Hide Forgot
Created attachment 476266 [details] Visual corruption in Firefox Description of problem: Screen shows visual corruption on semi-random triggers When a full-screen movie ends in VLC (rpmfusion) using standard output, the screen looks as in the attachment (or similar). On certain hover events in Firefox, the status bar shows login-screen visuals as in the other attachment. Device is an nVidia ION motherboard (Asus AT3IONT-I), carrying a NVAC graphics chip. Using the nouveau driver from the kernel noted, and using mesa-dri-drivers-experimental for 3d support. Compiz is enabled. Version-Release number of selected component (if applicable): Kernel 2.6.35.10-74.fc14.x86_64 mesa-dri-drivers-experimental-7.9-5.fc14.x86_64 xorg-x11-server-Xorg-1.9.3-4.fc14.x86_64 compiz-0.8.6-6.fc14.x86_64 (and compiz-fusion) xorg-x11-drv-nouveau-0.0.16-11.20100826git065576d.fc14.x86_64 How reproducible: Semi-random "triggers" Steps to Reproduce: NA Actual results: Corrupted visuals Expected results: Shiny visuals Additional info: dmesg shows no strange nouveau-errors.
Created attachment 476267 [details] Visual corruption in VLC
Created attachment 476268 [details] dmesg output after reproducing VLC corruption
I don't think mesa-dri-drivers-experimental is expected to work very well with nouveau?
Well, yes and no. The -experiemtal package exists solely for nouveau_dri and nouveau_vieux_dri, and is the only way of getting 3d acceleration in Nouveau. Basically it is expected to work well with nouveau, but perhaps not so good with nVidia hardware. Judging on the latest mesa build in koji, these dri modules are stable enough to ship by default. Anyway, I am aware that these experimental modules are, well.. experimental. However, if nobody tells anyone about problems, they never get fixed.
Of course, we are very glad when somebody tests experimental drivers (knowing that these are experimental) and files bugs. It would be also helpful, if you could attach to this bug uncompressed /var/log/Xorg.0.log. I don't see anything obviously bad in dmesg output (is it output of dmesg after the corruption and before reboot? just to be sure). Thank you
Created attachment 478735 [details] Xorg log Indeed dmesg was printed before reboot and after corruption. I do however see more visual corruption on the weirdest places. Yesterday for instance when Rhythmbox was emptying the list of artists (I removed tons of those) the fonts were rendered on top of each other, changing every second from orange to grey to black to any of those three colours. The attached Xorg.0.log includes the visibility of small corruptions, as I'm having reproducing the hover corruption. The exact trigger is still unknown.
Created attachment 486524 [details] dmesg from NVAC machine with all updates installed My current dmesg might be related, but could just as well be a different bug. Visual corruption included shadows just messed up and unreadable gnome-terminal. Both with compiz on. With compiz of, errors continue to exist and be reported. Those could be triggered by the default output module of VLC.
Fedora 15 seems to have fixed these problems.