Hi On Tue, Jun 14, 2016 at 10:46 AM, Christophe Fergeau <cferg...@redhat.com> wrote: >> Imho, the fix should be on server or guest/driver side, it shouldn't >> use a display monitor config with a scanout offset if there is none to >> be applied by the client. > > Regarding fixing this on the server, isn't it going to have similar > issues? Ie once we get QXL_IO_MONITORS_CONFIG_ASYNC in QEMU, we could > make sure that display channels with just one monitor will get (0, 0) as > their primary position, but this is not correct with virgl, so we'd need > some gl check in the server before doing the zero-ing out?
virtio-gpu/virgl do not use QXL_IO_MONITORS (or any QXL io/device). It's not an issue if a QXL driver want to use a single monitor with a position different from +0+0 imho. Apparently, this change was made by Sandy to fix #1202419, although I fail to reproduce the problem (I am still using older drivers/agents). I think we can just use the solution I proposed for the virgl case, until someone else is hurt by this again and we can find a proper solution. -- Marc-André Lureau _______________________________________________ Spice-devel mailing list Spice-devel@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/spice-devel
