Hi, Just moved to xbmc (which uses gl), from freevo and more often than not, I get extreme diagonal rendering artifacts on the gui. Pretty much each gui object (a box, a single character, etc) has a diagonal split and either of the halves renders o, and the other is mostly black. The diagonal flickers between the corners at a rapid pase (~ 4 changes / second). I files report on it at xbmc (http://trac.xbmc.org/ticket/14017) which also has screenshots and a short movie grab attached. Freevo worked ok, so I guess this is somehow related to new code paths executed by xbmc (both 11.0 and 12-git suffers). I have no idea if it's kernel driver related or somewhere higher in the stack. Ideas to narrow this down appreciated. xbmc is the only app which behaves like this. -ath
This is more likely an issue in Mesa than in the kernel. Please attach Xorg.0.log and the output of glxinfo and dmesg.
Created attachment 73876 [details] dmesg
Created attachment 73877 [details] glxinfo
Created attachment 73878 [details] Xorg.0.log
Might be interesting if you could try a recent Mesa Git snapshot.
Tried mesa git as of yesterday. No improvement. I've also tried to flip all USE flags on/off for mesa 9.0.1 with no improvement. Doing some surfing, I noticed the z buffer issues being a common problem, so I decided to try the car racing game "trigger". It does show the same effect as I've seen in screenshots on youtube (flickering triangular misrenderings on objects). The dynamic behavior seems to be the same. However, I see the same effect on the fist screen in Trigger, where the three helmets are shown. I'm not sure if they are supposed to be animated or static. The zbuffer issues I saw in other reports were all related to movements (Stuff move back/front, when I move a little). I also have zbuffer issues on the terrain once the game starts. If xbmc makes use of depth to paint its screen with a bunch of objects, I can see how this might be z related. However, xbmc should show a static screen (I think), whereas common z reports are all related to movement of the observer (I think). Any ideas how one could test this theory? Another crazy thing is that every once in a while it renders perfectly ok. While "bisecting" through the USEflags, I had a run of "good, good, good,..." and each test was a restart of Xserver, starting xbmc. After going back to the known bad set of USE flags (which now rendered good!!), I did some cursing and warm rebooted the box, and it suddenly rendered bad again :-(. This kind of suggests an uninitialized HW register to me... Any ideas how one can make xbmc not exercise the (gl?) related areas of the HW? I'm lacking a solid known-good state at the moment.
-- GitLab Migration Automatic Message -- This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity. You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/mesa/mesa/issues/349.
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.