If one can come up with a solution involving QXL > driver check, could we have it before? That would be good enough, yes.
actually, as I explained in comment #1, I don't think there is any reason to check for Spice. The llvm pipe check should be enough. please consider the patch as is. thanks
(In reply to Bastien Nocera from comment #7) > Let me know when it's actually all upstream, and we'll commit it, hopefully > for 3.20. ping All the spice bits are now upstream, including libvirt/libvirt-glib bits. we are only missing a few pending patches for virt-manager & osinfo/boxes & releases.