Description of Problem: When starting up a second X server (startx -- :1) there is corruption on the screen depending on what applications are being run. The corruption can usually seen as black areas in applications (screenshot attached), and can be seen in a default GNOME/KDE desktop almost instantly. TWM + creating a few windows and moving them around also shows similar problems. It used to work (tm), my guess is that hardware acceleration of Xrender only works fine for a single-head (and single server) configuration, the driver disables acceleration if you have a dual head config. The combination of magic Xaa options that remove the corruption seem to be (**) MGA(0): Option "XaaNoSolidFillRect" (**) MGA(0): Option "XaaNoSolidFillTrap" (**) MGA(0): Option "XaaNoSolidTwoPointLine" (**) MGA(0): Option "XaaNoScreenToScreenColorExpandFill" (couldn't get any visible corruption after these were in use although there might be a few cases these don't cover). Not worth disabling by default, I suppose, but maybe some smart workaround can be devised. X probably isn't smart enough to know if another instance is running, so this may not be possible... Version-Release number of selected component (if applicable): 4.2.0-67.6 How Reproducible: Steps to Reproduce: 1. startx 2. startx -- :1 3. Actual Results: Screen corruption on second X server Expected Results: Both X servers are fine
Created attachment 74344 [details] screenshot of corruption
There is aparently a patch to disable render acceleration floating around out there but I can't seem to find it currently.
This should be fixed in rawhide XFree86 CVS now. I'm going to backport the fix to 4.2.1 as soon as someone lets me know if it works or not. Can you test rawhide X and report back to me if it works?
Fixed in my rawhide builds, and also backported to latest test packages for RHL 8.0, 7.2, 7.1, AS2.1 (latest XFree86 CVS build, latest 4.2.1 build, latest 4.1.0 build).
4.1.0-43 has this for 7.[12] and AS 2.1 4.2.1-10 has it for 8.0 current rawhide has it for 8.0/rawhide